Tectonic and orbital forcing of the South Asian monsoon in central Tibet during the late Oligocene

Significance

The modern Asian monsoon system plays a critical role in transporting rainfall from the tropics to the extratropics and supports the livelihood of billions of people. Understanding the timing of the Asian monsoon is an important scientific issue that could constrain its driving mechanism. We present a high-resolution paleoclimatic record (27.32 to 23.24 million years ago, Ma) during the late Oligocene epoch to show that the South Asian monsoon already reached central Tibet at 27.3 Ma. Subsequently, the South Asian monsoon intensified at ~25.8 Ma due to the uplift of the Tibetan Plateau. On the orbital timescale, monsoon precipitation is strongest when eccentricity is maximum. 100-kyr cycles of precipitation were forced by Northern hemisphere solar radiation, rather than the Antarctic ice sheet.

Abstract

The modern pattern of the Asian monsoon is thought to have formed around the Oligocene/Miocene transition and is generally attributed to Himalaya–Tibetan Plateau (H–TP) uplift. However, the timing of the ancient Asian monsoon over the TP and its response to astronomical forcing and TP uplift remains poorly known because of the paucity of well-dated high-resolution geological records from the TP interior. Here, we present a precession-scale cyclostratigraphic sedimentary section of 27.32 to 23.24 million years ago (Ma) during the late Oligocene epoch from the Nima Basin to show that the South Asian monsoon (SAM) had already advanced to the central TP (32°N) at least by 27.3 Ma, which is indicated by cyclic arid–humid fluctuations based on environmental magnetism proxies. A shift of lithology and astronomically orbital periods and amplified amplitude of proxy measurements as well as a hydroclimate transition around 25.8 Ma suggest that the SAM intensified at ~25.8 Ma and that the TP reached a paleoelevation threshold for enhancing the coupling between the uplifted plateau and the SAM. Orbital short eccentricity-paced precipitation variability is argued to be mainly driven by orbital eccentricity-modulated low-latitude summer insolation rather than glacial-interglacial Antarctic ice sheet fluctuations. The monsoon data from the TP interior provide key evidence to link the greatly enhanced tropical SAM at 25.8 Ma with TP uplift rather than global climate change and suggest that SAM’s northward expansion to the boreal subtropics was dominated by a combination of tectonic and astronomical forcing at multiple timescales in the late Oligocene epoch.

During the Cenozoic era, the collision between the Indian and Eurasian plates resulted in the Himalaya–Tibetan Plateau (H–TP) uplift and the Asian monsoon formation (1). The modern Asian monsoon system including the South Asian monsoon (SAM) and East Asian monsoon (EAM) plays a critical role in transporting rainfall from the tropics to the extratropics and supports the livelihood of billions of people. Different from other low-latitude global tropical monsoon systems driven by the seasonal migration of the Intertropical Convergence Zone (ITCZ), the Asian monsoon system spans the tropical oceans to subtropical and temperate East Asia (2) and has been conventionally assumed to have formed approximately 23 million years ago (Ma) during the Oligocene/Miocene transition (3, 4) and more recently has been argued to exist in the Eocene, such as around 40 Ma (5, 6). When the monsoon migrates northward beyond the tropics to the mid-latitude areas is an ongoing scientific debate.

Reconstructions of the Asian monsoon have helped to clarify its dynamics and interactions in response to many geological factors. Although the onset of the modern Asian monsoon has been generally attributed to high topographic relief of H–TP uplift (1, 7–14), global changes in solar radiation, atmospheric CO₂, paleotemperature, polar ice sheet, and the retreat of the Paratethys sea have also been proposed to be important factors which could influence or drive the Asian monsoon evolution (5, 15–20). Multiple drivers of Asian monsoon variability have been comprehensively reviewed, e.g., refs. 4, 13, 21, and 22. However, these drivers are sometimes contradictory. For the first example, the Himalaya insulation model (8) conflicts with TP thermal forcing for Asian monsoon intensification (12). The former emphasizes that the Himalayas and adjacent mountain ranges can block cold–dry air from the north and maintain warm-moist air from the south to produce a strong SAM through thermal insulation caused by orographically mechanical effects (8). However, subsequent simulation studies argued that the Asian summer monsoon systems are controlled mainly by the thermal forcing of the TP, whereas the mechanical effects of orography of the Himalayas are not essential to intensify the Asian monsoon (12, 23). For the second example, the dominant role of atmospheric CO₂ concentration in forcing the monsoon (5) was challenged by the orographic height forcing of the TP (10). High atmospheric CO₂ during enhanced greenhouse conditions in the Eocene was thought to intensify the hydrological cycle and produce the Asian monsoon (5). However, other simulation studies showed that even four times of CO₂ concentration during the Eocene could not promote the establishment of the extratropical Asian monsoon without the plateau topography (24). Therefore, the origin of the Asian monsoon should be attributed to the uplift of the TP, rather than high atmospheric CO₂ (10, 24). For the third example, some studies emphasize that orbital Asian monsoon variability was paced by polar ice sheets (15, 25), while, other studies emphasize that the Asian monsoon was driven by low-latitude solar radiation or by a combination of high-latitude ice sheets and low-latitude solar radiation (13, 26–29). To understand how the Asian monsoon--its timing, intensity, and migration process--responded to these factors, especially to the astronomical forcing and H–TP uplift, it requires the development of new paleoclimate records. This is another key issue that addresses the relationship between tectonism and climate.

To clarify these issues, paleoclimatic records from the extratropical TP interior are crucial for providing key constraints. At present, there are few such records with sufficient temporal resolution from the TP interior to decipher the early origin and migration of the ancient Asian monsoon across the TP and to elucidate the coupling between H–TP uplift and monsoon evolution (30). Previously reported paleoclimatic records of the ancient Asian monsoon are primarily located in the surrounding areas of the TP, e.g., refs. 5, 7, 11, and 28. Moreover, for pre-Miocene monsoon records, most studies have primarily focused on long-term trends of nonmonsoon aridification to long-term trends of monsoon-induced humidification. Astronomically forced orbital-scale monsoon variability and arid–humid hydrological cycles remain undetermined (28, 31), especially in the TP interior. This limits the identification of pre-Miocene Asian monsoon signals to constrain its processes and mechanisms.

