Absolute 230Th/U chronologies and Δ47 thermometry paleoclimate reconstruction from soil carbonates in Central Asian loess over the past 1 million years

Abstract

Pleistocene loess records of the Khovaling Loess Plateau (KLP) in Tajikistan provide rich collections of lithic artifacts demonstrating past human presence in the region. To understand the timing of human activity and environmental conditions prevailing at that time U–Th dating and clumped/stable C/O isotope measurements have been applied to modern and Pleistocene soil carbonates (SCs) collected at several sites on the KLP and surroundings. U–Th ages were corrected by two methods: 1) assuming an initial [²³⁰Th/²³²Th] activity ratio of 0.85 ± 0.25 based on gamma spectrometry of loess/paleosol samples, and 2) the isochron technique using leachates and fully dissolved subsamples. Diagenetic alteration and potential U/Th mobilization and related isotope fractionation due to alpha-recoil was also modelled and found to be minor in the studied soil carbonates. Compared to model ages as references, uncorrected ²³⁰Th ages are only acceptable if measured [²³⁰Th/²³²Th] activity ratios of leachates are high (>30), while ²³⁰Th ages derived using method 1 are mostly overcorrected. It appears that SCs can be reliably dated by the U-series disequilibrium method in this sedimentary setting, but isochron dating cannot be spared. Application of the isochron method is required to derive ²³⁰Th_model ages, which ensures that the non-zero initial ²³⁰Th and possible U–Th gain/loss due to alpha-recoil can be simultaneously corrected and reliable U–Th ages obtained. U–Th ages of Pleistocene SCs clearly demonstrate post-pedogenic ingrowth of multiple, non-contemporaneous populations of SCs within loess/paleosol units, and that SC formation happened in many cases under cold, presumably dry glacial climate conditions. Considering that U–Th ages of SCs provide minimum ages of the sediment in which they form, these ages can be useful in developing loess stratigraphic models and for correlation of paleosols with marine isotope stages. This implies that the age of a given paleosol and any lithic artifacts it may contain, indicating human activity, cannot be younger than the age of SCs formed in that paleosol. This is due to the nature of soil carbonates, which can be the product of both syn- and post-depositional processes.

Clumped isotope thermometry of SCs collected from modern soils at three sites in Tajikistan provide evidence for SCs dominantly recording summer season soil temperatures, while the calculated soil water oxygen isotope signatures reflect annual signals and carbonate precipitation from source waters incorporating rainfall from prior to and during SC formation. In contrast, some Pleistocene SCs record soil temperatures and stable isotope compositions more appropriate to glacial conditions, confirming the findings of U–Th ages, and highlighting the primary role of aridity-driven soil moisture changes in SC precipitation in this setting. Considering the interpretative complexities of SC stable isotope compositions, involving issues such as SC formation depth within a soil/paleosol profile, seasonality of SC growth and violation of the law of superposition, SC stable isotope proxy records of past climates cannot be considered as a set of clearly sequential data through time. This implies that such SC-based stable isotope records must be accompanied by U–Th dating of carbonates to be meaningful.