The role of the westerlies and orography in Asian hydroclimate since the late Oligocene

Central Asia constitutes the largest extratropical arid zone on Earth. However, during the early Cenozoic, most of this region was warmer and wetter than present day and periodically occupied by the Paratethys, an epicontinental seaway that stretched from the Atlantic Ocean to the Tarim Basin. Numerous studies suggest a complex aridification history of Central Asia during the Cenozoic, but the mechanisms responsible for this drying trend remain controversial (An, 2014).

Asian aridification has been related to the retreat of the Paratethys (Ramstein et al., 1997), the uplift of the Tibetan Plateau (Liu et al., 2015), and/or global cooling (Lu et al., 2010). Recent work suggests that interactions between the westerlies and the Pamir–Tian Shan played an important role in controlling hydroclimate changes over Central Asia (Caves et al., 2015; Heermance et al., 2018). However, the timing when this topographic-atmospheric framework was established remains poorly constrained.

Precipitation in Central Asia concentrates in boreal winter seasons (Fig. S1in the Supplemental Material¹), and its variability is mainly influenced by the location and intensity of the westerlies (Chen et al., 2019). Blocked by the Pamir and Tian Shan, the modern westerlies are split into two branches, with precipitation higher on the western flanks (windward) of the Pamir–Tian Shan and significantly lower on the eastern side (leeward; Fig. 1B). Seasonally, the regions west of these mountains have winter-spring precipitation maxima, whereas regions to the east have summer precipitation maxima (Fig. S1). These distinct west-east hydroclimate differences are closely related to the orographic barrier, which intercepts moisture in the westerlies, blocks winter precipitation, and enhances subsidence over the leeward side (Baldwin and Vecchi, 2016).

The Pamir formed as a result of accretion of allochthonous terranes to Eurasia during the Mesozoic and subsequent significant uplift during the Cenozoic (Chapman et al., 2019). The Tian Shan is a Paleozoic orogen that was reactivated during the late Cenozoic (Sobel and Dumitru, 1997). The Tajik and Tarim Basins (Fig. 1C) are foreland basins related to the uplift of the Pamir that developed by no later than ca. 40 Ma (Carrapa et al., 2015; Chapman et al., 2019), as supported by geochemical and geological data from the Tarim Basin indicating high elevations in the eastern Pamir by no later than Oligocene time (Bershaw et al., 2012). Cenozoic sedimentary sequences in these basins (Fig. S2) provide ideal archives for understanding paleoclimate changes in Central Asia. Although a connection between the two basins through what is now the Alai valley existed throughout the Cenozoic, the growth of the Pamir (Wang et al., 2019; Blayney et al., 2019; Chapman et al., 2019) and of the western Tian Shan (Bande et al., 2017; Jepson et al., 2018) as early as the Oligocene suggests that the westerly atmospheric flow was disrupted by high topography in the region beginning in the early Cenozoic.

We applied stable oxygen isotope geochemistry to well-dated Eocene–Miocene sedimentary rocks from the Tajik Basin and to Eocene sedimentary rocks from the western Tarim Basin (Fig. 2). We then integrated our results with published stable oxygen isotopic data from adjacent regions (Kent-Corson et al., 2009; Zhuang et al., 2011; Charreau et al., 2012; Caves et al., 2014, 2017; Macaulay et al., 2016) to evaluate regional precipitation changes across topography. We also conducted elemental geochemical and clay mineralogical analyses of siltstone samples (modal grain size: 5–63 μm) from the Tajik Basin as a proxy for weathering in the basin and its catchment. Finally, we performed climate simulations, using the Atmosphere General Circulation Model (Stepanek and Lohmann, 2012), to evaluate the impact of topographic changes in the Pamir and Tian Shan on Central Asia climate. Detailed methods are described in the Supplemental Material.

The δ¹⁸O values (relative to Vienna Peedee belemnite [VPDB]) of sandstone and mudstone cements from Cenozoic nonmarine sedimentary rocks (Fig. S3) in the Tajik Basin ranged from −7.51‰ to −13.48‰, with an average value of −10.46‰ (Fig. 2; Table S1). The δ¹⁸O_cc (cc—carbonate cements) values were highest (∼−8.67‰) in the Paleocene to Middle Eocene continental deposits. The δ¹⁸O_cc values were observed to decrease sharply (∼−10.84‰) in Upper Eocene strata and increase gradually to intermediate and stable values in Lower Oligocene strata (∼−9.45‰), followed by a decreasing trend up section (Fig. 2). The δ¹⁸O values measured across growth lines of an oyster fossil yielded high variability along with growth increments (Fig. S4). The δ¹⁸O and δ¹³C values for shell, limestone, and adjacent continental red-bed samples were notably different (Fig. S4). In the western Tarim Basin, δ¹⁸O_cc values from Upper Eocene nonmarine strata yielded a similar upward-decreasing trend as those from the Tajik Basin (Fig. 2).

