J. D. Hays, J. Imbrie, and N. J. Shackleton (1976, Science 194, 1121–1132) showed that the astronomical theory explained many features of late Quaternary ice-age climates, but they did not specify the physical mechanisms involved. Here it is proposed that interlocked variations of ice-sheet heat sinks in both polar hemispheres amplified and transmitted Milankovitch summer half-year insolation changes (a version of the astronomical theory) between 45° and 75°N into the globally synchronous climate changes recorded in geologic records. It is suggested that late Quaternary ice sheets had terrestrial components (grounded above sea level, melting margins, fluctuations controlled by climate) and marine components (grounded below sea level, drained largely by ice streams, limited melting margins, fluctuations controlled primarily by sea level and secondarily by climate, interior surface elevations coupled to downdraw through ice streams). Northern Hemisphere ice sheets were largely marine (with minor melting margins) in the Arctic and terrestrial (with major melting margins) in the midlatitudes. West Antarctic and peripheral East Antarctic ice was marine-based and lacked melting margins. Because of their geographic array, these terrestrial and marine components formed an ice-sheet system whose variations were coupled on a global scale. Milankovitch summer isolation changes near midlatitude Northern Hemisphere melting margins controlled most variations of this system, because advance or retreat of melting margins initiated concurrent eustatic sea-level change. Such sea-level change afforded the critical interlocking mechanism between terrestrial and marine components because it forced simultaneous expansion or contraction of marine margins in both polar hemispheres. This initiated an amplifying feedback loop among all marine components and influenced interior downdraw through ice streams. Arctic summer insolation change was less important because northern melting margins were relatively minor. Its greatest influence was on surface ablation of ice streams that controlled interior downdraw. This affected eustatic sea level and activated global linkage of marine sectors. By analogy with present-day Antarctica, late Quaternary ice sheets were enormous planetary heat sinks due to their reflective and radiative surface characteristics. It is suggested that the effectiveness of these ice-sheet heat sinks varied with their areal extent and interior surface elevation. Thus, it is postulated that concurrent growth or decay of these interlocked ice-sheet heat sinks in both polar hemispheres served as the global amplifier of regional Milankovitch summer insolation.

Age data for about 660 speleothems and about 140 spring-deposited travertines were collected, including many unpublished results. These data were plotted as histograms and also as error-weighted frequency curves on a 350,000-yr scale. These plots clearly show periods of increased speleothem/travertine growth as well as times of cessation. The periods of most frequent speleothem growth were between approximately 130,000 and 90,000 yr ago and since about 15,000 yr ago. Such periods before 150,000 yr ago, however, cannot be yet recognized because of a lack of sufficient data and the associated uncertainties of dates in this age range. A comparison with the oxygen-isotope record of deep-sea core V28–:238 shows a clear relationship, indicating that terrestrial calcite formation is controlled by paleoclimatic fluctuations. The evident climatic stimulation of Quaternary calcite formation is readily explained geochemically and is substantiated by the obvious difference in speleothem/travertine growth as a function of geographic position.

A substantially modified history of the last two cycles of Lake Bonneville is proposed. The Bonneville lake cycle began prior to 26,000 yr B.P.; the lake reached the Bonneville shoreline about 16,000 yr B.P. Poor dating control limits our knowledge of the timing of subsequent events. Lake level was maintained at the Bonneville shoreline until about 15,000 yr B.P., or somewhat later, when catastrophic downcutting of the outlet caused a rapid drop of 100 m. The Provo shoreline was formed as rates of isostatic uplift due to this unloading slowed. By 13,000 yr B.P., the lake had fallen below the Provo level and reached one close to that of Great Salt Lake by 11,000 yr B.P. Deposits of the Little Valley lake cycle are identified by their position below a marked unconformity and by amino acid ratios of their fossil gastropods. The maximum level of the Little Valley lake was well below the Bonneville shoreline. Based on degree of soil development and other evidence, the Little Valley lake cycle may be equivalent in age to marine oxygenisotope stage 6. The proposed lake history has climatic implications for the region. First, because the fluctuations of Lake Bonneville and Lake Lahontan during the last cycle of each were apparently out of phase, there may have been significant local differences in the timing and character of late Pleistocene climate changes in the Great Basin. Second, although the Bonneville and Little Valley lake cycles were broadly synchronous with maximum episodes of glaciation, environmental conditions necessary to generate large lakes did not exist during early Wisconsin time.

