Loess is one of the most widespread subaerial deposits in Alaska and adjacent Yukon Territory and may have a history that goes back 3 Ma. Based on mineralogy and major and trace element chemistry, central Alaskan loess has a composition that is distinctive from other loess bodies of the world, although it is quartz-dominated. Central Alaskan loess was probably derived from a variety of rock types, including granites, metabasalts and schists. Detailed stratigraphic data and pedologic criteria indicate that, contrary to early studies, many palaeosols are present in central Alaskan loess sections. The buried soils indicate that loess sedimentation was episodic, or at least rates of deposition decreased to the point where pedogenesis could keep ahead of aeolian input. As in China, loess deposition and pedogenesis are likely competing processes and neither stops completely during either phase of the loess/soil formation cycle. Loess deposition in central Alaska took place before, and probably during the last interglacial period, during stadials of the mid-Wisconsin period, during the last glacial period and during the Holocene. An unexpected result of our geochronological studies is that only moderate loess deposition took place during the last glacial period. Our studies lead us to conclude that vegetation plays a key role in loess accumulation in Alaska. Factors favouring loess production are enhanced during glacial periods but factors that favour loess accumulation are diminished during glacial periods. The most important of these is vegetation; boreal forest serves as an effective loess trap, but sparsely distributed herb tundra does not. Thus, thick accumulations of loess should not be expected where tundra vegetation was dominant and this is borne out by modern studies near the treeline in central Alaska. Much of the stratigraphic diversity of North American loess, including that found in the Central Lowlands, the Great Plains, and Alaska is explained by a new model that emphasizes the relative importance of loess production factors versus loess accumulation factors.

Peoria Loess‐derived soils on uplands east of the Mississippi River valley were studied from Louisiana to Iowa, along a south‐to‐north gradient of decreasing precipitation and temperature. Major element analyses of deep loess in Mississippi and Illinois show that the composition of the parent material is similar in the northern and southern parts of the valley. We hypothesized that in the warmer, wetter parts of the transect, mineral weathering should be greater than in the cooler, drier parts of the transect. Profile average values of CaO/TiO2, MgO/TiO2, K2O/TiO2 and Na2O/TiO2, Sr/Zr, Ba/Zr, and Rb/Zr represent proxies for depletion of loess minerals such as calcite, dolomite, hornblende, mica, and plagioclase. All ratios show increases from south to north, supporting the hypothesis of greater chemical weathering in the southern part of the valley. An unexpected result is that profile average values of Al2O3/TiO2 and Fe2O3/TiO2 (proxies for the relative abundance of clay minerals) show increases from south to north. This finding, while contrary to the evidence of greater chemical weathering in the southern part of the transect, is consistent with an earlier study which showed higher clay contents in Bt horizons of loess‐derived soils in the northern part of the transect. We hypothesize that soils in the northern part of the valley received fine‐grained loess from sources to the west of the Mississippi River valley either late in the last glacial period, during the Holocene or both. In contrast, soils in the southern part of the valley were unaffected by such additions.

Geomorphologists have long recognized that eolian sand transport pathways extend over long distances in desert regions. Along such pathways, sediment transport by wind can surmount topographic obstacles and cross major drainages. Recent studies have suggested that three distinct eolian sand transport pathways exist (or once existed) in the Mojave and Sonoran Desert regions of the southwestern United States. One hypothesized pathway is eolian sand transport from the eastern Mojave Desert of California into western Arizona, near Parker, and would require sand movement across what must have been at least a seasonally dry Colorado River valley. We tested this hypothesis by mineralogical, geochemical and magnetic analyses of eolian sands on both sides of the Colorado River, as well as sediment from the river itself. Results indicate that dunes on opposite sides of the Colorado River are mineralogically distinct: eastern California dunes are feldspar-rich whereas western Arizona dunes are quartz-rich, derived from quartz-rich Colorado River sediments. Because of historic vegetation changes, little new sediment from the Colorado River is presently available to supply the Parker dunes. Based on this study and previous work, the Colorado River is now known to be the source of sand for at least three of the major dune fields of the Sonoran Desert of western Arizona and northern Mexico. On the other hand, locally derived alluvium appears to be a more important source of dune fields in the Mojave Desert of California. Although many geomorphologists have stressed the importance of large fluvial systems in the origin of desert dune fields, few empirical data actually exist to support this theory. The results presented here demonstrate that a major river system in the southwestern United States is a barrier to the migration of some dune fields, but essential to the origin of others. Published by Elsevier Science Ltd.