The chemical proxy of alteration (CPA) in the Tajik Basin yielded the highest values (∼90.01) in Upper Eocene strata, decreasing to low values (∼80.35) in Lower Oligocene strata and returning to relatively high values (∼84.44) in Upper Oligocene to Lower Miocene strata (Fig. 2; Table S2). X-ray diffraction (XRD) analyses indicated that clay minerals mainly comprise illite, smectite, and kaolinite (Figs. S5A and S5B). Scanning electron microscopy (SEM) data show that clay minerals in Upper Oligocene to Lower Miocene strata comprise a large portion of authigenic clays (Figs. S5C–S5F), such as honeycomb-shaped smectite, whereas those from Upper Eocene and Lower Oligocene strata are dominated by detrital clays (Figs. S5G–S5H).

Our climate simulations, using pre-industrial boundary conditions with a large areal extent of the Paratethys (Markwick, 2007; see Fig. S6 and Table S3), show an arid and semiarid climate over Central Asia (Fig. 3A). Modeled east-west climatological differences are present when relief of the Pamir–Tian Shan convergence zone is higher than 75% of modern elevation (∼3000 m; Fig. 3B), whereas these differences disappear below 50% of modern elevations (∼2000 m; Fig. 3C). Modeled seasonal precipitation patterns show that the regions west of the Pamir–Tian Shan have winter-spring precipitation maxima, whereas regions to the east have summer precipitation maxima (Fig. 3D). The west-east differences in precipitation seasonality become much weaker with progressively reducing topography in the Pamir–Tian Shan orogen (Figs. 3E and 3F). This is consistent with other simulations using a regional climate model (RegCM4.1, https://www.ictp.it/research/esp/models/regcm4.aspx) with late Oligocene boundary conditions (Liu et al., 2015).

If continental sedimentary carbonates are formed early and under shallow burial conditions, the δ¹⁸O values of these carbonates will reflect the isotopic composition and local temperature of surface waters (Quade and Roe, 1999). The following lines of evidence suggest that the δ¹⁸O_cc values in this study are mainly related to syndepositional meteoric water or very early cementation. We observed distinctly different δ¹⁸O and δ¹³C values among shell, limestone, and adjacent continental red-bed samples (Fig. S4). This provides evidence that diagenesis has not reset the isotopic chemistry of the carbonates. Thin-section point counting of the sandstones suggests that a minor proportion (2.2%) of the total lithic grains in the samples are detrital carbonates (Table S4), and therefore they are unlikely to contaminate the stable isotopic data. High intergranular volume proportions (19%–40%; Table S4) suggest early sandstone cementation at shallow burial depth to prevent significant sandstone compaction (Zhuang et al., 2011). The δ¹⁸O_cc values of Upper Eocene nonmarine strata are not identical (Fig. 2), suggesting that dissolved calcite from marine limestones, if any, had minimal impact on the δ¹⁸O_cc compositions (Bershaw et al., 2012). Finally, the similarity between δ¹⁸O values of carbonate cements and paleosol samples from an adjacent section (∼60 km apart) in the Tajik Basin (Fig. 2) provides evidence that the cements were formed soon after deposition, and in environments very similar to the co-occurring soil carbonates.

After the ultimate retreat of the Paratethys at ca. 37.4 Ma (Wang et al., 2019), an arid to semiarid climate regime was established in the Tajik Basin, as supported by the high δ¹³C values of paleosols from the Dashtijum (DS) and Zidi (ZD) sections (Fig. S7) and the occurrence of eolian deposits in the Oligocene strata of the Peshtova (PE) section (Carrapa et al., 2015). The δ¹⁸O_cc data from the Tajik Basin, the western Tarim Basin, and other sections from northwestern China (Kent-Corson et al., 2009) yielded comparable Paleogene decreasing trends (Fig. 4; Fig. S8), suggesting that there were no fundamental differences in the source of moisture between the windward and leeward sides of the Pamir–Tian Shan orogen, nor were there any high topographic barriers between the two regions during most of the Paleogene (Fig. S9). Paleogene δ¹⁸O_cc decreasing trends likely were linked to the westward retreat of the Paratethys arising from the “continental effect” and/or due to late Eocene global cooling (Zachos et al., 2001), leading to lower δ¹⁸O values of the regional rainfall (Hoefs, 2015).