Here, we present a time series analysis of precession-scale resolution environmental magnetic records of fluvial-lacustrine sediments from the Nima Basin in the central TP to constrain the midlatitude hydrological changes, the SAM northward progression, and potential forcing mechanisms on tectonic through orbital time scales during the late Oligocene. During the late Oligocene, there is only a unipolar ice sheet on Antarctica and the Earth experienced the Late Oligocene warming event (32–35). The TP experienced significant uplift during the late Oligocene (36–39). Therefore, the late Oligocene records at Nima present an opportunity to investigate SAM evolution in the light of the unipolar icehouse climate state and TP uplift.

Results

Chronology and Time Series Analysis.

The Nima Basin is in the Bangong–Nujiang suture zone of the central TP between the Lhasa and Qiangtang terranes with an elevation of about 4,500 to 5,000 m (Fig. 1). It trends generally E–W with a north–south width of about 50 km. It is bounded by the Gaize-Siling Co thrust to the south and the Muggar thrust to the north (40). The present climate is alpine, subarid with a mean annual temperature of about −4 °C and mean annual precipitation of about 150 mm under the combined influence of the Asian monsoon and the midlatitude Westerlies, and is therefore sensitive to hydroclimate changes (41).

Fig. 1.

Geologic setting and site location. (A) Topographic map of the present Tibetan Plateau and location of the Nima Basin. Yellow and blue arrows mean the prevailing summer and winter wind directions of the present-day Asian monsoons. SAM, South Asian monsoon. EAM, East Asian monsoon. (B) Paleogeographic reconstruction derived from ODSN (https://www.odsn.de/odsn/services/paleomap/adv_map.html) with a red dot showing the position of the sampled Dazecuo section at 26 Ma. The green dot shows the paleolatitude (33.7°/−6.8°/+5.2°N) of the Nima Basin for the 25.1 to 22.3 Ma time interval based on paleomagnetic data (42). The red thick line indicates the northern boundary of the simulated Intertropical Convergence Zone (ITCZ) at 25 Ma (24). (C) Geological map of the southern Nima Basin based on ref. 40. Q, Quaternary sediments; N-Q, Neogene-Quaternary; Nf, Neogene fan conglomerates; Tr, Tertiary red beds; Kr, Albian-Cenomanian red beds, volcanic-clast sandstone, and conglomerate; Kv, Albian volcanic flows, tuffs, breccias volcaniclastic sandstone, and conglomerate; J-K, Jurassic-Cretaceous shale, siltstone, turbiditic sandstone. The thick blue line on the right indicates the sampled Dazecuo section. (D) Alternation of 21 reddish-brown--grayish-green layers above 420 m of the section (Fig. 2).

The sampled Dazecuo section of the Dingqing Formation has a thickness of 1,018.27 m and is situated on the south bank of Dazecuo Lake in the southern Nima Basin, about 40 km east of the Nima town (31.82°N, 87.68°E, 4,605 m elevation) (Figs. 1 and 2 and SI Appendix, Figs. S1 and S2). It was deposited during the late Oligocene (41, 43). Please see SI Appendix, Method 1 for sampling.

Fig. 2.

Lithology, paleoclimate sequences, and age model of the Dazecuo section. (A) Lithostratigraphy. (B) Redness (a*). (C) SIRM, saturation isothermal remanent magnetization. (D) S-ratio. (E) HIRM, hard IRM. (F) Hm/Gt, hematite/goethite (in a log coordinate). Numbers G1–G21 indicate 21 grayish-green layers marked using light green bars. (G) PCA-F1, the first principal component of redness, SIRM, S-ratio, and HIRM. (H) Thickness filter of PCA-F1 for 405-kyr astronomical tuning. (I and J) Correlation between the measured polarity sequences with GTPS2020 (44). Red stars mark the ages and locations of D1 and D4 tuffite samples of ref. (41). (K and L) Eccentricity solution (45) (thin gray line) and its 405-kyr Gaussian filtered curve (passband: 0.00196 to 0.00298 cycles/kyr) (thick black line). (M–O) 405-kyr tuned PCA-F1 and its ~100-kyr (passband: 0.007 to 0.012 cycles/kyr) (thin gray line) and 405-kyr (passband: 0.00196 to 0.00298 cycles/kyr) (thick black line) Gaussian filtered curves, and polarity zones based on the tuned age model. Except (D, K, and L), other curves are drawn on a reversed scale. Numbers of E₅₇ to E₆₈ in H and K mean the 405-kyr Eccentricity Cycles. Stratigraphic positions in meters and tuned ages of the measured polarity zones are shown in SI Appendix, Table S1.

A magnetostratigraphy study observes four normal (N1-N4) and three reversed (R1-R3) polarity zones (Fig. 2 and SI Appendix, Fig. S4). Three zircon U-Pb ages of two tuffite layers in this section have been reported (41). Sample D4 is at ~890 m (23.4 ± 1.1 Ma). Samples D1 and D3 are at ~588 m and have ages of 24.87 ± 0.44 and 24.83 ± 0.37 Ma, respectively. Based on the constraint of these tuffite ages, we correlate R2 with C6Cr. Then N1, N2, and N3 can be naturally correlated with C6Cn.2n, C6Cn.3n, and C7n.2n, respectively. The top of N4 can be correlated with C8n.1n., but the base of N4 remains questionable. Four poorly determined polarity zones (N2-1, R3-1, R4-1, and R4-2) may be correlated with C7n.1n, C7r, C8n.1r, and C8r, respectively.

To better constrain the chronology and polarity correlations, PCA-F1 (the first principal component of Redness, SIRM (saturation isothermal remanent magnetization), S-ratio, and HIRM (hard IRM)) was tuned to 405-kyr eccentricity solution using the Acycle software (version 2.4.1) (46) (Fig. 2 G–O and SI Appendix, Figs. S6–S11 and Table S2). Then, we can obtain the astronomically tuned ages of the Dazecuo section which range from 27.32 to 23.24 Ma. This constrains that the base of N4 is correlated with the lower C9n and N1 is correlated with C6Cn.3n. In addition, magnetostratigraphic linear ages of paleoclimatic sequences which range from 27.44 to 22.88 Ma were estimated by piecewise linear interpolation between age control points according to the magnetostratigraphy excluding the poorly determined polarity zones. Please see SI Appendix, Method 2 for detailed chronology establishment.

