Introduction

The Middle Miocene Climatic Optimum (~17–14 million years ago) encompasses an interval of relatively warmer temperatures and enhanced precipitation during the mid-Miocene1,2,3,4. This period was immediately followed by the Middle Miocene Climate Transition, which was characterized by a global sea level drop of ~40–90 m5,6,7, a cooling of ~6–7 °C and ~2 °C for sea surface temperatures near Antarctica and in the Equatorial Pacific, respectively8,9,10, and a pCO2 decrease of ~100 ppm11,12. This substantial global cooling and the permanent establishment of the East Antarctic ice sheet likely had an influence on global precipitation, sea and land surface temperatures, oceanic circulation, and the biosphere13,14,15,16. Despite this cooling and its potential ramifications, the mechanisms behind it are still debated, with orbital forcing, reorganization of past ocean circulation, and/or atmospheric CO2 drawdown all having been proposed to trigger the middle Miocene Climate transition8,9,11,17,18.

The hydrological cycle of East Asia was altered across the middle Miocene Climate transition via impacts from shifts in the location of the tropical rain belt13, intensification of Atlantic and/or global Meridional Overturning Circulation14,19,20, varied meridional heat/water vapor transport21,22 and decreased global sea-level23. Specifically, pollen records24,25, magnetic proxies26,27, geochemical element concentrations28, and carbonate oxygen isotopes29 all suggest that the relatively humid environment of inland Asia around 17–15 million years ago became arid after the middle Miocene Climate transition. However, these results contrast with proxy records from South China Sea sediments30 and expansion of paleolakes in the Guide, Xunhua and Linxia basins31,32, which imply increased summer monsoonal precipitation in East Asia around 11–10 million years ago. Modeling results may support this notion, as simulations with an expanded Antarctic ice sheet but no Greenland ice sheet show enhanced summer monsoon precipitation in East Asia33. Considering the discrepancies between prior studies, the long-term evolution of East Asian hydroclimate variability is still uncertain.

Sediments from Northeast (NE) Tibet can provide sensitive archives of past changes in the regional hydroclimate of East Asia related to the middle Miocene Climate transition34,35. Specifically, they potentially allow for the discernment of orbital-scale shifts in the contribution of precipitation between moisture delivered via the westerlies from Central Asia and East Asian summer monsoon-related moisture coming from the Pacific Ocean and the South China Sea during and after this important interval of the geologic past32,36. However, previous studies in NE Tibet have primarily focused on the long-term, million-year climate evolution of this region, and lack sufficiently high-resolution paleohydrological records spanning the middle-to-late Miocene. Here we present orbital-scale geochemical records of element concentrations and clay mineral ratios for the interval of ~13.7–10 million years ago from the Guide Basin, located on the NE Tibetan margin, and interpret our data with the aid of recent climate model simulations for the Middle Miocene37. Our results reveal an antiphase relationship between westerly moisture contribution and East Asian summer monsoon-derived moisture on 100-kyr timescales.

Results and discussion

Orbital cycles in Middle Miocene hydroclimate records on the NE Tibet

The Guide Basin is surrounded by the Waligong Shan to the west, the Zamazari Shan to the east, the Laji Shan to the north, and the West Qinling Mountains to the south (Fig. 1). The Neogene deposits of the Guide Basin have been divided into the Guidemen (~20.8–19 million years ago), Garang (~19– < 16 million years ago), Ashigong (~> 11.5–7.8 million years ago), Heerjia (~7.8–3.6 million years ago), and Ganjia (~3.6–2.6 million years ago) formations based on a combination of magneto- and mammalian stratigraphy32,38. The middle Miocene Ashigong Formation is composed of massive red-brown, blocky mudstones interbedded with grayish-green, finely-laminated mudstones, sandstones, and siltstones in the lower member, changing to deposits dominated by marls and gypsum-rich grayish-green mudstones in the middle member, and finally prominent, dense red-brown and grayish-green mudstones in the upper member32. The lower and upper members indicate a marginal lake depositional environmental, while the middle member reflects a more permanent lake setting32. A recent age model constructed for the Ashigong Section in the Guide Basin spanning ~13.7–10 million years ago suggests a dominant ~100 kyr cyclicity in the Guide paleolake evolution32. The age uncertainties of the Ashigong Section are based on the paleomagnetic boundary uncertainties, with about ± 40 kyr error in pairs of normal+reversed polarity32. Recently, Yao et al.39 used the lightness (L*) of the Ashigong Formation sediments for astronomical tuning using the magnetostratigraphy as initial constraints. Specifically, these authors correlated the ~50 m wavelengths in the L* series to the orbital eccentricity metronome. Their tuning results indicate the presence of 405 kyr spectral peaks with high significance as well as ~100 kyr, 40 kyr and 20 kyr spectral peaks in the L* series. In particular, the ~100-kyr cycles in the L* are antiphase with the La2004 short orbital eccentricity model39. These findings suggest that our interpretations regarding moisture-derived variability, at least on the 100-kyr orbital scale, are robust.