A detailed chronology of hydrological changes that occurred in northeastern Mali during the Holocene is based on approximately 30 14C dates of molluscs which were collected from the lacustrine and paludal deposits of the Ine Sakane sand sea. The presence of cemented paleodunes indicates that the end of the Pleistocene was and arid period. This was followed by an early Holocene lacustrine episode (ca. 9500–6500 yr B.P.) during which numerous lakes occupied depressions formed in the Cretaceous limestones and between the sand ridges. These lakes reflect a significant rise in the water table. Between ca. 6500 and 5500 yr B.P. the lakes dried up, although the water table remained close to the ground surface in several areas. Calcareous concretions formed at the water table fringe during this time, thereby “fossilizing” some of the middle Holocene dunes. A second lacustrine episode is dated between ca. 5500 and 4500 yr B.P.; it corresponds to the establishment of numerous Neolithic settlements in the area. Arid conditions have developed since about 4500 yr B.P. to attain the conditions of the present day. The isotopic composition (18O and 13C) of the fauna reflects highly variable hydrological conditions. Groundwaters were recharged by precipitation which was occasionally very depleted in heavy oxygen (δ18O ≅ −13%.). Evaporation induced an enrichment in 18O; high concentrations characterize some lakes. The 13C content of fossil shells reflect: (1) species ecological requirements, (2) isotopic composition of the total inorganic dissolved carbon (TIDC) in groundwaters, and (3) the lake metabolism.

Data on glacier, tree-line, tree-ring, pollen, and ice-core variations in North America, Greenland, and Europe during the last 2000 yr (up to A.D. 1800) are compared in detail on the century time scale. Only data that may be indicative of summer temperature changes are included, since these comprise most of the available paleoclimatic information, although some variations (especially of glaciers) may have been in response to precipitation changes instead. Radiocarbon dates and 14C-dated records are converted to calendar (dendrochronological) years using the calibration of M. Stuiver (1982, Radiocarbon 24, 1–26). Despite the basic uncertainties in dating, interpretation, response times, and “noise level” of proxy climatic data, there is at times good agreement among different kinds of evidence from within a region to suggest an episode of generally warmer or cooler summers. Three previously identified episodes find expression in records from all of the regions considered: the “Little Ice Age” of the last few centuries, a “Medieval Warm Period” around the 12th century A.D., and an earlier cold period some time between the 8th and 10th centuries. However, the timing of minima and maxima within these episodes seems to have varied from region to region (although the evidence is consistent within regions). In the 15th century, summers were warm in the eastern Canadian Arctic and southern Greenland while there was a cold episode in Europe and western North America.

Variations of ice-sheet volume during the Quaternary ice ages are simulated using a simple ice-sheet model for the Northern Hemisphere. The basic model predicts ice thickness and bedrock deformation in a north-south cross section, with a prescribed snow-budget distribution shifted uniformly in space to represent the orbital perturbations. An ice calving parameterization crudely representing proglacial lakes or marine incursions can attack the ice whenever the tip drops below sea level. The model produces a large ∼ 100,000-yr response in fair agreement (correlation coefficient up to 0.8) with the δ18O deep-sea core records. To increase confidence in the results, several of the more uncertain model components are extended or replaced, using an alternative treatment of bedrock deformation, a more realistic ice-shelf model of ice calving, and a generalized parameterization for such features as the North Atlantic deglacial meltwater layer. Much the same ice-age simulations and agreement with the δ18O records, as with the original model, are still obtained. The model is run with different types of forcing to identify which aspect of the orbital forcing controls the phase of the 100,000-yr cycles. First, the model is shown to give a ∼ 100,000-yr response to nearly any kind of higher-frequency forcing. Although over the last 2-million yrs the model phase is mainly controlled by the precessional modulation due to eccentricity, over just the last 500,000 yr the observed phase can also be simulated with eccentricity held constant. A definite conclusion on the phase control of the real 100,000-yr cycles is prevented by uncertainty in the deep-sea core time scales before ∼600,000 yr B.P. The model is adapted to represent West Antarctica, and yields unforced internal oscillations with periods of about 50,000 yr.