The sampling interval has an average temporal resolution of ~2-kyr based on the chronology of the section (27.32 to 23.24 Ma, 2,129 bulk samples), sufficient to assess orbital forcing responses. The 2π multi-taper method power spectrum of the tuned PCA-F1 shows significant short (~100-kyr) and long (405-kyr) eccentricity, obliquity (41-kyr), and precession (19- and 23-kyr) cycles (SI Appendix, Fig. S12G). The 95.5-kyr cycle has the strongest spectral power. Other nondominant obliquity and precession cycles, such as 1,203-, 176-, 36.5-, 15.4-kyr, etc., are described in SI Appendix, Method 3. Wavelet analyses of the PCA-F1 suggest distinct short and long eccentricity and obliquity cycles based on both the tuned and untuned linear age models (Fig. 3 A and B). Precession cycles are significant in the tuned age model, mainly in the lower part of the section. This was supported by wavelet and power spectra when the 27.3 to 25.8 Ma and 25.8 to 23.2 Ma intervals of the section are separately calculated (SI Appendix, Fig. S12). The 117- and 95-kyr cycles before and after 25.8 Ma are potential responses to short eccentricity forcing. The 53- and 40.7-kyr cycles before and after 25.8 Ma are all responses to obliquity forcing. The 23.6- and 19.5-kyr are potential responses to precession forcing before 25.8 Ma. There are no significant precession cycles after 25.8 Ma, but the 95-kyr cycle is robust. Spectral and wavelet analyses show a distinct shift in the periodicity of dominant cycles from combined eccentricity, obliquity, and precession before 25.8 Ma to eccentricity and obliquity after 25.8 Ma. The short eccentricity cycles strengthen remarkably after 25.8 Ma, indicating dominant eccentricity-paced paleoclimatic oscillations (Fig. 3 A and B).

Fig. 3.

Wavelet power spectra of different sequences during the late Oligocene. (A and B) PCA-F1 C2-8 is based on the 405-kyr tuned age and untuned magnetostratigraphic linear age models, respectively. C2–8 is EEMD components 2~8 of PCA-F1 after the Hilbert–Huang transform (SI Appendix, Fig. S14 and Method 3). (C) Gamma-ray record from the Jianghan Basin (31). (D) CENOGRID benthic δ¹⁸O from ODP Site 1264 (32, 35, 47). (E) La2010d-Ecc3L eccentricity solution (45). (F–I) Wavelet spectra of (A–D) after removal of >40-kyr periodicities through highpass filtering with a frequency of 0.025 cycles/kyr to display the potential precession signals, respectively. (J) 30°N summer (June-July-August) insolation (48). Global wavelet power spectra of (F–J) are displayed on their right. The 95% CI for the wavelet spectrum is outlined in black. White dashed lines mean locations of the ~100-kyr eccentricity periodicities.

Lithology and Proxy Fluctuations Indicate Monsoon Variability.

The top part of the Dazecuo section (1,020 to 1,100 m) consists of coarse-grained light red sandstone and gray conglomerate packages with high-energy flood alluvial facies (Fig. 2A). The lower part (0 to 420 m) is characterized by interbedded reddish-brown mudstone and sandstone, representing the alternating occurrence of shallow lacustrine and fluvial facies (27, 43). This indicates the 0 to 420-m interval has weak hydrological cycles with cyclicity in a relatively small amount of precipitation when compared with the 420 to 1,020-m interval. For the middle part (420 to 1,020 m), cyclic intercalations of grayish-green layers with reddish-brown layers are typical features (hereafter referred to as the red-green layers). The red layers comprise fine-grained laminated mudstones, siltstones, and fine sandstone successions, reflecting lake retreat and occasional oxidizing conditions in a distal alluvial fan environment and/or ephemeral subaerial exposure at the margin of a paleolake (43). Twenty-one green layers mainly comprise finely laminated mudstone except for G14, G18, and G19 layers that comprise mudstone and fine-grained siltstone (Figs. 1D and 2A).

Sediment color has a close link with the oxidation state of iron (49). Hematite, as a pigment of sediment, occurs mainly in oxidizing environments and it will dissolve in reducing conditions (49, 50). Sediment color will transform from red, brown, or brown-red to olive green, grayish-green, or light to dark gray or even black colors in association with a reduction of Fe³⁺ to Fe²⁺ (49, 51). Therefore, redness is a useful indicator of diagenetic alteration associated with variation in the sedimentary environment. For the 420 to 1,020-m interval, redness is higher for red layers compared with green layers, indicating hematite is enriched in the former (Fig. 2B). During deposition, when the Dingqing Formation was in an arid or semiarid environment, it accumulated a large amount of hematite, which gives the reddish-brown color for the sediments. When paleoclimate became wetter allowing high rates of organic matter deposition and burial, oxygen is rapidly consumed in the sediment, resulting in an anoxic environment (52) in which reduction could induce the release of soluble ferrous iron into sediment porewaters (53). Fe²⁺ was then adsorbed or displaced into clay minerals, and the sediment took on a grayish-green color (54). Twenty-one green layers indicate subaqueous lake and perennial anoxic conditions with higher lake levels (43). Cyclic lithology and sedimentary color represent periodically hydrodynamic changes of lacustrine transgressive–regressive cycles and alternation of relative reducing and oxic environments in the paleolake.