Fig. 1: Map showing the locations of the Guide Basin (this study), Aktau Section55 and marine sediment core ODP 114830.
Fig. 1: Map showing the locations of the Guide Basin (this study), Aktau Section55 and marine sediment core ODP 114830.
Full size image

A The Aktau Section is in the arid zone predominantly influenced by moisture derived from the westerlies. The Ashigong Section in the Guide Basin receives precipitation from both the westerlies and the East Asian summer monsoon. Finally, ODP 1148 is proximal to East Asian summer monsoon moisture sources. The base map was taken from https://www.ngdc.noaa.gov/mgg/global/ that is a globally shareable website. B, C are field photographs from the Guide Section. EAWM-East Asian winter monsoon; EASM-East Asian summer monsoon; ISM-Indian summer monsoon.

We measured several proxies to reconstruct hydroclimate variability in the Ashigong Formation in the Guide Basin, including manganese-to-iron ratios (Mn/Fe, ~10 cm sampling resolution) and clay mineral content (~1 m sampling resolution) (Figs. 2 and 3). In sediment porewaters, Mn reduction occurs prior to Fe reduction as oxygen is depleted, and as such lower Mn/Fe rations typically indicate reducing conditions while elevated Mn/Fe ratios point to more oxygenated porewaters40,41,42. In the Ashigong Section, periods of low magnetic susceptibility (MS) and redness and high Mn/Fe correspond to intervals of gray-green mudstone. Individually, Fe content is generally low and Mn is relatively high (Fig. 2 and Supplementary Fig. S1) within these gray-green mudstone layers. Broadly, the trends in Mn/Fe ratios, redness, and MS are consistent. Therefore, we suggest that the Mn/Fe ratios can reflect the sedimentary environment during the mid-to-late Miocene, and specifically indicate regional lake level fluctuations (Fig. 2 and Supplementary Fig. S1). While the correspondence between Mn/Fe ratios and lake level may seem counterintuitive, we suggest that increased regional precipitation, and in turn elevated lake levels, resulted in i) enhanced post-depositional dissolution of Fe-bearing magnetic minerals, ii) precipitation of Mn carbonates, or iii) a combination of these two processes, ultimately increasing Mn/Fe ratios on these layers. Post-depositional dissolution of Fe-bearing magnetic minerals is common in lacustrine deposits of northern Chinese basins22,43 (Supplementary text 1). In summary, based on other data from the section and our posited diagenetic impacts, we assume that higher Mn/Fe ratios in our record generally correlate to higher lake levels and are linked to periods with higher regional precipitation.

Fig. 2: Lithology, magnetic stratigraphy and paleoclimate proxies of the Ashigong Formation in the Guide Basin, NE Tibet.
Fig. 2: Lithology, magnetic stratigraphy and paleoclimate proxies of the Ashigong Formation in the Guide Basin, NE Tibet.
Full size image

(A) redness (a*). (B) magnetic susceptibility (MS). (C) Mn/Fe ratios. (D) illite content, and (E) smectite/(illite+chlorite) ratios. Magnetic stratigraphy, redness and MS data are from Wang et al.32.

Fig. 3: Evolutionary spectra of global δ18Obenthic, δ13Cbenthic, smectite/(illite+chlorite) and Mn/Fe ratios over the period of ~13.7–10 Ma.
Fig. 3: Evolutionary spectra of global δ18Obenthic, δ13Cbenthic, smectite/(illite+chlorite) and Mn/Fe ratios over the period of ~13.7–10 Ma.
Full size image

A Global δ18Obenthic timeseries across the Miocene64. B Global δ18Obenthic timeseries for the interval of ~13.7–10 Ma64. C Global δ13Cbenthic time series for the interval of ~13.7–10 Ma72. D Smectite/(illite+chlorite) ratio from the Guide Section spanning ~13.7–10 Ma (this study). E Mn/Fe ratios from the Guide Section spanning ~13.7–10 Ma (this study). The yellow inverted triangles are the paleomagnetic age tie-points. MCT Miocene Climate transition, MCO Miocene Climatic Optimum.