The thickness of varves in the sediments of Skilak Lake, Alaska, are correlated with the mean annual temperature (r = 0.574), inversely correlated with the mean annual cumulative snowfall (r = −0.794), and not correlated with the mean annual precipitation (r = 0.202) of the southern Alaska climatological division for the years 1907–1934 A.D. Varve thickness in Skilak Lake is sensitive to annual temperature and snowfall because Skilak Glacier, the dominant source of sediment for Skilak Lake, is sensitive to these climatic parameters. Trends of varve thickness are well correlated with trends of mean annual cumulative snowfall ( = −0.902) of the southern Alaska climatological division and with trends of mean annual temperature of the southern ( = 0.831) and northern ( = 0.786) Alaska climatological divisions. Trends of varve thickness also correlate with trends of annual temperature in Seattle and North Head, Washington ( = 0.632 and 0.850, respectively). Comparisons of trends of varve thickness with trends of annual temperature in California, Oregon, and Washington suggest no widespread regional correlation. Trends of annual snowfall in the southern Alaska climatological division and trends of annual temperature in the southern and northern Alaska climatological divisions are reconstructed for the years 1700–1906 A.D. Climatic reconstructions on the basis of varve thickness in Skilak Lake utilize equations derived from the regression of series of smoothed climatological data on series of smoothed varve thickness. Reconstruction of trends of mean annual cunulative snowfall in the southern Alaska climatological division suggests that snowfall during the 1700s and 1800s was much greater than that during the early and mid-1900s. The periods 1770–1790 and 1890–1906 show marked decreases in the mean annual snowfall. Reconstructed trends of the annual temperature of the northern and southern Alaska climatological divisions suggest that annual temperatures during the 1700s and 1800s were lower than those of the early and mid-1900s. Two periods of relatively high annual temperatures coincide with the periods of low annual snowfall thus determined.

The Savanna Terrace, composed of alternating red and gray clayey sediments of late Wisconsinan age, can be found in five states along the upper Mississippi valley from Pepin County, Wisconsin, to Jackson County, Illinois. The terrace is the highest glaciofluvial-lacustrine deposit without a loess cover in the upper Mississippi valley. Chemical, physical, and mineralogical data show that two different sources provided sediment. The red clay is believed to have come from Lake Superior sources, while the gray clay is believed to have come from sources farther west. Large-scale flood events from glacial Lakes Agassiz, Grantsburg, and Superior were probably the main contributors of the sediments. The red clay in the terrace is similar in composition to red glaciolacustrine sediment found in eastern and northern Wisconsin. It also is mineralogically similar to the Hinckley Sandstone and the Fond du Lac Formation, which occur under and around Lake Superior. Radiocarbon dates obtained from the lower Illinois valley indicate that the terrace sediments were deposited sometime between about 13,100 and 9500 yr ago. Soils developed on the terrace are variable in their physical, chemical, and mineralogical properties, which reflect the composition of the clayey sediments.

Rock varnish coats many surfaces of geomorphic and archaeologic interest in arid lands. All varnish dating techniques are limited by the time lag between the exposure of a surface to subaerial processes and the onset of varnishing. They are valid only where manganese is not remobilized after deposition, for example, in most arid environments. The premise of a new age-determination method, cation-ratio dating, is that the ratio of the more mobile cations (e.g., K and Ca) to titanium in varnish decreases with time. Although there are many inherent assumptions and potential limitations, cation-ratio dating has been verified on relative age-sequences from a Death Valley debris cone, Negev Desert talus flatirons, and prehistoric lake levels at Searles Lake in California. Varnish cation ratios have been calibrated to independently dated surfaces in the Coso volcanic field and vicinity in California. Tentative absolute dates have been assigned to geomorphic surfaces in the Coso area. Cation ratios have been used to distinguish relative ages of archaeologic artifacts in southwestern North America and to demonstrate that varnish at the South Stoddard locality, Mojave Desert, did not form in 25 yr.