Four parameters, including SIRM, S-ratio, HIRM, and Hm/Gt (hematite/goethite) are used to trace the paleoclimate response of magnetic minerals with variable coercivity. While the IRM is imparted with a 2.3 T field, SIRM approximatively represents the total contribution of all magnetic minerals in a bulk sample; the S-ratio and HIRM are widely used to observe the absolute and relative concentrations of antiferromagnetic minerals in mineral mixtures (55). For the 420 to 1,020-m interval, SIRM is higher in the red layers (Fig. 2C). Rock magnetic results indicate that magnetic minerals of red layers are dominated by hematite (SI Appendix, Method 4). Therefore, the low SIRM of green layers was caused by decreased hematite concentration. This was supported by a higher S-ratio in the green layers, which indicates decreased concentrations of antiferromagnetic minerals relative to ferrimagnetic material (55) (Fig. 2D). HIRM at Nima mainly represents hematite because goethite contributes little to an IRM imparted with a 2.3 T field due to the high coercivity and weak magnetization of goethite (51). Low HIRM of green layers further demonstrates that hematite concentration is decreased (Fig. 2E). Hm/Gt, as a proxy for the relative abundance of hematite and goethite, has a higher value representing a drier environment (51). Low Hm/Gt of green layers was also caused by decreased hematite concentration, indicating a wetter environment (Fig. 2F).

Recent iron mineralogy research on a Mesoproterozoic lacustrine record of shales, siltstones, and sandstones suggests that the deepest water lithologic facies is characterized by a lack of hematite and a very weak magnetization carried by trace magnetite with possible pyrite formed by reductive dissolution and sulfidation, indicating an anoxic environment (50). At intermediate water depths, sediments are dominated by a mixture of hematite and magnetite indicative of low oxygen conditions; in the shallowest water sediments, almost all the iron oxide was oxidized to its highest oxidized form, hematite (50). At Nima, red layers are dominated by hematite, indicating that shallow water had oxic conditions. Green layers are dominated by low-coercivity components and have negligible hematite (SI Appendix, Method 4), indicating that green layers were formed in the deeper paleolake in an anoxic environment that favors the reduction dissolution of iron oxides (49, 50). Magnetic mineral dissolution played an important role in controlling variations of environmental magnetism parameters in the Dazecuo section. A qualitative model is used to illustrate the mechanism of magnetic property changes in the section (SI Appendix, Fig. S17). Paleoclimatic explanations of environmental magnetism parameters in this study are shown in SI Appendix, Table S3.

Environmental magnetic parameters are all characterized by a shift to a higher amplitude at 420 m (~25.8 Ma) when the green layers emerge, which coincides with a significant lithological shift (Fig. 2 and SI Appendix, Fig. S13). Distinct low Redness, SIRM, HIRM, and Hm/Gt values combined with high S-ratio values in green layers indicated a significant decrease in total magnetic mineral concentration and hematite contribution. Magnetic proxies with the same rhythmic variation as the lithology and color confirm the presence of changing redox conditions in terms of magnetic minerals. The significant loss of hematite in green layers suggests partial postdepositional hematite dissolution under reductive and anoxic subaqueous conditions, indicating a relatively humid climate with a deeper lake compared with the red layers (49–51).

An analog occurs in Mediterranean sapropel-bearing sediments where the reductive dissolution of magnetic minerals led to weak magnetizations in the sapropels (56). When primary productivity increased and/or the ventilation of the bottom waters diminished, a sapropel formed under conditions of oxygen consumption and organic matter retention. Thus, the sapropel cycles of the Mediterranean have been linked to the northward penetration of the African summer monsoon precipitation (57). Stratigraphic units with low magnetization intensities or low magnetic susceptibility in sedimentary sequences from the Tianshui, Guide, and Lanzhou basins in the northern TP (58–60) and Nihewan Basin in North China (61) have been attributed to partial postdepositional detrital magnetic mineral dissolution in anoxic subaqueous conditions and related to high regional precipitation and a relatively deep/large lake.

Therefore, we interpret the lithology changes from red to green layer in the Dazecuo section to reflect the transition from shallow water hematite-rich to deep-water iron oxide-poor. The repetitive alternations of lithology and proxies reflect the climatic forcing of lake expansion–contraction and indicate orbitally paced, arid–humid hydrological oscillations. Although there was active thrusting that constrained the Nima Basin, it is hard to imagine that tectonics could have influenced the sedimentary facies at astronomically orbital timescales based on the distinctive cyclostratigraphic signals in the Dazecuo section. We posit that provenance changes and local tectonics did not dominate orbital scale proxy variations at Nima. For the 0 to 420-m interval, there is no green layer, indicating a constant and steady oxic condition. Proxy variations suggest water-depth fluctuations but in a shallow range (50). This scenario is different from the alternating shallow and deep-water changes for the 420 to 1,020-m interval.

Continued regression of the Paratethys from the northern TP since the Eocene would have reduced the moisture transported by westerly circulation and contributed to progressive aridification of the Asian inland (18, 62). A recent study demonstrated that westerlies were weaker and deflected northward during the Pliocene warm period (63). In this context, westerly moisture into the TP would be further reduced progressively under the stepwise retreat of Paratethys and the late Oligocene warming, and would not have been the main water vapor source for the Nima humidification. Carbon and oxygen isotope analyses of carbonates from the central TP support a dominant moisture source from the Indian Ocean after the Eocene (64, 65). Fossil biota including fish, insects, and plants also indicate that the TP hinterland involving the Nima and Lunpola basins was a warm lowland influenced by tropical humidity from the Indian Ocean during the late Oligocene (66). The hydrological cycle interacts with monsoon circulation which transported moisture from the tropical oceans to inner Asia (21). Therefore, we propose that arid–humid cycles at Nima from 27.3 to 23.2 Ma can be attributed to a SAM trigger scenario. When the summer monsoon is strong, it would bring more summer rainfall into the studying areas, leading to a wetter environment. Conversely, a weak summer monsoon would result in a relatively dry environment when compared with the environment under a strong monsoon. This was supported by pronounced short eccentricity and precession cycles in the paleoclimatic series of the Nima Basin (Fig. 3 and SI Appendix, Fig. S12), which suggest low-latitude forcing of monsoon variability (15, 21, 25, 27, 28). This scenario is like dry–wet analogs of paleolake sediments in the northeastern TP (15, 25, 28), central China (27, 31, 67) as well as the Chinese loess–paleosols sequences (20) where short eccentricity and/or precession cycles in sediments have also been interpreted as precipitation responses to Asian monsoon variability.