In certain environments (i.e., regions and time periods where provenance changes are minimized), clay mineralogy can be used to reconstruct the silicate weathering intensity of the larger catchment44,45. Chlorite and illite primarily reflect cold and/or arid environments, whereas smectite indicates temperate-humid climates45. Thus, higher smectite/(illite+chlorite) ratios generally represent periods with stronger chemical weathering resulting from either higher temperature, precipitation, or a combination thereof. According to the classical diagenesis model by Weaver46, when digenesis plays a major role in regulating illite and smectite contents, the illite content is expected to exhibit gradual increases with the rise of the burial depth. Our results show that illite is relatively stable and does not show an obvious change with depth (Fig. 2 and Supplementary Fig. S1), suggesting minimal diagenetic impacts during and after deposition of the studied Ashigong Section.

Evolutive spectral analysis of our Mn/Fe proxy displays dominant ~100-kyr cycles between ~13.7 and 10 million years ago (Fig. 3). We interpret these ~100-kyr cycles to represent eccentricity-forced paleolake level/precipitation variability in the Guide Basin during the period of ~13.7-10 million years ago32. Additionally, our smectite/(illite+chlorite) record contains prominent ~405-kyr and ~100-kyr cycles throughout the period of ~13.7–10 million years ago (Supplementary Fig. S2). Both the Mn/Fe and smectite/(illite+chlorite) records display weak obliquity (41-kyr) and precession (23- and 19-kyr) signals (Fig. 3 and Supplementary Fig. S2). It could be argued that the lack of 41-kyr, 23-kyr, and/or 19-kyr cycles in the two records, particularly the lower resolution clay mineral ratios, may be an artifact of sampling resolution. Sedimentation rates based on magnetic reversal stratigraphy are approximately 6.5–14.4 cm/kyr32. This leads to sampling intervals spanning ~0.7–1.5 kyr for Mn/Fe and ~7–15.4 kyr for the clay mineral ratios. However, even in the interval containing the highest sedimentary rates (~14.4 cm/kyr; ~11.6–12.1 million years ago), the 100-kyr cycles are still dominant compared to either obliquity or precession signals (Fig. 3 and Supplementary Fig. S2). As such, we suggest that the findings of our spectral analysis for both proxy records are robust. Finally, we note that the Mn/Fe ratio shows a stronger obliquity signal between 12.4 and 13 million years ago compared to the clay mineral ratios (Fig. 3). This difference in orbital expression between the Mn/Fe and smectite/(illite+chlorite) data may be attributed to the fact that variability in the smectite/(illite+chlorite) ratio is not only affected by precipitation, but can also be linked to changes in temperature as well as biological activity44,45.

Sources of water vapor for NE Tibet across ~ 14–10 million years ago

A comprehensive review of pedogenic and lacustrine carbonate stable oxygen isotope data in Asia shows that south Asian monsoonal-derived moisture has not been transported toward the Central Tibetan Plateau since the Cenozoic47,48. Thus, the hydrological budget of the NE Tibet is a balance of East Asian summer monsoonal moisture originating from the Pacific Ocean and the South China Sea, and winter, westerly-derived moisture from the North Atlantic and the Mediterranean48,49. Analyses of pollen records from the late-Oligocene Lanzhou Basin50, temporal-spatial distributions of carbonate δ18O and δ13C across Asia47, and late Oligocene-early Miocene eolian deposits in the Tianshui Basin51 suggest that NE Tibet predominantly experienced a monsoonal climate like that of the present day since the late Oligocene, with a smaller contribution of moisture delivered via the westerlies. Although some thermochronological data and sedimentary records reveal that the Pamir Mountains experienced substantial uplift prior to the early Miocene and thus acted as an important barrier to precipitation in inland Asia36,52, an open channel between the Pamir Mountains and the Tianshan Mountains may have existed during the middle Miocene53. As such, moisture brought from the westerlies could have penetrated the Tarim Basin and/or the Qaidam Basin at this time53,54, indicating that previous studies may have underestimated the contribution of westerlies-derived moisture to NE Tibet.