Global sea-level changes are expressed in the coastal landforms and deposits of northern San Clemente Island. Stratigraphic, radiometric, amino acid, and pedologic dating techniques have allowed the development of a chronology of sea-level changes for about the last 500,000 yr. A uranium-series date on coral of about 127,000 yr for the 2nd terrace serves as a calibration point for amino acid age estimates of four other mapped terraces. Two of these terraces have age estimates of about 80,000–105,000 yr, another has an age estimate of about 127,000 yr, and the 5th terrace on the west side of the island is estimated to be about 415,000–575,000 yr old. These dates correlate reasonably well with marine terraces dated elsewhere and with stages of the oxygen-isotope record that are thought to represent high stands of the sea. Weakly cemented calcareous dune sands (eolianites) are moderately extensive on northern San Clemente Island and appear to represent low stands of the sea, since calcareous shelf sands were the most likely source. A radiocarbon date of about 22,000 yr suggests that the youngest eolianite was deposited during the last glacial maximum. An older eolianite is estimated to be about 140,000–195,000 yr old based on stratigraphic relations and degree of soil development. The suggested ages for the eolianites also correlate well with oxygen-isotope estimates of low sea levels.

Pollen records in the Kootenai and Fisher River drainages in western Montana reveal a fivezone sequence of Holocene vegetation change. Deposition of Glacier Peak Ash-Layer G (ca. 10,540 ± 660 yr B.P.) in the lowermost sediments (clay intermixed with pebbles) at Tepee Lake gives a minimum date for the initiation of sedimentation. Initial vegetation on the newly deglaciated terrain was dominated by Pinus (probably white bark pine) with small amounts of Gramineae, Picea and Abies, reflecting a relatively cool, moist macroclimate. Two vegetation units appear to contribute to Pollen Zone II (ca. 11,000–7100 yr B.P.): arboreal communities with pines, along with Pseudotsuga or Larix, or both, and treeless vegetation dominated by Artemisia. Pollen Zone II represents an overall warmer macroclimate than occurred upon ice withdrawal. After ca. 7100 yr B.P. (Pollen Zone III) diploxylon pines became a major pollen contributor near both Tepee Lake and McKillop Creek Pond, indicating an expansion of xerophytic forest (P. contorta and P. ponderosa) along with an increase in the prominence of Pseudotsuga menziesii or Larix occidentalis, or both. Artemisia briefly expanded coverage near Tepee Lake concomitant with the Mazama ashfall ca. 6700 yr B.P. A short-term climatic trend with more available water began after ca. 4000 yr B.P. as Abies (probably A. grandis) along with Picea engelmannii became a more regular component of the forest surrounding both sites. Emergence of the modern macroclimate is indicated primarily with the first regular appearance of Tsuga heterophylla in the pollen record by ca. 2700 yr B.P., synchronous with the development of western hemlock forest within the same latitudes in northern Idaho and northeastern Washington.

Since 1965, ice-wedge casts have been reported in deposits of sand and gravel in Connecticut. These are wedge forms up to 1.1 m wide and many meters high. Most are single forms, not in polygonal array. They are found in adjoining states as well. Their distribution, dimensions, structure, and fabric and an assessment of the former physical environment preclude their origin as permafrost features. They appear to be tension fractures produced by the loading of coarse clastics on fine clastics near and below the water table where sediments creep toward a stream or depression. Locally movement started with kettle formation during deglaciation. However, some wedges cut horizontal layers of iron-coated sand and gravel and must be younger than those distinctly postglacial phenomena. Moreover, modern B horizons of the overlying soil have moved down into some wedges more than 2 m, indicating that fracturing is still active today. Complex fracture fillings in bedrock also have been attributed to a permafrost origin, but this too seems unlikely.

