The Dawson Cut Forest Bed lies in the lower part of thick, late Cenozoic loess deposits in the Fairbanks area. It is associated with several distal tephra beds that provide age control and offer the opportunity of its recognition elsewhere in central Alaska. EC tephra (named herein) occurs in the uppermost part of the Dawson Cut Forest Bed and its petrographic and chemical properties point to a co-magmatic relationship with PA tephra, which has not been found in direct association with the forest bed. Both tephra beds are pink and have unusually high Cl in their glass shards, which readily separates them from all other tephra beds in the Fairbanks area. They were produced by discrete eruptions, closely spaced in time. PA tephra has a glass-fission-track age of 2.02 ± 0.14 myr, indicating that the Dawson Cut Forest Bed must be about 2 million years old. The Palisades tephra (named herein) has very similar properties to these two tephra beds, suggesting that the buried forest bed just above it at the Palisades site on the Yukon River, about 250 km west of Fairbanks, correlates with the Dawson Cut Forest Bed.

The “upper pit” at the Lost Chicken placer gold mine in east central Alaska contains fossils that provide information on the flora and insect fauna of interior Alaska just before the onset of global cooling at 2.5 myr. Fossils come from sediments interbedded with the Lost Chicken tephra (dated at 2.9 ± 0.4 myr—early Late Pliocene) and portray the floodplain and valley of a small creek within a region dominated by a coniferous forest richer in genera and species than the present one. Climate was wetter and less continental, and there was probably little or no permafrost. At least one other Pliocene tephra (the Fortymile tephra) occurs at the site and is also associated with plant and insect fossils. Among these fossils are extinct plants and insects like those found at other Tertiary sites in northern Canada and Alaska. The Lost Chicken sequence is the same age as the Beaufort Formation on Meighen Island, more than 1000 km to the north. Like Lost Chicken, Meighen Island sediments contain fossils representing a diverse boreal environment. This shows that the latitudinal climate gradient during early Late Pliocene time was shallower than at present and the boreal forest had a far greater latitudinal span than now.

Pollen analysis of a sediment core from Zagoskin Lake on St. Michael Island, northeast Bering Sea, provides a history of vegetation and climate for the central Bering land bridge and adjacent western Alaska for the past ≥30,000 14C yr B.P. During the late middle Wisconsin interstadial (≥30,000–26,000 14C yr B.P.) vegetation was dominated by graminoid-herb tundra with willows (Salix) and minor dwarf birch (Betula nana) and Ericales. During the late Wisconsin glacial interval (26,000–15,000 14C yr B.P.) vegetation was graminoid-herb tundra with willows, but with fewer dwarf birch and Ericales, and more herb types associated with dry habitats and disturbed soils. Grasses (Poaceae) dominated during the peak of this glacial interval. Graminoid-herb tundra suggests that central Beringia had a cold, arid climate from ≥30,000 to 15,000 14C yr B.P. Between 15,000 and 13,000 14C yr B.P., birch shrub-Ericales-sedge-moss tundra began to spread rapidly across the land bridge and Alaska. This major vegetation change suggests moister, warmer summer climates and deeper winter snows. A brief invasion of Populus (poplar, aspen) occurred ca.11,000–9500 14C yr B.P., overlapping with the Younger Dryas interval of dry, cooler(?) climate. During the latest Wisconsin to middle Holocene the Bering land bridge was flooded by rising seas. Alder shrubs (Alnus crispa) colonized the St. Michael Island area ca. 8000 14C yr B.P. Boreal forests dominated by spruce (Picea) spread from interior Alaska into the eastern Norton Sound area in middle Holocene time, but have not spread as far west as St. Michael Island.

The Palisades Site is an extensive silt-loam bluff complex on the central Yukon River preserving a nearly continuous record of the last 2 myr. Volcanic ash deposits present include the Old Crow (OCt; 140,000 yr), Sheep Creek (SCt; 190,000 yr), PA (2.02 myr), EC (ca. 2 myr), and Mining Camp (ca. 2 myr) tephras. Two new tephras, PAL and PAU, are geochemically similar to the PA and EC tephras and appear to be comagmatic. The PA tephra occurs in ice-wedge casts and solifluction deposits, marking the oldest occurrence of permafrost in central Alaska. Three buried forest horizons are present in association with dated tephras. The uppermost forest bed occurs immediately above the OCt; the middle forest horizon occurs below the SCt. The lowest forest bed occurs between the EC and the PA tephras, and correlates with the Dawson Cut Forest Bed. Plant taxa in all three peats are common elements of moist taiga forest found in lowlands of central Alaska today. Large mammal fossils are all from common late Pleistocene taxa. Those recovered in situ came from a single horizon radiocarbon dated to ca. 27,000 14C yr B.P. The incongruous small mammal assemblage in that horizon reflects a diverse landscape with both wet and mesic environments.

