We reconstruct the provenance of aluminosilicate sediment deposited in Ulleung Basin, Japan Sea, over the last 12 Ma at Site U1430 drilled during Integrated Ocean Drilling Program Expedition 346. Using multivariate partitioning techniques (Q-mode factor analysis, multiple linear regressions) applied to the major, trace and rare earth element composition of the bulk sediment, we identify and quantify four aluminosilicate components (Taklimakan, Gobi, Chinese Loess and Korean Peninsula), and model their mass accumulation rates. Each of these end-members, or materials from these regions, were present in the top-performing models in all tests. Material from the Taklimakan Desert (50–60 % of aluminosilicate contribution) is the most abundant end-member through time, while Chinese Loess and Gobi Desert components increase in contribution and flux in the Plio-Pleistocene. A Korean Peninsula component is lowest in abundance when present, and its occurrence reflects the opening of the Tsushima Strait at c. 3 Ma. Variation in dust source regions appears to track step-wise Asian aridification influenced by Cenozoic global cooling and periods of uplift of the Tibetan Plateau. During early stages of the evolution of the East Asian Monsoon, the Taklimakan Desert was the major source of dust to the Pacific. Continued uplift of the Tibetan Plateau may have influenced the increase in aeolian supply from the Gobi Desert and Chinese Loess Plateau into the Pleistocene. Consistent with existing records from the Pacific Ocean, these observations of aeolian fluxes provide more detail and specificity regarding the evolution of different Asian source regions through the latest Cenozoic.

A 4000-yr sediment core record from Lake Salpetén, Guatemala, provides evidence for Maya-induced forest clearance and consequent soil erosion between ∼1700 cal yr B.C. and 850 cal yr A.D. Radiocarbon ages of wood, seeds, and charcoal support an age-depth model with average errors of ±110 cal yr. Relatively low carbonate δ18O values between 1300 and 400 cal yr B.C. coincide with pollen evidence for forest loss, consistent with increased surface and groundwater flow to the lake. Minimum δ18O values between 400 cal yr B.C. and 150 cal yr A.D. suggest a high lake level, as do 14C-dated aquatic gastropods as much as 7.5 m above the present lake stage. High lake levels resulted from reduced evaporation-to-precipitation ratios, increased hydrologic input caused by anthropogenic deforestation, or both. The Preclassic abandonment (150 A.D.) and Early Classic/Late Classic boundary (550 A.D.) are marked by relatively high δ18O values indicating reduced lake levels. Oxygen isotope composition increased further coincident with the Terminal Classic Maya demographic decline between 800 and 900 A.D. This period of high δ18O may have been caused by the greater aridity that has been documented in northern Yucatán lakes or by decreased hydrologic input to the lake as a consequence of forest recovery. Reduced soil erosion after 850 cal yr A.D. coincided with the Terminal Classic Maya demographic decline and permitted forest recovery and resumption of organic sedimentation.

Lake Cobá, within the archaeological site of Cobá, provides evidence bearing on lowland Maya development. Palynological and geochemical data record multidecadal precipitation cycles from a 8.80-m, >8370-yr lake-sediment sequence terminating on bedrock. Late Classic sedimentation rates are rapid, but an anthropogenically derived colluvium layer is lacking. Initial vegetation was medium semi-deciduous and swamp forest. Forest clearance began 1650 B.C. (Early Preclassic) and maize first occurred at 850 B.C. (Middle Preclassic). Lakeside milpas existed until A.D. 720 (Late Classic) and then were moved from the city center as urbanization intensified and Lake Cobá was diked as a reservoir. Cobá was at most briefly vacated during the Classic Collapse and was abandoned after A.D. 1240, although some habitation persisted. The paleoecological record matches the archaeological history for Cobá, but pervasive disturbance muted the climatic signal, as the Late Classic drought is barely evident. The question whether economic trees were maintained within the city is unresolved. Maize cultivation allowed the Maya to develop a complex society and support a large population, but dependence on maize was ultimately doomed by variable rainfall. Precipitation in extreme years was insufficient to support crops, while native vegetation was not directly affected by drought that devastated Maya agriculture.