To explore moisture sources and orbital-scale moisture variability of the NE Tibet during the interval of ~14–10 million years ago, we utilize hematite/goethite records from Ocean Drilling Program (ODP)1148 in the South China30 (Fig. 1), which lies near East Asian summer monsoon moisture sources, and CaSO4 contents from the Aktau Section in the Ili Basin, which receives moisture predominantly via the westerlies55 (Fig. 1). In a lake system, a low CaSO4 content reflects enhanced aquifer recharge and is associated with lake freshening accompanied by rising lake levels, whereas a high CaSO4 content is related to a negative water budget and higher solute concentration55. Hematite/goethite records from marine sediments may reflect continental chemical weathering processes because during soil formation, humid conditions are favorable for goethite formation while drier conditions are favorable for hematite formation30. For ODP 1148 in the South China Sea specifically, the hematite/goethite ratio is proposed to be a proxy for East Asian monsoonal precipitation variability based on two lines of evidence: 1) iron oxide formation and mineralogy in the source region are sensitive to East Asian summer monsoon precipitation56,57, and 2) the influences of post-depositional redox changes on hematite/goethite ratios are limited as high-coercivity hematite and goethite are rather resistant to porewater reduction processes56,57. In addition, the hematite/goethite ratios from the ODP 1143 core correspond well with magnetic susceptibility of the Chinese Loess Plateau58 (suggested to be a proxy of East Asian summer monsoon variability) during the late Pliocene57. Overall, more goethite deposition in the South China Sea corresponds to a stronger summer monsoon, whereas weaker summer monsoon periods would result in more hematite deposition30,56,57. Broadly, the hematite/goethite ratios from the South China Sea could reflect East Asian summer monsoon variability.

Hematite/goethite ratios in ODP 1148, CaSO4 content of sediment from the Ili Basin (except for near zero values encountered in the interval of ~11.5–12.5 million years ago) and our Mn/Fe records reveal distinct ~100-kyr cycles, and imply a short eccentricity control of regional hydroclimate in the Middle Miocene (Fig. 4). This prominent ~100-kyr eccentricity signal is also recognized in hydrological records from the fluviolacustrine deposits in the Qaidam Basin, NE Tibet21 (~8.5–7 million years ago), lake sediments from the Tianshui Basin, NE Tibet22 (~10.25–8 million years ago), and fluviolacustrine sequences from the Weihe Basin, North China59,60 ( ~ 5.2–2.6 million years ago). These suggest that the hydrological records across Asia are particularly sensitive to ~100-kyr eccentricity forcing under unipolar icehouse states.

Fig. 4: Spectral and filter analysis from the Guide Basin in the NE Tibet (this study), and comparison with other climatic records over the interval of 14–10 Ma.
Fig. 4: Spectral and filter analysis from the Guide Basin in the NE Tibet (this study), and comparison with other climatic records over the interval of 14–10 Ma.
Full size image

A Calculated eccentricity curve with a 405-kyr filter85 (red curve, 0.0025 ± 0.0005 cycles/kyr). B Global δ18Obenthic data with a 100-kyr filter64 (red curve, 0.01 ± 0.003 cycles/kyr). C Mn/Fe ratios in the Guide Basin with a 100-kyr filter (red curve, 0.01 ± 0.003 cycles/kyr). D Hematite/goethite ratios from the ODP1148 in the South China Sea with a 100-kyr filter30 (red curve, 0.01 ± 0.003 cycles/kyr). E CaSO4 content from the Aktau section in the Ili Basin and the 100-kyr filter55 (red curve, 0.01 ± 0.003 cycles/kyr). F Smectite/(illite+chlorite) ratios in the Guide section with 405-kyr (pink curve, 0.0025 ± 0.0005 cycles/kyr) and 100-kyr (red curve, 0.01 ± 0.003 cycles/kyr) filters. G Global δ13Cbenthic data with a 405-kyr filter72 (pink curve, 0.0025 ± 0.0005 cycles/kyr). The grey bars indicate relatively low values of the theoretical 405-kyr eccentricity.