Pollen analysis at two sites, correlated by the presence of the 190,000 yr-old Sheep Creek tephra, documents fluctuations in vegetation and climate consistent with this date and indicates that the records span marine oxygen isotope stage 7 and the stage 6/7 transition. Dawson Cut, near Fairbanks, Alaska, provides a 5.2-m-long pollen record of interglacial boreal forest succeeded by shrub tundra and then forest/tundra. Ash Bend, Stewart River, central Yukon, provides a 9.5-m-long record of interglacial boreal forest succeeded by forest/tundra, shrub tundra, and herbaceous tundra. The replacement of forest at both sites by more open or tundra vegetation indicates warm interglacial conditions giving way to cold and arid climate. It is not clear whether stage 7 was warmer than the present. The warm–cool–warm climate oscillation evident at both sites may correlate to Lake Baikal substages 7a, 7b, and 7c. Sheep Creek tephra fell on forest/tundra vegetation.

Recurring glacial outburst floods from the Yukon-Tanana Upland are inferred from sediments exposed along the Yukon River near the mouth of Charley River in east-central Alaska. Deposits range from imbricate gravel and granules indicating flow locally extending up the Yukon valley, to more distal sediments consisting of at least 10 couplets of planar sands, granules, and climbing ripples with up-valley paleocurrent indicators overlain by massive silt. An interglacial organic silt, occurring within the sequence, indicates at least two flood events are associated with an earlier glaciation, and at least three flood events are associated with a later glaciation which postdates the organic silt. A minimum age for the floods is provided by a glass fission track age of 560,000 ± 80,000 yr on the GI tephra, which occurs 8 m above the flood beds. A maximum age of 780,000 yr for the floods is based on normal magnetic polarity of the sediments. These age constraints allow us to correlate the flood events to the early-middle Pleistocene. And further, the outburst floods indicate extensive glaciation of the Yukon-Tanana Upland during the early-middle Pleistocene, likely representing the most extensive Pleistocene glaciation of the area.

Two widespread tephra deposits constrain the age of the Delta Glaciation in central Alaska. The Old Crow tephra (ca. 140,000 ± 10,000 yr), identified by electron microprobe and ion microprobe analyses of individual glass shards, overlies an outwash terrace coeval with the Delta glaciation. The Sheep Creek tephra (ca. 190,000 yr) is reworked in alluvium of Delta age. The upper and lower limiting tephra dates indicate that the Delta glaciation occurred during marine oxygen isotope stage 6. We hypothesize that glaciers in the Delta River Valley reached their maximum Pleistocene extent during this cold interval because of significant mid-Pleistocene tectonic uplift of the east-central Alaska Range.

Thermoluminescence (TL) and infrared-stimulated luminescence (IRSL) sediment-dating methods have been applied to paleosol- and tephra-bearing loess sequences younger than marine oxygen isotope stage (MIS) 7 at three important sites. TL ages indicate the development of significant paleosols ∼75,000 and ∼30,000 yr ago in the loess sequence at the Gold Hill site. Relatively minor soil development occurred ∼70,000 and ∼48,000 yr ago. Like the ∼75,000-yr-old soil, the 30,000-yr-old soil is apparently of global extent, and consistent in timing with inferred warm intervals elsewhere (e.g., Greenland, Europe, western and central China). At Birch Hill, replicate TL dating of primary loess combined with two earlier TL results from the same site, and with an earlier mean fission-track-glass-shard age of 140,000 ± 10,000 yr for the associated Old Crow tephra, yield a more precise numeric age of 142,300 ± 6600 yr for this Alaska/Yukon chronostratigraphic marker ash bed. Three of the TL ages at the Halfway House site are difficult to interpret, but combined with other evidence, they indicate: (1) the upper 5–6 m of loess from Halfway House is not part of the Gold Hill Loess (equivalent to pre-MIS 5 age) as long thought by T.L. Péwé, but rather is much younger; (2) the regionally significant variegated tephra, found in the Fairbanks and Klondike areas and previously thought to be older than MIS 5, has an age of 77,800 ± 4100 yr (late MIS 5).