Asphalt-impregnated sediments at Rancho La Brea (Los Angeles County) and McKittrick (Kern County) in California provide a rich Quaternary insect record. Ages of various sites at Rancho La Brea range from more than 40,000 14C yr B.P. to modern. McKittrick insects studied by W. D. Pierce are not contemporaneous with the late Pleistocene vertebrate fauna, but are only about 7000 14C yr old. The major paleoecological groupings are: (1) ground dwellers, (2) aquatics. (3) scavengers, and (4) miscellaneous. Contrary to conclusions of earlier authors, most specimens represent modern species. Only two apparent terminal Pleistocene extinctions are recognized. both dung beetles (Scarabaeidae).

A new quantitative Soil Development Index based on field data has been applied to chronosequences formed under different climatic regimes. The four soil chronosequences, developed primarily on sandy deposits, have some numeric age control and are located in xeric-inland (Merced, Calif.), xeric-coastal (Ventura, Calif.), aridic (Las Cruces, N. Mex.), and udic (Susquehanna Valley, Pa.) soil-moisture regimes. To quantify field properties, points are assigned for developmental increases in soil properties in comparison to the parent material. Currently ten soil-field properties are quantified and normalized for each horizon in a given chronosequence, including two new properties for carbonate-rich soils in addition to the eight properties previously defined. When individual properties or the combined indexes are plotted as a function of numeric age, rates of soil development can be compared in different climates. The results demonstrate that (1) the Soil Development Index can be applied to very different soil types, (2) many field properties develop systematically in different climatic regimes, (3) certain properties appear to have similar rates of development in different climates, and (4) the Profile Index that combines different field properties increases significantly with age and appears to develop at similar rates in different climates. The Soil Development Index can serve as a preliminary guide to soil age where other age control is lacking and can be used to correlate deposits of different geographical and climatic regions.

Five pollen diagrams reveal late Wisconsin and Holocene vegetation changes in the Walker Lake/Alatna Valley region of the central Brooks Range, approximately 100 km west of the area studied by D. A. Livingstone (1955, Ecology 36, 587–600). New insights into the vegetation history of this region are provided by calculations of pollen influx and by the use of linear discriminant analysis to separate Picea glauca and P. mariana pollen. Three major pollen zones are identified: (1) a basal herb zone, characterized by high percentages of Cyperaceae, Gramineae, Salix, and Artemisia, and low total pollen influx; (2) a shrub Betula zone with increased total pollen influx and very high percentages of Betula pollen, predominantly in the size range of B. nana and B. glandulosa; and (3) and Alnus zone dominated by Alnus pollen. Lakes currently within the boreal forest or near tree line show relatively high percentages of Picea pollen in the Alnus zone. Several striking vegetation changes occurred between ca. 10,000 and 7000 yr B.P. Between ca. 11,000 and 10,000 yr B.P., Populus balsamifera pollen percentages as great as 30% indicate that this species was present at low-elevation sites near Walker Lake. These populations declined abruptly ca. 10,000 yr ago and have never regained prominence. About 8500 yr B.P., Picea glauca pollen reached 10–15%, indicating the arrival of P. glauca in or near the study area. P. glauca populations evidently decreased ca. 8000 yr ago, when Picea pollen percentages and influx fell to low values. About 7000 yr B.P., Alnus pollen percentages and influx rose sharply as alder shrubs became established widely. Picea once more expanded ca. 5000 yr ago, but these populations were dominated by P. mariana rather than P. glauca, which increased slowly at this time and may still be advancing northward. Some vegetation changes have been remarkably synchronous over wide areas of interior Alaska, and probably reflect responses of in situ vegetation to environmental changes, but others may reflect the lagged responses of species migrating into new areas.

As the late Wisconsin Cordilleran Ice Sheet retreated, sediment accumulated in shallow depressions at the Manis Mastodon Archaeological site on the Olympic Peninsula, near Sequim, Washington. Pollen, plant macrofossils, and bones of mastodon, caribou, and bison occur within the lower 47 cm of these deposits. The fossil pollen and seed assemblages indicate persistence for 1000 yr (11,000–12,000 yr B.P.) of an herb-and-shrub-dominated landscape at a time when forest species appear elsewhere in Washington and in adjacent British Columbia.

