The evolution of arid environments in northern China was a major environmental change during the Quaternary. Here we present the dating and environmental proxy results from a 35 m long core (A-WL10ZK-1) collected from the Ulan Buh Desert (UBD), along with supplemental data from four other cores. The UBD is one of the main desert dune fields in China and our results indicate the UBD has undergone complex evolution during the late Quaternary. Most of the present UBD was covered by a Jilantai-Hetao Mega-paleolake lasting until ~ 90 ka ago. A sandy desert environment prevailed throughout the UBD during the last glacial period and early Holocene. A wetland environment characterized by the formation of numerous interdunal ponds in the northern UBD occurred at ~ 8–7 ka, although a dune field persisted in the southern UBD. The modern UBD landscape formed after these wetlands dried up. During the last 2000 years, eolian sand from the Badain Jaran Desert has invaded the northern UBD, while farming and overgrazing resulted in the formation of the eastern UBD. We suggest that the formation of UBD landforms is related to the disintegration of the megalake Jilantai-Hetao and to summer monsoon changes during the last glaciation and Holocene.

The interpretation of trace element/calcium ratios of speleothems as indicators of local hydroclimatic variability in the vicinity of caves has led to controversy in reconstructing the evolution of moisture conditions in arid central Asia (ACA) during the Holocene. Here we present records of Mg/Ca, Sr/Ca, Ba/Ca, and U/Ca from precisely dated stalagmites from Baluk cave in Xinjiang (northwest China), spanning the past 9370 years. The co-variations of the trace element ratios, together with the slopes of the regression lines of the corresponding logarithmically transformed data, suggest that they are dominated by prior calcite precipitation (PCP) and thus can be used as reliable proxies of changes in moisture/precipitation. The trace element ratios are relatively high during ~9 to 5 ka and lower from 5 ka to the present, indicating a trend of increasing mid–late Holocene moisture in ACA. The long-term trend of variation of the trace element ratios is correlative with two other records of speleothem trace element ratios from caves in ACA: Kesang cave (western Xinjiang) and Ton cave (Uzbekistan). This spatial coherency of the trend of inferred moisture conditions from three caves that are separated by hundreds of kilometers demonstrates that speleothem trace element ratios are indicative of large spatial scale rather than local hydroclimatic variability in ACA during the Holocene. However, the long-term trend of variation of the trace element ratios is the inverse of the corresponding oxygen isotope (δ¹⁸O) records from the same cave sites, which implies that Holocene speleothem δ¹⁸O records do not represent changes in the precipitation amount in ACA; rather, they most likely reflect moisture sources and related water vapor transport controlled by Northern Hemisphere summer insolation (NHSI). Our findings provide new evidence for a ‘westerlies-dominated climatic regime’, which influenced hydroclimatic changes in ACA during the Holocene.