Comparison with neighbouring regions

The western foreland of the Central Asian mountains features one of the most complete Quaternary loess-palaeosol records of the world (Bronger et al., 1998, Dodonov, 2002). In this record, warm, moist climate phases are indicated by pedocomplexes (PC), while cold, dry phases are indicated by unweathered loess. The record of the last glacial cycle begins with the first pedocomplex (PC1), which is associated with the MIS 5. The luminescence age of PC1 at Darai Kalon, western Tajikistan, has been measured to between >117 ka at the base and 60-80 ka at the top (Frechen & Dodonov, 1998). Mestdagh et al. (1999), using soil micromorphology, have shown that during this time precipitation in the region has dropped from ~700-800 mm at the beginning of soil development to below 400 mm at its end. During the rest of the last glacial, conditions have been adverse for soil formation, as loess deposited during the MIS 3 is only slightly weathered and does not contain a fully de-veloped palaeosol (Frechen & Dodonov, 1998). Mestdagh et al. (1999) attribute this in-creasing drought to a successive strengthening of the Siberian anticyclone. The same has been inferred from the successively more constrained glaciation in the Siberian Arctic in

the course of the last glacial cycle (Velichko et al., 1997, Svendsen, 2003), and has also been described by a climate model (Krinner et al., 2003). The successive drying-out of the Central Asian mountain foreland seems to be reflected by the decreasing extent of gla-ciation which is documented in the Pamir.

5.3.6.2 Kunlun Shan & Tibetan Plateau

In the Kunlun Shan, southeast of the Pamir, the most extensive glacier advances date from the early and middle Pleistocene. One of these advances may correlate with the advance having deposited the isolated boulders left today above 4400 m in the Pamir (e.g. UK1, GU1) or above the highest intact lateral moraine in the Aksu Valley (AK1). In the Kunlun Shan, thermoluminescence (TL) ages of 700-500 ka (Wu et al., 2001), as well as TL and ESR ages around 330 ka, and around 206 ka have been reported from early and middle Pleistocene deposits (Zheng & Rutter, 1998). Late Pleistocene advances in the Kunlun were dated to 67 cal. ka B.P. (TL from correlated sandy loess, Zheng & Rutter, 1998), be-tween 23 and 21 ¹⁴C ka B.P. (27-24 cal. ka B.P.), and between 18 and 16 ¹⁴C ka B.P. (ca.

21-19 cal. ka B.P., Derbyshire et al., 1991). The last dates have been confirmed by other workers dating a correlative advance between 16 and 18 ¹⁴C ka B.P. as well (Li & Shi 1992, cited in Gillespie & Molnar, 1995). These late Pleistocene ages are in excellent agreement with the glacier advances from the Pamir dated in this work.

Thompson et al. (1997) have provided an ice core δ¹⁸O record from the Guliya ice cap in the Kunlun Shan, which can serve as a temperature indicator for the northwestern part of High Asia. It has recently been shown to have been influenced mostly by the westerly cir-culation during the Holocene, in contrast to the Dunde ice core record from the Qilian Shan in northeastern Tibet (Thompson et al., 1989), which has been influenced mostly by the Asian monsoon (He et al., 2004). In the late Pleistocene, a rough correlation with 30°N in-solation (Berger & Loutre, 1991) is noticeable (Fig. 5.12). The Guliya record shows pro-nounced cold phases at 75-62 cal. ka B.P., 52-45 cal. ka B.P. and 32-15 cal. ka B.P., with which glacial activity in the Pamir most probably is contemporaneous (Fig. 5.12).

In the Tanggula Shan in east-central Tibet, ¹⁰Be exposure ages of >125-130 ka and 58-70 ka have been determined for two successively less extensive moraines (Schaefer et al., 2002, recalculated data). The younger ages are synchronous with the oldest late Pleistocene Pamir ages found in this study (Fig. 5.10). In northeastern Tibet, Owen et al. (2003a, b)

have found exposure ages on late Pleistocene moraines of 22-23 ka, 18-20 ka, and 15-16 ka, again in excellent agreement with dated advances in the Pamir (Fig. 5.11).

5.3.6.3 Hindu Kush, Karakoram & northwestern Himalaya

In the Hindu Kush, southwest of the Pamir, the most extensive Drosh glaciation, with an ELA depression of 1200 m, has been dated by Kamp (1999) to have occurred during the MIS 6 (ages >200-149 ka), using optically stimulated luminescence (OSL). Another advance with an ELA depression of 1000 m, the Pret stage, has been OSL-dated to 54-43 ka. A third, not very well defined stage called Sonoghar has been assumed to have occur-red during the MIS 2. Kamp's (1999) OSL data, however, have been corrected by Owen et al. (2002b), placing the Drosh advance into the MIS 3 (55-31 cal. ka B.P.), the Pret stage into the MIS 2 (>7.7 cal ka B.P.) and the Sonoghar stage into the Lateglacial (Kamp et al., 2003). If these recalculated ages are correct, the early late Pleistocene glacier advance in the Hindu Kush is antiphased with respect to that in the Pamir. Owen et al. (2002b) and Kamp et al. (2003) accordingly argue for a strong monsoonal influence on glaciation in the Hindu Kush. A comparison of OSL and ¹⁰Be surface exposure ages in the neighbouring Karakoram by Spencer & Owen (2004), however, has shown that OSL ages as calculated by L. A. Owen's group may be significantly lower than ¹⁰Be ages of the same deposits. The revised chronology of the Hindu Kush therefore remains in doubt, and should be tested by an SED study.

The glacial chronology of the Karakoram, south-southeast of the Pamir, has been defined in the Hunza Valley. The recalculated ¹⁰Be ages of Owen et al. (2002c) show that the Borit Jheel stage has occurred prior to 47-57 ka (Fig. 5.10), which is corroborated by an earlier TL age of >50-65 ka (Derbyshire et al., 1984). The MIS 2 stages Ghulkin I, II and Batura at the Pasu glacier accordingly occurred 27-20, 18, and ~14.5 cal. ka B.P. (Fig. 5.11, 5.12).

These ages agree well with our results from the Pamir.

In the Swat Himalaya, Richards et al. (2000) have measured an OSL age of ~77 ka for the maximum glacier advance (Gabral), and an OSL age of >38 ka for the next younger one (Kalam I). In the upper Indus Valley, the maximum stage (Shatial) yielded an OSL age of

~60 ka, while three younger and smaller advances yielded OSL ages of 27 ka, 21-23 ka, and 15 ka (Richards et al., 2000). Phillips et al (2000), using ¹⁰Be SED, have dated the maximum late Pleistocene advance at the Nanga Parbat to 52-65 cal. ka B.P. (ages recalcu-lated), while one boulder from a younger moraine gave an age of ~20 ka. This 20 ka age

indicates an MIS 2 advance, as suggested by Richards et al. (2001), if the other Holocene ages from this deposit are interpreted as degradation ages. Again, the agreement with the Pamir data is good, especially for the MIS 2 ages. Even in the Zanskar Range and the Garhwal Himalaya, both still further southeast, the maximum late Pleistocene stages have OSL ages of 40-78 cal. ka (Batal stage, Taylor & Mitchell (2000), and around 63 ka (Bhagirathi stage, Sharma & Owen, 1996), respectively.