Kathryn E. Fitzsimmons, Tobias Sprafke, Christoph Zielhofer, Christina Günter, Jean-Marc Deom, Renato Sala, Radu Iovita

• Whilst correlations have been made between the loess of Europe and China, deposits in Central Asia have remained largely overlooked by scientific investigation. The nature of the relationship between loess accumulation and palaeoclimate at the core of the Eurasian loess belt is particularly poorly understood. Here we reconstruct palaeoenvironmental change in Central Asia over the last 40 ky based on data from the Remizovka loess profile, in the northern foothills of the Tian Shan in southern Kazakhstan. Our interpretations are based on synthesis of chronostratigraphic, colour and magnetic susceptibility data, supported by chronostratigraphies from two additional sites nearby, Maibulak and Valikhanova. All three sites record substantially increased loess accumulation during late MIS 3 into the global last glacial maximum (gLGM). At Remizovka, increased loess flux occurred in two pulses at c. 38-25 ka and 22-18 ka, with the intervening period involving incipient pedogenesis. At Maibulak, two loess pulses at c. 40-30 ka and c. 28-22 kaWhilst correlations have been made between the loess of Europe and China, deposits in Central Asia have remained largely overlooked by scientific investigation. The nature of the relationship between loess accumulation and palaeoclimate at the core of the Eurasian loess belt is particularly poorly understood. Here we reconstruct palaeoenvironmental change in Central Asia over the last 40 ky based on data from the Remizovka loess profile, in the northern foothills of the Tian Shan in southern Kazakhstan. Our interpretations are based on synthesis of chronostratigraphic, colour and magnetic susceptibility data, supported by chronostratigraphies from two additional sites nearby, Maibulak and Valikhanova. All three sites record substantially increased loess accumulation during late MIS 3 into the global last glacial maximum (gLGM). At Remizovka, increased loess flux occurred in two pulses at c. 38-25 ka and 22-18 ka, with the intervening period involving incipient pedogenesis. At Maibulak, two loess pulses at c. 40-30 ka and c. 28-22 ka are separated by a weakly developed paleosol which may date to the same time as pedogenesis at Remizovka. There is additional possible periglacial influence at Maibulak from c. 40-33.5 ka. At Valikhanova, there is some age overlap between paleosol and loess samples, but overall loess accumulation appears to have increased at c. 42-35 ka, c. 30 ka and the gLGM, with pedogenesis occurring >40 ka and c. 32 ka. At all three sites, Holocene loess accumulation is minimal; this period is characterised by pedogenesis. The chronostratigraphic variability between our sites highlights a need to interrogate climate-driven models for loess formation in piedmont environments. We interpret our data in the context of regional palaeoenvironmental archives to indicate that loess accumulation increased coeval with MIS 3 glacial advance in the Tian Shan, which was facilitated by northward expansion of the Asian monsoon and associated increase in precipitation. We hypothesise that increased ice volume impeded teleconnections with the temperate zone westerlies to the north; these were compressed against the piedmont resulting in increased wind strength and facilitating increased loess flux. Peak loess accumulation during the gLGM occurred under colder, drier climatic conditions, with reduced but sustained glacial ice volume and persistent influence of the westerlies in the arid Central Asian piedmont loess belt. In the absence of more widespread, reliably dated palaeoenvironmental records from the region, our data become of critical importance for understanding past environmental conditions in Central Asia, relative to elsewhere in Eurasia and globally.…