The origin of an eccentricity control on Asian hydrological cycles prior to ~2.6 million years ago is suggested to be related to a non-linear amplification mechanism associated with Antarctic ice-sheet dynamics and/or eccentricity modulation of low-latitude summer insolation through a non-linear interaction with the global carbon cycle21,22,60,61,62. Inferred Middle Miocene high-water levels in the Guide and Ili basins correspond to low 405-kyr eccentricity intervals, which leads to decreased net incoming solar radiation to Earth’s surface through its effect on precession amplitude (Fig. 4). At present, it is generally considered that the eccentricity maximum causes high-amplitude variability of precession cycle through amplitude modulation, enhancing the amount of solar radiation at low latitudes and thus strengthening the intensity of the EASM and increasing summer precipitation in northern China59,62,63. However, our results show that some periods of high paleolake water levels in NE Tibet correspond to low 405-kyr eccentricity periods (Fig. 4), which may indicate that precession-induced low-latitude insolation variability plays a secondary role in the paleolake evolutions in the Guide Basin. Jin et al.62 suggested that there was a short eccentricity-control on hydroclimate records in the Nima Basin of the Tibetan Plateau interior over ~27.32–23.24 million years ago, with precipitation being modulated by low-latitude summer insolation variability rather than glacial-interglacial Antarctic ice sheet fluctuations. However, tuning paleoclimatic proxy records to the 405-kyr orbital curve may cause large phase deviations between the benthic δ18O (‰) record and the paleoclimatic proxy in question on the ~100-kyr eccentricity scale, an important consideration for records from the Nima Basin. In fact, for Northern Hemisphere ice-free periods from the latest Eocene to the late Pliocene, global climate variability was substantially influenced by Southern Hemisphere summer insolation, amplified by a dynamic Antarctic ice sheet64.

After the middle Miocene Climate transition, increases in lake levels occurred in the NE Tibetan Guide and Ili basins (Figs. 2 and 4), with maximum lake levels found at about 11–10.5 million years ago32,55. Although the Mn/Fe and smectite/(illite+chlorite) show a relatively weak response to the overall rise of water level at 11–10.5 million years ago, the changes of lithology, MS, redness, and Sr content at ~11–10.5 Ma point to a substantial rise in water level at 11–10.5 million years ago32. Provenance65 and paleomagnetic38 studies suggest that the Guide Basin was not significantly impacted by renewed uplift until ~8 million years ago, after the deposition of the lacustrine Ashigong Formation. It is particularly noteworthy that the 11.4–10.5 million years ago time window corresponds very well to the peak of Asian monsoonal intensity during the Neogene66. Stable isotopes from the South China Sea indicate a remarkable warming episode at ~10.8–10.7 million years ago67, which fits with maximum lake expansion in our section. We thus argue that climate changes, rather than tectonic activity, are the most likely driver of water level fluctuations during this time. In addition, this period (11−10 million years ago) of lake expansion not only occurred in the Guide Basin, but also in the adjacent Xunhua and Linxia basins31. Therefore, this is not a local event, but a common phenomenon in several NE Tibetan basins.

The intensity of chemical weathering in the South China also strengthened after the middle Miocene Climate transition, with an extended period of elevated monsoon intensity around 18-10 million years ago, followed by a weakening of the East Asian summer monsoon30 (Fig. 4). These results may indicate that spatial differences of precipitation variability occurred across Asia after the middle Miocene Climate transition25,27. The decrease in atmospheric CO2 concentrations after the middle Miocene Climate transition, which was accompanied by a decline in global temperature, will generally reduce the amount of water vapor in the atmosphere. This could have led to a long-term intensification of Central Asian aridity and/or precipitation decrease in East Asia, as is suggested by pollen records that are sensitive to pCO2 variability. However, atmospheric CO2 is not the only potential control on the Miocene climate of East Asia. Specifically, the expansion of the East Antarctic ice sheet could amplify the interhemispheric temperature gradient, and thus shift tropical rain belts northward into the warmer hemisphere13,68. This could produce an overall increase in Northern Hemisphere precipitation69. In addition, after the middle Miocene Climate transition, a more vigorous meridional overturning circulation in the Pacific Ocean and intensification of the Atlantic Meridional Overturning Circulation (AMOC) following global cooling occurred due to increased production of circumpolar deep water as well as waters equivalent to modern Antarctic Intermediate Water14. Intensified AMOC, and generally enhanced overturning globally, would trigger a growth in northward oceanic heat transport, resulting in Northern Hemispheric warming and elevated rainfall in NE Tibetan basins22,69.