At present, Sequim is near the northern coastal limits of both Cactaceae and Ceratophyllum. Mean annual precipitation is 42.7 cm and summer temperatures average 15°–16°C in July. The absence of coniferous trees and the presence of cactus and Ceratophyllum in late-glacial sediments are explained by a regional climate that was drier and at least as warm as today. These conditions persisted in the rain shadow of the Olympic Mountains until at least 11,000 yr B.P.

Pingos, or ice-cored mounds, are described for the first time from Antarctica and for the first time on ice-cored moraine. Seven pingos up to 4 m high and 12 m in diameter occur on former dering of lakes on the moraine exposes water-saturated sediments to freezing and consequent lake sediments on Flanders Moraine, Vestfold Hills, Antarctica (68°40'S 78°00'E). Lateral wandering formation of closed-system pingos. The pingos are probably only a few hundred years old. The apparent absence of pingo scars in the Southern Hemisphere may be due to lack of suitable substrates, rather than to unsuitable climatic conditions.

The late Quaternary mammalian zoogeographic history of eastern Washington as revealed by archaeological and paleontological research conforms to a set of past environmental conditions inferred from botanical data. During the relatively cool and moist late Pleistocene and early Holocene, Cervus cf. elaphus, Ovis canadensis, Vulpes vulpes, Martes americana, Alopex lagopus, and perhaps Rangifer sp., taxa with ecological preferences for mesic steppe habitats, were present in the now xeric Columbia Basin. As the climate became progressively warmer and drier during the late Pleistocene and early Holocene, Antilocapra americana, Onychomys leucogaster, Spermophilus townsendii, and Neotoma cinerea, taxa with ecological preferences for xeric steppe habitats, appear in the Columbia Basin. Bison sp. and Taxidea taxus may have been present in eastern Washington for the last 20,000 yr. Middle and late Holocene records for Oreamnos americanus, Spermophilus columbianus, S. townsendii, Lagurus curtatus, and Urocyon cinereoargenteus in central eastern Washington suggest fluctuations in the ranges of these taxa that conform to a middle Holocene period of less effective precipitation and a ca. 3500-yr-old period of more effective precipitation before essentially modern environmental conditions prevailed.

An expanded study of the radiolarian Cycladophora davisiana in late-Pleistocene North Atlantic marine sediments shows that over the last several hundred thousand years this species exhibits large variations in relative abundance. The C. davisiana curves in the North Atlantic cores are quite similar, with easily recognizable features common to all records. Minor deviations from the general pattern of this species' abundance apparently reflect the response of C. davisiana to specific oceanographic conditions characteristic of a particular area within the North Atlantic. C. davisiana occurs today in high abundance (>20%) only in the Sea of Okhotsk. Extensive winter and early spring sea-ice cover coupled with low surface-water salinities during summer and fall is responsible for maintaining near-freezing subsurface temperatures in this northwest Pacific marginal sea as well as relatively stable temperatures and salinities at depths below a shallow subsurface temperature minimum. During periods in the late Pleistocene, high C. davisiana abundances (>20%) in the North Atlantic were probably associated with oceanographic properties similar to those that exist in the Sea of Okhotsk today. Because of the relationship between relatively stable subsurface temperatures and salinities and high abundance levels of C. davisiana, analysis of this species' abundance pattern at several locations throughout the high-latitude North Atlantic should assist in identifying source areas of deep-water formation and determining the duration of deep convective processes at these sites.

The pulse of large mammal extinctions at the close of the last ice age was a pronounced ecological catastrophe in the Western Hemisphere. Available paleoenvironmental, paleontological, and archeological data are largely concordant with the “overkill” hypothesis proposing that human predation was the cause of these extinctions. Climatic explanations fail to postulate specific mechanisms accounting for the event's timing, differential impact, geographic extent, and stratigraphic singularity. Lack of provisional acceptance of the overkill hypothesis, when no viable alternatives exist, may be in part because this unconventional hypothesis violates a strong sociocultural bias, the image of prehistoric man as the “noble savage”.