Geochemical evidence for the provenance of middle Pleistocene loess deposits in southern China

Abstract

The provenance of middle Pleistocene loess in the middle and lower reaches of the Yangtze River, the most intensively investigated loess deposits outside the Loess Plateau region in China, remains controversial. Identification of the provenance will provide crucial insight into the environmental implications of this valuable sedimentary archive, and into the potential role of the East Asian winter monsoon in transporting the dust from deserts in the Asian interior. In this study, geochemistry was used to compare the provenance of loess in the lower reaches of the Yangtze River in southern China with that inferred for loess deposits on the Loess Plateau in northern China. Compared with samples from the Loess Plateau, the <20 μm fraction in the loess deposits of southern China has higher TiO 2/Al 2O 3, Zr/Nb, Zr/Al, Zr/Ti, Zr/Hf, Y/Al and La N/Sm N ratios, and lower Eu/Eu*, Th/Nb, Y/Nb and Al/Nb ratios. The clear distinction in immobile element ratios between samples from the two regions indicates that the loess deposits in the two regions have different provenances. The inferred difference in source area is also supported by variations in the major element composition of bulk samples obtained in this study and collected from published data. These lines of evidence indicate that the deserts in the Asian interior are not the primary provenance for the southern loess. It is suggested that the adjacent floodplains to the north of Yangtze River are the dominant dust sources, and the occurrence of sustained loess deposits in the lower reaches of the Yangtze River, currently an area of northern subtropical climate, is an indication of local aridification and strengthened winter monsoon activity during glacial periods as a regional response to the Middle Pleistocene climate transition around 0.8 Ma. The role of the East Asian winter monsoon in transporting the dust from northern deserts to southern China has been overestimated in previous studies.