Climate model results for the Middle Miocene37 show that the scale of Central Asian aridity induced by the combination of decreasing atmospheric CO2 and expansion of the Antarctic ice sheet is sensitive to the mean state CO2 concentration itself (Supplementary Fig. S3). The largest reductions in annual precipitation occur at the lower CO2 concentrations simulated (Supplementary Fig. S3A). However, at higher simulated CO2 concentrations, Central Asia becomes wetter as CO2 decreases (Supplementary. Fig S3C). These results suggest that when Central Asian aridification did take place, atmospheric CO2 must have been comparatively low. The most recent estimates for the CO2 decrease over the middle Miocene Climate transition is from at least 500 ppm to ~400 ppm12. We therefore suggest that the most appropriate model simulation comparison for the long-term Middle Miocene hydroclimate evolution in Central Asia over our period of interest is Supplementary Fig. S3B (relatively higher atmospheric CO2), which suggests an overall increase in annual precipitation in this region of ~5-10%. These previous studies, along with our new records, suggest there might be no long-term drying trend across ~14-10 million years ago in NE Tibetan basins. Considering the data and modeling results, we hypothesize that aridification instead occurred after this interval, and was driven by a combination of further reductions in atmospheric CO2, and the eventual glaciation of Greenland and other regions of the high-latitude Northern Hemisphere. The loss of the Paratethys likely removed an important moisture source70, while the development of continental ice sheets in the Northern Hemisphere may have altered the position of the westerly jet71.

Eccentricity-driven variations in moisture sources for NE Tibet may be the cause of the ~100-kyr cycles after the middle Miocene Climate transition. Specifically, higher lake levels occurred in the Guide and Ili Basins55 during maximum values of eccentricity, generally consistent with lower global δ18O values, indicating warmer climate intervals (Fig. 4). However, during these warm periods, East Asian continental weathering intensity (and the strength of the East Asian monsoon) was predominantly weaker (shaded intervals 1, 4, 5 and 6 in Fig. 4), although there are some periods of enhanced weathering/summer monsoon strength (shaded intervals 2 and 3 in Fig. 4). Climate model simulations with orbital configurations favorable for Antarctic ice sheet expansion and contraction37 show that in the warmer climate/ice sheet retreat cases, East Asian monsoon weakening occurs relative to the colder climate/ice sheet expansion cases, regardless of the atmospheric CO2 concentration (Supplementary Fig. S4A and S4B compared to S4C and S4D). As such, at least on ~100-kyr timescales after the middle Miocene Climate transition, the variations of lake levels in NE Tibet are not synchronized with weathering intensity fluctuations in southern China. This may indicate that moisture derived from the westerlies provided an important contribution to NE Tibetan precipitation during eccentricity maximum over ~14–10 million years ago, an interpretation supported by the correspondence between the Guide and Ili Basins and the climate model simulations for the Middle Miocene37 that indicate that the position of the Northern Hemisphere westerly jet shifts to the north when the Antarctica ice sheet expands (Fig. 5). We recognize that additional summer monsoonal moisture from the South China Sea and Pacific Ocean to NE Tibet also contributed to overall precipitation on 100-kyr timescales, as evidenced by some correspondence between the hematite/goethite ratios from ODP 1148 and higher smectite/(illite+chlorite) ratios from the Guide Section (shaded interval 3 in Fig. 4; Fig. S5). Despite previous studies suggesting that long-term (~405 kyr) shifts in the global carbon cycle (Fig. 4) and associated global temperature variations triggered long-term changes in the chemical weathering intensity across the continents72, such similar fluctuations between Mn/Fe ratios and the smectite/(illite+chlorite) in our records suggest that precipitation substantially contributes to variability in chemical weathering, at least on ~100-kyr timescales in East Asia.

Fig. 5: Simulated westerly jet position and intensity under Middle Miocene boundary conditions with various orbital configurations favorable for Antarctic ice sheet expansion and retreat at different atmospheric CO2 concentrations.
Fig. 5: Simulated westerly jet position and intensity under Middle Miocene boundary conditions with various orbital configurations favorable for Antarctic ice sheet expansion and retreat at different atmospheric CO2 concentrations.
Full size image

A 280 ppm and an orbital configuration that favors a warmer climate globally without an Antarctic ice sheet. B 850 ppm and an orbital configuration that favors a warmer climate globally without an Antarctic ice sheet. C 280 ppm and an orbital configuration that favors a cooler climate globally with an Antarctic ice sheet. D 850 ppm and an orbital configuration that favors a cooler climate globally with an Antarctic ice sheet. The purple regions in these latitudinal cross-sections through the atmosphere at the ~200 mb pressure levels show the position of the westerlies. The latitudinal positions of the Aktau Section and the Guide Section are shown as white circles.