We conclude that the SAM had already advanced into the central TP (~32ºN, an extra-tropical region) at least by ~27.3 Ma, although monsoon precipitation before 25.8 Ma is relatively weak compared to the time interval after 25.8 Ma. Our paleoclimate simulations also suggest that winds and precipitation of SAM had reached the TP interior during the late Oligocene (Fig. 4 and SI Appendix, Figs. S18–S22). These findings indicate that the timing of the northward migration of SAM across the TP and the start of the modern Asian monsoon pattern should be in the Paleogene rather than the traditionally considered Oligocene/Miocene transition. The Late Oligocene warming event which seems likely to have initiated at ~27.5 Ma (32, 34, 35) could account for the appearance of SAM at Nima at 27.3 Ma because global warming could intensify monsoonal circulation via northward migration of the thermal equator (19, 20). This inference is supported by the similarity of trends between PCA-F1 and CENOGRID (Fig. 5 A–C). The trend of PCA-F1 encompasses 22% of the total variation (SI Appendix, Fig. S14A) and represents the overall process of getting more humid in the Nima Basin. It is a piece of important evidence to support that the appearance of SAM at Nima at 27.3 Ma is a response to the Late Oligocene warming event.

Fig. 4.

Simulated Asian climate in response to the Tibetan Plateau evolution. Differences in the simulated annual precipitation (units: mm/year) due to the uplift of modern Asian topography with the maximum heights increased from 1,000 to 2,000 m (A and E), from 1,000 to 3,000 m (B and F), from 2,000 to 3,000 m (C and G), and the uplift and northward motion of the main TP (D and H) under the early Oligocene (A–D) and early Miocene (E–H) boundary conditions. Levels of topographic differences equal to 500, 1,000, and 1,500 m are highlighted with gray contours. Please see SI Appendix, Figs. S18–S22 for the raw simulated results and SI Appendix, Method 6 for experiments. The red dots represent the modern positions of the Nima Basin (31.82°N, 87.68°E, ~4,600 m elevation). Paleoposition of the Nima Basin is marked in Fig. 1B.

Fig. 5.

Comparison of Nima paleoclimate records with global change during the late Oligocene. (A) CENOGRID sea surface air temperature (32). (B) CENOGRID benthic δ¹⁸O from ODP Site 1264 (32, 35, 47) (left axis). The blue area represents the minimum ice volume contribution (right axis) to the δ¹⁸O record (35). MOGI, Mid-Oligocene Glacial Interval, means a generally cold but highly unstable mid-Oligocene time interval (~28.0 to 26.3 Ma) (35). OMT, the Oligocene–Miocene transition (~23.7 to 20.4 Ma) (35). Small arrows with ages indicate the selected significant ice sheet changes. (C) Paleoclimatic oscillation (405-kyr tuned PCA-F1) at Nima. The black line is the trend of PCA-F1 after the Hilbert–Huang transform (SI Appendix, Fig. S14A). Numbers G1–G21 mean 21 grayish-green layers marked using green shadows. Numbers g1–g5 mean potentially absent green layers. (D and E) ~100-kyr (passband: 0.007 to 0.012 cycles/kyr) (thin lines) and 405-kyr (passband: 0.00196 to 0.00298 cycles/kyr) (thick lines) Gaussian filtered curves for PCA-F1 and eccentricity solution, respectively. Short gray lines indicate potential correlations between the filtered curves. (F) La2010d–Ecc3L eccentricity variability (45) (thin ice blue line) and its amplitude envelope line calculated using the “Origin” software (thick blue line and area) (left axis), and 30°N summer (June–July–August) insolation (thinner black line, right axis). (B–D) are drawn on a reversed scale. Six vertical gray bars (numbered 1-6) represent the low-value areas of 405-kyr eccentricity after 25.8 Ma, respectively. Numbers of E₅₈–E₆₇ in E mean the 405-kyr eccentricity Cycles.

Discussion

Given the intertwined nature of monsoon processes and mechanisms at multiple timescales from tectonic and orbital to millennial and even interdecadal scales (21), discussion of the monsoon should be based on considering different timescales. Here, we discuss the SAM evolution and its mechanism first on the tectonic timescale, and then on the astronomically orbital scale. The “tectonic timescale” in this study refers to a timescale in which paleoclimate changes slowly at greater than a million years. The “orbital scale” in this study mainly refers to a timescale <405-kyr in which paleoclimate changes were paced by eccentricity, obliquity, and precession.

Tectonic Forcing of the Intensified SAM at ~25.8 Ma.

At 25.8 Ma, the occurrence of the first green layer and the increased amplitude of the proxy series reveal a remarkable environmental shift from stable shallow water conditions before 25.8 Ma to alternating shallow, intermediate, and/or deep-water conditions after 25.8 Ma. This indicates a step-like increase in rainfall associated with a greatly intensified SAM. The SAM intensification occurred when the mean annual precipitation changed from 400 to 850 mm before 25.8 Ma to 800 to 1,000 mm after 25.8 Ma based on sporopollen records of the Dazecuo section (41) and is supported by the increased biological diversification in the south TP during the late Oligocene (68). We attribute the pronounced hydroclimate change at 25.8 Ma to the coeval Asian climate-pattern rearrangement.

In general, if there was no land–ocean thermal contrast, the tropical monsoon would fluctuate in low-latitude areas in response to the ITCZ. The Dazecuo section was situated at ~34.8°N, 79.9°E at 26 Ma according to paleogeographic reconstruction derived from Ocean Drilling Stratigraphic Network (ODSN) (https://www.odsn.de/odsn/services/paleomap/adv_map.html); paleomagnetism also demonstrates that the paleolatitude of the Nima Basin was approximate to its present position in the late Oligocene (42) (Fig. 1B). Therefore, the basin was not located within the tropics during the late Oligocene and would not experience an ITCZ-driven tropical monsoon. Enhanced greenhouse conditions due to higher atmospheric CO₂ content during the Eocene have been proposed to have caused modern-like Eocene monsoonal rainfall (5). However, proxy reconstructions do not support atmospheric CO₂ playing a defining role in late Oligocene SAM intensity (6). The persistent decrease of atmospheric CO₂ from the late Eocene to Miocene would have caused global cooling, and consequently, intensified Asian aridity and the winter monsoon rather than intensifying the summer monsoon (4, 60, 69). Furthermore, the Late Oligocene warming event initiated at ~27.5 Ma is about 2 Ma earlier than the inferred SAM intensification at 25.8 Ma. Therefore, atmospheric CO₂ and global warming are evidently not determining factors for the 25.8 Ma SAM intensification. This is also supported by recent paleoclimate numerical simulation studies, which suggest that Asian monsoon evolution was controlled by paleogeography, and not CO₂ (10, 13, 24).