The most significant changes in lithofacies and paleoclimate in the Tajik Basin occurred during the late Oligocene (ca. 25 Ma). Lithofacies in the Tajik Basin changed gradually from eolian sandy loess to fluvial-dominated facies (Fig. 2), consistent with a relatively wetter climate in the region after ca. 25 Ma (Wang et al., 2019). CPA values increase systematically starting in the late Oligocene (Fig. 2), suggesting enhanced chemical weathering and a relatively wetter climate (Ren et al., 2019) in the Tajik Basin and its catchments after ca. 25 Ma. SEM observations indicate that clay minerals in Lower Oligocene strata are dominated by detrital clays, whereas those in Upper Oligocene to Lower Miocene strata comprise a large portion of authigenic clays (Fig. S5), consistent with enhanced chemical weathering and poorly drained conditions (Wilson, 1999) in the Tajik Basin after ca. 25 Ma. These independent lines of evidence suggest a change in climate toward wetter conditions in the Tajik Basin after ca. 25 Ma, which is comparable to late Oligocene to early Miocene wetter climate changes in the Issyk Kul (Macaulay et al., 2016) and Ili basins (Hellwig et al., 2018) on the Tian Shan western flanks (Fig. 1A), but it is in contrast with the late Oligocene to early Miocene enhanced aridification in northwestern China and Mongolia on the eastern side of the orogen (Guo et al., 2002; Zheng et al., 2015; Richoz et al., 2017).

Stable oxygen isotopic records from the windward and leeward sides of an orogen can be used to evaluate precipitation changes across topography and forcing mechanisms. The δ¹⁸O_cc values decreased systematically in the Tajik Basin between ca. 25 and 23.3 Ma (Fig. 2). Comparison with the records from the Issyk Kul (Macaulay et al., 2016), Zaysan (Caves et al., 2017), and Junggar (Charreau et al., 2012) basins reveals that all of these basins that are windward or near-windward show decreasing δ¹⁸O_cc starting in the late Oligocene (Fig. 4A). These decreasing δ¹⁸O_cc trends contrast with nearly all δ¹⁸O_cc records from northwestern China and Mongolia (Kent-Corson et al., 2009; Zhuang et al., 2011; Caves et al., 2014), on the leeward side of the orogen, which yield overall increasing trends since the late Oligocene (Fig. 4B).

Global temperature and ice-volume change no doubt played a role in driving regional climate changes, but these are unlikely to have been the main contributor for the establishment of the west-east hydroclimate differences over Central Asia (Fig. 4), as these factors would have affected the two regions similarly. Instead, we posit that significant topography associated with the Pamir–Tian Shan orogen best explains the opposite trends in δ¹⁸O_cc records between the windward and leeward sides of the orogen since the late Oligocene. First, a wetter climate, associated with increased orographic precipitation on the windward side of the mountains (Figs. 3A–3C), can explain a decrease in δ¹⁸O_cc, consistent with increased humidity and lower postcondensation evaporation (Hoefs, 2015). Second, an increase in the elevation of the study region’s catchments could shift the basinal δ¹⁸O_cc to lower values, as authigenic carbonates recorded high-elevation waters from the growing Pamirs to the south (Wang et al., 2019). Third, a change in precipitation seasonality toward dominant winter precipitation may have influenced the timing of pedogenic carbonate formation, thereby decreasing δ¹⁸O_cc values due to lower δ¹⁸O values of precipitation in the winter (Caves et al., 2017). The overall increasing δ¹⁸O_cc values during the Neogene in records from northwestern China and Mongolia on the leeward (east) side of the Pamir and Tian Shan are broadly interpreted as a result of enhanced aridification (Zhuang et al., 2011). All these mechanisms require the establishment of a modern-like tectonic-climatic configuration during the late Oligocene (Fig. 1B).

When integrated with geological evidence, our new data suggest that a significant part of the Pamir–Tian Shan mountains had reached elevations of ∼3 km and acted as a moisture barrier for the westerlies since ca. 25 Ma. Interestingly, in our simulations, monsoon precipitation in the Himalaya also decreases when the Pamir–Tian Shan convergence zone is lowered (Figs. 3B and 3C), suggesting a possible control by Pamir–Tian Shan topography on westerlies and the Asian monsoon.

We thank Huayu Lu, John Bershaw, and anonymous reviewers for their constructive comments. This work is supported by the National Natural Science Foundation of China (grant 41672158), the second Tibetan Plateau Scientific Expedition and Research Program (grant 2019QZKK0602), the U.S. National Science Foundation (grant EAR-1450917), and the Helmholtz Postdoc Program (grant PD-301).