Paleoclimatic records and climate model simulations from other studies reveal that a reduced northward migration of the westerlies during the Quaternary alters the timing and duration of the various rainfall intraseasonal stages (namely spring, pre-meiyu, meiyu, and midsummer), resulting in spatial variability in East Asian rainfall73,74. Due to the accelerated uplift of the Pamir and Tianshan Mountains in the late Miocene53,75, the circulation patterns of the westerlies in the Quaternary are likely different from that of the Miocene period. This is because most of water vapor carried by the westerlies is rained out on the windward slope of the Tianshan Mountains and Pamir Plateau today, and thus less moisture can reach NE Tibet since the uplift of the Pamir Mountains76. During the Middle Miocene, however, a narrow water-vapor channel may have existed between the Tianshan and Pamir Plateau53, resulting in more moisture transported to NE Tibet compared to the Quaternary22. Our results demonstrate alternating moisture-derived variability between westerly winds and East Asian summer monsoon on the northeastern Tibetan Plateau, potentially related to the waxing and waning of the Antarctic ice sheets on orbital timescales.

Materials and Methods

Data measurement and spectrum analysis

A total of 3700 samples (~10 cm sampling resolution) were collected after removal of the weathered outcrop surface (this removed surface corresponds to a minimum of the uppermost 0.5 m). Ages of the samples are based on the age model of Wang et al.32. Approximately 5 g of each sample were ground into powder and passed through a 200-mesh sieve. To determine the abundance of Fe and Mn, samples were measured by a portable energy-dispersive Innov-X Systems X-ray fluorescence spectrometer in geochemistry mode using beam 1 (50 kv) and beam (10 kv) at the Chinese University of Geoscience (Wuhan). The measurement time for each sample was at least 50 s, taking the average value of three measurements. About 0.5 g of each powdered sample was further sieved using a 150-mesh screen, and mineral contents were subsequently determined using a Olympus Terra BTX II X-Ray Diffraction Analyzer at the Chinese University of Geoscience (Wuhan). The measurement time for each sample was at least 25 min. The compiled δ18Obenthic and δ13Cbenthic data are derived from nine δ18Obenthic and δ13Cbenthic records64,72, with an average temporal resolution of 2.89 kyr and 2.80 kyr, respectively.

We calculated the phase relationships between Mn/Fe-δ18Obenthic, Mn/Fe-δ13Cbenthic, and hematite/goethite-smectite/(illite+chlorite) using a cross-spectrum fast Fourier transform, implemented in the IRISSeismic package in R77. We calculated the cross spectra using a 200-kyr sliding window and a 10-kyr step size in individual datasets. Phase analysis calculated at the ∼100 kyr eccentricity scale was chosen for a frequency range of 90–135 kyr with the highest frequency coherence. The R code is taken from De Vleeschouwer et al.72. Evolutionary spectral analysis was carried out using evolutionary Fast Fourier transform spectrograms with evofft.m software78, with a window of 800 kyr and simulation step size of 6 kyr. The ∼100 kyr eccentricity components were extracted using Gaussian band-pass filtering in AnalySeries 2.0.879, with a center frequency and bandwidth of 0.01 and 0.003 cycles/kyr, respectively.

Model simulations

The climate model simulations use the fully coupled atmosphere–ocean GCM HadCM3LB-M2.1aE80 with the interactive vegetation model TRIFFID81. This is the low-resolution ocean version of HadCM382, and both the atmosphere and the ocean components have a resolution of 2.5° latitude by 3.75° longitude. The paleogeography used83 is representative of Middle Miocene conditions with an open Panama Gateway and Indonesian Seaway and a closed Bering Strait. The Tibetan Plateau, the Andes and Greenland are more than 2000 m lower than modern, and Greenland is also ice-free. Simulations were conducted with Antarctica being either ice-free or with a modern-like ice sheet configuration of 55 m sea level equivalent. Where orbital parameters have been altered, they represent a cold orbital configuration favorable for Antarctic glaciation (low obliquity (22.1°), high eccentricity (0.054), perihelion during boreal summer, 374 W m–2 summer insolation at 70° S) and a warm orbital configuration favorable for Antarctic deglaciation (high obliquity (24.5°), high eccentricity (0.054), perihelion during austral summer, 483 W m−2 summer insolation at 70° S). See Bradshaw et al.37 for full details of the model setup.