Consequently, we propose that regional orography due to the TP uplift must have played a dominant role in forcing the tectonic-scale SAM intensification (4, 10, 13, 24). Previous studies have documented that extensive tectonic deformation and magmatism occurred across most of the H–TP orogen during the late Oligocene (36, 70), indicating tectonic uplift of the H–TP on a large scale that was synchronous with SAM intensification. Paleoelevation reconstructions suggest an elevation transition of the Himalaya orogen from 1,000 m above sea level to more than 2,000 m during the late Oligocene (37, 38) and uplift of 1,500 to 3,000 m of the Qiangtang and Songpan-Ganzi terranes during the late Oligocene-early Miocene (39). The major turnover ecosystem from tropical/subtropical ecosystems to a cooler alpine biota on the TP around the Oligocene/Miocene transition and increased biological diversification in South Asia and the Himalayas since the late Oligocene also suggest a rapid H–TP uplift (66, 68). The widespread tectonic uplift in the late Oligocene would reach a paleoelevation threshold of the TP for enhancing the coupling between the uplifted TP and the SAM at 25.8 Ma, even though we cannot precisely constrain the amplitude and scope of the uplift. Simulations show that increased elevations and summer solar radiation are most effective in strengthening the Asian monsoon when the TP elevation is only half that of today (9).

An important factor is the thermal effect of the H–TP (23). The plateau acts as a heat source during summer, which can contribute to a strong atmospheric ascent due to thermal convection locally and the large-scale South Asian high in the upper troposphere (9). This would greatly enhance the north–south temperature gradients and, consequently, the SAM (9, 21). Intensified heating of a high plateau can also amplify the low-pressure system on the TP and further amplify the sea–land pressure contrast, resulting in northward propagation of monsoon and penetration further inland with additional precipitation (12). Moreover, the released latent heat in the monsoon rains associated with the radiative heating of clouds could form positive feedback between the monsoon circulation and rainfall (21). A higher TP represents a large air pump and enhances both sensible heat south of it and latent heat on it, and provides an enhanced summer heat source thus strengthening the positive feedback and monsoon circulation (9, 12, 23, 29).

We hypothesize that the late Oligocene H–TP uplift could enhance these thermal effects, and land–ocean thermal and atmospheric pressure contrast, leading to the intensified SAM and its northward migration into the central TP. To test this hypothesis, 14 sensitivity experiments were conducted with Community Atmosphere Model version 4 (CAM4) under both early Oligocene and early Miocene boundary conditions (SI Appendix, Method 6 and Table S4). These two periods were chosen because they are approximately close to the beginning and end of the deposition of the Dazecuo section and can better constrain the paleoclimatic effects of TP uplift during the time interval of our section. The effects of TP uplift under the two boundary conditions are similar, indicating common features for the early Oligocene (Fig. 4 A–D) and the early Miocene (Fig. 4 E–H). In a hypothetical scenario of linear uplift of modern Asian topography, the elevated TP could strengthen the winds of the South Asian summer monsoon and increase precipitation in the northern Indian Peninsula and also over the TP interior through the strengthened land–sea thermal contrast (SI Appendix, Figs. S18–S21). In particular, the precipitation in the TP interior increased more when the maximum heights of the topography increased from 1,000 to 2,000 m (Fig. 4 A and E) compared with uplifts from 1,000 to 3,000 m (Fig. 4 B and F) and from 2,000 to 3,000 m (Fig. 4 C and G). The Asian topography that was used in the above experiments may be idealized although the paleomagnetic reconstruction suggests that the paleolatitude of the Nima Basin was approximate to its present position in the late Oligocene (42). If we consider the uplift and northward motion of the main-TP between the late Oligocene and the early Miocene (37, 39), simulations suggest that these tectonic processes also increased the annual precipitation over the TP interior by increasing the water vapor supply and intensifying the regional ascent of the atmosphere (Fig. 4 D and H and SI Appendix, Fig. S22). Meanwhile, due to the limited extent of the main TP, the anomalous dry northwesterlies to the west of the main TP decreased the amount of water vapor over the northern Indian Peninsula, and the precipitation thus decreased in this region (SI Appendix, Fig. S22). No matter what the scenario is, these model results confirm that the uplift of the TP increased the precipitation over the TP interior and support our geological records and interpretations.

Simultaneously, the late Oligocene TP uplift also increased regional aridification in the Asian inland owing to the retreat of the Paratethys sea and the evolution of a TP rain-shadow effect (7, 62), and humidification in south China associated with the EAM (4). Reorganization of Asian climate patterns during the late Oligocene at the continental-scale highlights the orographic and thermal effects associated with the surface uplift of the H–TP orogen (4, 10, 13).

Paleorecords indicate that the ITCZ may have migrated toward a warming hemisphere together with a migration of the concomitant rain belt (71). The northern edge of the ITCZ appears to have migrated to near 28ºN at 25 Ma, when the TP significantly rose and the land-sea thermal difference increased (24) (Fig. 1B). This migration would enhance summer rainfall in the TP interior. The rainfall belt of the ITCZ would cover the Nima Basin. In this context, rainfall in Nima was essentially influenced by the seasonal migration of the ITCZ associated with its concomitant rainfall zone. These mechanisms can account for the SAM intensification and shift of hydrological cycles at 25.8 Ma on the tectonic timescale.

Orbital Variation of SAM Precipitation.

Although the tectonic-scale SAM was strengthened by plateau uplift, oscillations of monsoon precipitation were orbitally forced by combined eccentricity, obliquity, and precession cycles before 25.8 Ma and predominately by eccentricity cycles after 25.8 Ma (Fig. 3 A and B and SI Appendix, Fig. S12). The shift of paleoclimate proxy and periodicity represents the shift of climate response to orbit. When comparing the filtered time series, the 100-kyr Gaussian band-pass filters for the Nima paleoclimatic oscillation and eccentricity show that their phases are highly consistent after 25.8 Ma (Fig. 5 D and E), demonstrating that eccentricity forcing and increased precipitation associated with SAM intensification were strongly coupled after 25.8 Ma.

