Detailed quartz and feldspar luminescence chronologies of loessic Palaeolithic sites in Tajikistan

Over the past ~ 2.1 Ma, an intense dust flux has been a major component of the climate of the Khovaling region in Southern Tajikistan, Central Asia, leading to the formation of loess deposits up to 200 m thick. These deposits record palaeoclimatic and palaeoenvironmental changes throughout the Quaternary. During interstadials, soil formation produces dark brown palaeosols, while stadials are characterised by dust accumulation under cold and arid conditions, forming yellowish-light brown loess units. Superposed palaeosols formed within a single interglacial period are referred to as a pedocomplex (PC). Thus alternation of loess and pedocomplexes in the Khovaling plateau is taken to reflect glacial-interglacial cycles. Chronostratigraphies of these deposits typically rely on correlation-based methods, which assume continuous sediment deposition. It is known from elsewhere, particularly in China, that large discontinuities in loess sedimentary sequences exist (sometimes of more than an entire glacial cycle); this failure of the assumption of continuity casts serious doubt on the reliability of chronologies based on correlative methods, and on any climate interpretations based on proxy records within the loess.

The Khovaling plateau is also a key region for studying human dispersals, with several Lower and Middle Palaeolithic sites excavated in Southern Tajikistan. The Lower Palaeolithic industries of Kuldara are estimated to be ~ 0.8-0.9 Ma (PCs 12-11), and other notable Loessic Palaeolithic sites date from ~ 0.6-0.4 Ma (PCs 6-4). In Khonako III, Middle Palaeolithic industries were found in PC2, L2, and PC1. Accurate dating of these sites is essential for understanding the timing of human occupation in Central Asia; if correlation-based chronostratigraphies are unreliable, independent numerical dating methods become essential.

Luminescence dating is widely used to date loess deposits and relies on the ability of quartz and feldspar crystals to store energy, and for that stored energy to be released by exposure to light or heat. This study applied high spatial resolution luminescence dating to three loess sections in Southern Tajikistan — Khonako II, Kuldara, and Khonako III — using Optically Stimulated Luminescence (OSL) and post-IR IRSL (pIRIR) dating. We expected the luminescence chronological range to be limited to the PC2-L3 units. Thus, the upper part of the Kuldara section, and the entirety of the Middle Palaeolithic industries of Khonako III, fall within our dating range.

In Khonako II, we compared quartz OSL and polymineral pIRIR_200,290 chronologies down to the middle of the L1 unit. The pIRIR_200,290 residual dose was calculated based on the offset between the quartz OSL and pIRIR_200,290 ages and through comparisons with modern analogues from Tajikistan and Uzbekistan. After correcting for the residual dose, we identified an accumulation period from ~ 39 to ~ 3-4 ka, with a hiatus of ~ 7-9 ka at the beginning of the Holocene.

In Kuldara, the luminescence study revealed significant breaks in the upper section, with material from Marine Isotopic Stage (MIS) 1, MIS 2, and part of MIS 3 missing. Loess L2 and the upper part of PC2 underwent major deflation, leading to the loss of ~ 110 ka of deposits, and L3 and PC3 were also lost. Despite these disturbances, the chronology confirmed the attribution of PC2 to MIS 7, L2 to MIS 6, and PC1 to MIS 5.

In Khonako III, extensive testing of the pIRIR_200,290 signal showed generally satisfactory luminescence characteristics, but with a notable dependence of the dose on the size of the test dose. However, several limitations in the Khonako III study were highlighted: the pIRIR_200,290 chronology appeared to suffer from systematic underestimation, likely because of water content assumptions, and the magnetic susceptibility record seemed slightly shifted compared to the stratigraphic record. After correcting for this, the chronology suggests continuous accumulation from ~ 230 to ~ 18 ka, with Middle Palaeolithic industries attributed to MIS 7, MIS 6, and MIS 5.

We conducted a comparative study of three age-depth models (Bacon, CosmoChron, and ArchaeoPhases) using the Khonako III dataset to assess their performance with high spatial resolution data. The models were tested for sensitivity to section truncation (shortened stratigraphic sections) and changes in sampling resolution. ArchaeoPhases proved less suitable for our dataset, likely due to computational limits encountered when calculating individual Bayesian luminescence ages using BayLum. While Bacon and CosmoChron models generally produced consistent results, differences in Dust Accumulation Rate (DAR) values were observed, with all three models showing sensitivity to sampling resolution. This study advises caution when interpreting individual DAR values, emphasizing the importance of focusing on broader trends.

We calculated DAR for all three sections examined here. MIS 6 appears to have been characterised by an intense accumulation peak in Khonako III and deflation in Kuldara, presumably reflecting unusually strong wind activity in this period. The interglacial periods MIS 7 and MIS 5 are marked by a general decrease in dust accumulation. Dust accumulation increased again at the end of MIS 5 and the beginning of MIS 4. The dust flux was also intense during MIS 3 and MIS 2, before decreasing during the Holocene.

Overall, this work establishes a detailed chronological framework for ongoing palaeoclimatic and palaeoenvironmental studies and documents continuous dust activity spanning the past ~ 250 ka, with the Khonako III section providing new insights into human occupation during the Middle Palaeolithic in Central Asia.