The fossil vole Microtus deceitensis occurs in Early Pleistocene deposits at Fort Selkirk, Yukon Territory, and Late Pliocene beds at the type locality, Cape Deceit, Alaska. Analyses of simple vs complex morphotypes in the cheek teeth, and of differentiation of tooth enamel, show that the Cape Deceit sample of M. deceitensis is less derived, and thus appears to be older, than the Fort Selkirk sample. The fossiliferous deposits at Fort Selkirk are well constrained by fission-track and radiometric dates and are 1.5 to 1.7 myr. Sediments at Cape Deceit bear a normal magnetic polarity, are correlated with the Olduvai subchron, and probably are latest Pliocene.

Thousands of Late Pleistocene remains are found in sites throughout Beringia. These specimens comprise an Ice Age genetic museum, and the DNA contained within them provide a means to observe evolutionary processes within populations over geologically significant time scales. Phylogenetic analyses can identify the taxonomic positions of extinct species and provide estimates of speciation dates. Geographic and temporal divisions apparent in the genetic data can be related to ecological change, human impacts, and possible landscape mosaics in Beringia. The application of ancient DNA techniques to traditional paleontological studies provides a new perspective to long-standing questions regarding the paleoenvironment and diversity of Late Pleistocene Beringia.

Loess-paleosol sequences of the last interglacial-glacial cycle are correlated from European Russia to central Siberia and the Chinese Loess Plateau. During cold periods represented by marine oxygen isotope stages (OIS) 2 and 4, loess deposition dominated in the Russian Plain and the Loess Plateau. In central Siberia, loess deposition took place also, but five to seven thin, weakly developed paleosols are identified in both stages. OIS 3, in the Chinese Loess Plateau near Yangchang, consists of a loess bed that is flanked by two weakly developed paleosols. At Kurtak, Siberia, OIS 3 is represented by two distinct, stacked paleosols with no loess bed separating the paleosols. In the Russian Plain, OIS 3 consists of a single, possibly welded paleosol, representing upper and lower stage-3 climates. Brunisols and Chernozems dominate the profiles in China and Siberia, whereas Regosols, Luvisols, and Chernozems are evident in the northern and southern Russian Plain, respectively. OIS 5 is represented in China and the Russian Plain by pedo complexes in a series of welded soils, whereas in contrast, the Kurtak site consists of six paleosols with interbedded loess. The paleosols consist largely of Brunisols and Chernozems. Although the three areas examined have different climates, geographical settings, and loess source areas, they all had similar climate changes during the last interglacial-glacial cycle.

Recent stratigraphic studies in central Alaska have yielded the unexpected finding that there is little evidence for full-glacial (late Wisconsin) loess deposition. Because the loess record of western Alaska is poorly exposed and not well known, we analyzed a core from Zagoskin Lake, a maar lake on St. Michael Island, to determine if a full-glacial eolian record could be found in that region. Particle size and geochemical data indicate that the mineral fraction of the lake sediments is not derived from the local basalt and is probably eolian. Silt deposition took place from at least the latter part of the mid-Wisconsin interstadial period through the Holocene, based on radiocarbon dating. Based on the locations of likely loess sources, eolian silt in western Alaska was probably deflated by northeasterly winds from glaciofluvial sediments. If last-glacial winds that deposited loess were indeed from the northeast, this reconstruction is in conflict with a model-derived reconstruction of paleowinds in Alaska. Mass accumulation rates in Zagoskin Lake were higher during the Pleistocene than during the Holocene. In addition, more eolian sediment is recorded in the lake sediments than as loess on the adjacent landscape. The thinner loess record on land may be due to the sparse, herb tundra vegetation that dominated the landscape in full-glacial time. Herb tundra would have been an inefficient loess trap compared to forest or even shrub tundra due to its low roughness height. The lack of abundant, full-glacial, eolian silt deposition in the loess stratigraphic record of central Alaska may be due, therefore, to a mimimal ability of the landscape to trap loess, rather than a lack of available eolian sediment.

Most recent summaries of eastern Beringian (Alaska and Yukon) archaeology present the Nenana complex, beginning 11,500–12,000 14C yr B.P., as the beginning of the regional archaeological record. Either explicitly or tacitly, these summaries dismiss or ignore a body of older putative evidence of human occupation that may span the late Wisconsin stade and even extend into part of the mid-Wisconsin interstade. This paper summarizes the interpretive problems surrounding the older findings, bringing together data that have accumulated over a period of two decades, in the hope that a coherent presentation will encourage more careful appraisals of the materials. The paper concludes with a family of testable hypotheses concerning the beginnings of human occupation in eastern Beringia. The hypothesis that people were present during the mid-Wisconsin interstade has not yet been falsified.