The eccentricity-paced EAM in the northeastern TP during the Miocene has been attributed to periodic expansion and contraction of the Antarctic ice sheet via ocean processes and/or the global carbon cycle (15, 25, 28). Periodic waxing and waning of the ice sheet could result in periodic variability of many geological processes of the ocean including sea level, sea surface temperature, intermediate and deep-water production in the Southern Ocean, the cross-equatorial pressure gradient, and the amount of latent heat release. These processes further regulate the orbital-scale changes of monsoon moisture, finally leading to orbital-scale changes in the Asian summer monsoon including its intensity, northeastern edge, and precipitation extent (15, 25, 28). For the 25.8 to 23.2 Ma time interval, six low-value phases of 405-kyr eccentricity cycles are shaded in gray (termed 1-6) (Fig. 5). The weakly developed green layers G12, G14, and G18–G19 occur at phase 4, 5, and 6, respectively. According to the 100-kyr filter correlation between proxy oscillation of the Dazecuo section and eccentricity, we speculate that several green layers may be missing at Nima, such as g1-g5 layers, which all occur within the time intervals of low 405-kyr eccentricity except phase 2. The g1-g3, g4, and g5 correspond to phases 1, 3, and 4, respectively. The weakly developed or possibly absent green layers indicate weak monsoon precipitation during these time intervals. Antarctic ice volume reconstruction suggests that these phases correspond well in time to intervals of ice sheet expansion (Fig. 5B). Ice volume was paced by eccentricity (35, 47). Therefore, weak monsoon precipitation at the six phases on the 405-kyr timescale could be directly caused by a response to the long eccentricity-modulated precession of Earth’s spin axis, or by Antarctic ice sheet expansion through interactions with the global carbon cycle (28, 35, 72). Although we cannot determine the causal relationship between Nima precipitation and the Antarctic ice sheet, they are in phase on the 405-kyr timescale.

However, the mechanism of Antarctic ice sheet forcing cannot explain the SAM variability on the short eccentricity timescale for the late Oligocene. The short eccentricity and obliquity cycles at Nima were not generated by the Antarctic ice sheet for the following reasons: 1) The tectonic-scale SAM intensification at 25.8 Ma was caused by the H–TP uplift rather than global change and ice sheet. The pattern of orbital eccentricity cycles at Nima differs from the persistent and stable 100-kyr cycles of La2010d eccentricity solution (Fig. 3E) and differs from the 100-kyr cycles of CENOGRID during 27.5 to 23 Ma, which is a combination of global ice volume and temperature (32, 35, 47) (Fig. 3D). The short eccentricity signals at Nima were significantly enhanced after 25.8 Ma, but the theoretical eccentricity series and CENOGRID do not show an increase at that period (Fig. 3 and SI Appendix, Fig. S12). Moreover, during periods of rapid glacial retreat or temperature increase at about 26.3, 26.1, 25.5, and 24.7 Ma, there is no green layer formation at Nima (Fig. 5 A–C). During ~24.2 to 23.9 Ma when distinct warming and ice volume significantly decreased, there are only weakly developed green layers G12 and G14 except G13. Moreover, at the beginning of the Oligocene–Miocene transition (around 23.7 Ma), the temperature decreased and ice volume increased significantly, but green layers G16 and G17 developed well. Therefore, the formation of alternating red-green layers and 100-kyr cycles at Nima was less responsive to the waxing and waning of the Antarctic ice sheet and its associated oceanic feedback. 2) It was the low-latitude tropical Indian Ocean, rather than the Antarctic ice sheet, that directly responds to the H–TP uplift (29). Conversely, some studies suggest that low-latitude interhemispheric insolation gradients primarily played an important role in influencing or facilitating the glacial–interglacial cycles of the Antarctic ice sheet via monsoon dynamics or cross-equatorial heat flow (28, 29, 73). 3) The precipitation at Nima is essentially a type of precipitation mainly influenced by the ITCZ. The ITCZ is a low-latitude tropical climate feature, although it can be influenced by the Antarctic ice sheet (29). Pronounced precession cycles in the Dazecuo section demonstrate a low-latitude forcing of monsoon variability (15, 21, 25, 27, 28). 4) Before the emplacement of the Antarctic ice sheet which occurred at the Eocene-Oligocene transition (32–34), an EAM already existed in the middle Paleogene (~40 Ma) (5). Eccentricity and obliquity cycles occurred in the extra-tropical Jianghan Basin (31°N) from 40 to 34 Ma (31), which is earlier than the emplacement of the Antarctic ice sheet. Therefore, these orbital responses are not necessarily related to the ice sheet. 5) Numerical simulation indicates that obliquity-induced changes in a tropical climate can occur without high-latitude ice sheet fluctuations (74). Obliquity can modulate the spatial distribution of insolation and exert a direct effect on monsoon climate (22) via the cross-equatorial meridional insolation gradient in the summer hemisphere that modulates the thermal contrast between the Asian continent and surrounding oceans (74). Therefore, the obliquity cycles in the Nima Basin are also not necessarily related to the Antarctic ice sheet.

Alternatively, the prominent eccentricity-paced variability of the SAM not only passively follows the glacial-interglacial signals from high latitudes but can be forced directly by low-latitude summer insolation (17, 21, 28). Eccentricity can influence insolation by modulating the precession cycles via a nonlinear response mechanism (28, 75, 76). Precession affects the seasonal cycle of incoming solar radiation and contributes to mid-low-latitude summer insolation on both hemispheric distributions and is therefore thought to be a major control of orbital changes in monsoon intensity (21, 22). There is a strong and positive response of boreal summer monsoon precipitation to boreal summer insolation forcing (77). During minimum precession, the seasonal cycle of tropical precipitation is enhanced. Numerical simulations indicate that precession dominated SAM variability over the last 30 Ma; the precessional effect on SAM rainfall is more prominent than the summation effects of both obliquity and eccentricity during 30 to 23 Ma (13).

The noteworthy relationships between eccentricity maximum and wetter conditions during the late Oligocene at Nima are in line with forcing from insolation (Figs. 3 F and J and 5 C–F). The formation of the green layers reflects the influence of high solar radiation. Therefore, orbital short eccentricity-paced SAM after 25.8 Ma at Nima demonstrates a pronounced nonlinear climate response to eccentricity-modulated precession forcing of insolation. As an analog, the Mediterranean sapropel layers, which reflect the African monsoon also formed during a high eccentricity period and correspond to high solar radiation during low precession in the Northern Hemisphere (57). The precession-paced precipitation before 25.8 Ma indicates a direct response of the SAM to low-latitude insolation forcing. These two types of responses are also in line with Chinese speleothem records (16).

The precession signals at Nima are weak after 25.8 Ma (Fig. 3 F and G and SI Appendix, Fig. S12). The TP uplift strengthened SAM precipitation after 25.8 Ma. We suggest that fluctuations of lake level on the eccentricity scale suppress precession-scale magnetic mineral diagenesis, thus weakening records of precession cycles. The distinct precession signals before 25.8 Ma represent the precession-forced hydrologic cycles. The water depth fluctuates under a limited range as indicated by the alternating shallow lacustrine and fluvial facies in the interbedded mudstone and sandstone units (Fig. 2). Therefore, below 420 m, variation of magnetic mineral diagenesis is easily affected by precession-scale climate fluctuation and can record the precession signals. After 25.8 Ma, eccentricity-scale red-green alternating units represent more intense hydrological cycles. A larger fluctuation of eccentricity-scale lake level was made up of eccentricity-modulated precession-scale precipitation cycles. On the eccentricity scale, the limited precession-scale fluctuation of water depth would not significantly affect the diagenetic environment of magnetic minerals under the eccentricity-scale water depth. In other words, the smaller amplitude precession signals may be overprinted by the larger amplitude and longer eccentricity periodicities (28). These orbital features are different from the EAM records. Wavelet analysis of gamma-ray from the Jianghan Basin shows that short eccentricity and precession signals are basically continuous and stable over time (Fig. 3 C and H). Enhanced SAM precipitation at Nima after 25.8 Ma is responsible for the enhancement of eccentricity signals and suppression of precession signals. Although the EAM and SAM were both induced by the TP uplift (4, 21), there are different climate responses in the east Asia and south Asia areas. During the late Oligocene, we speculate that the EAM may be more sensitively responsive to the orbital and/or Antarctic ice sheet forcing, while the SAM was regulated by the tectonic and orbital forcing.

We propose a mechanism to explain how insolation forces hydrological cycles in the Nima Basin on the orbital timescale. Increased northern hemispheric insolation drives warming during the boreal summer and enhances the land–sea thermal contrast and the atmospheric pressure gradient between the H–TP and the Indian Ocean (21, 77). Consequently, atmospheric circulation was strengthened. This triggered a northerly shift of the ITCZ and the strengthened Hadley circulation enhanced SAM rainfall across the central TP (13, 29). In sum, changes of low-latitude insolation caused by eccentricity-modulated precession cycles, in turn, influenced the relative strength of SAM by the meridional and cross-equatorial shifts of the ITCZ via Hadley Cells (10, 13), leading to periodic hydrologic cycling in the Nima Basin. While we emphasize the dominant low-latitude orbital forcing of the SAM at Nima, high-latitude climate impacts cannot be neglected in influencing the mid-low-latitude hydroclimate on orbital timescales (13, 15, 28). However, high-latitude effects should be weak on Nima rainfall. Finally, whether the Antarctic ice sheet or tropical processes dominate the paleoclimatic variability in the Nima Basin, they both need to regulate low-latitude hydrologic cycling, then influence the periodic precipitation in the basin.

In summary, during the late Oligocene, an enhanced tropical SAM and its northward expansion to the boreal subtropics were dominated by the effects of orogeny and Earth’s orbital configuration at multiple timescales. Two principal drivers involving internal forcing (plateau uplift) and extraterrestrial forcing (solar radiation) together controlled the evolution of the late Oligocene SAM. The former affected planetary-scale monsoon circulation and intensified monsoon and was the primary boundary forcing factor at the tectonic timescale. The latter regulates orbital monsoon variability and eccentricity-paced hydrological cycles and was the primary forcing factor at the orbital timescale. Tectonic and orbital forcing on intensified South Asian summer monsoon in the late Oligocene improves our current understanding of the controlling factors of the SAM. Finally, this study would supply boundary conditions for future paleoclimatic numerical simulations.

Materials and Methods

From the Dazecuo section (thick of 1,018.27 m; 31.82°N, 87.68°E), 565 oriented block samples and 2,129 bulk samples were sampled (SI Appendix, Method 1). To establish the section’s chronology, oriented samples were step-wise demagnetized thermally for magnetostratigraphy; then the first principal component (PCA-F1) of Redness, SIRM, S-ratio, and HIRM was astronomically tuned to better constrain the chronology (SI Appendix, Method 2). The 405-kyr tuned PCA-F1 series was then performed wavelet and power spectral analyses to investigate climate variations on orbital time scales (SI Appendix, Method 3). To observe magnetic mineralogy, 10 representative bulk samples were carried out for rock magnetism analysis, including temperature-dependent susceptibility (χ-T), three-axis IRM thermal demagnetization, IRM acquisition curves, and hysteresis loops (SI Appendix, Method 4). To construct paleoclimatology sequences, 2,129 bulk samples were measured for their environmental magnetic parameters, including Redness, SIRM, S-ratio, HIRM, and Hm/Gt (SI Appendix, Method 5). To test the paleoclimatic effects of the Tibetan Plateau uplift, 14 sensitivity experiments were conducted with CAM4 under both early Oligocene and early Miocene boundary conditions (SI Appendix, Method 6).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Dataset S05 (XLSX)

Dataset S06 (XLSX)

Dataset S07 (XLSX)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix. Redness, SIRM, S-ratio, HIRM, Hm/Gt, PCA-F1, and magnetostratigraphy data are available in SI Appendix, Dataset S1.