Eocene–Oligocene Crustal Thickening-Collapse of the Eastern Tibetan Plateau: Evidence from the Potassic Granitoids in SW China

The Cenozoic India–Asia convergence event generated the uplift that formed the Tibetan Plateau and surrounding regions, meaning that this area provides an excellent opportunity for research into the evolution of orogenic events and intracontinental deformation (e.g., [1–5]). However, the dynamic processes involved in the growth of uplifted plateaus remain controversial (e.g., [2, 6–10]), leading to the development of numerous models (e.g., quasi-rigid block extrusion, removal of thickened lithospheric mantle, and lower crustal channel flow) to explain this type of uplift (e.g., [2, 6, 9–13]). The rigid block extrusion model emphasizes the lateral extrusion of crustal blocks with strain localized along major tectonic boundaries represented by the Ailao Shan–Red River (ASRR) and Gaoligong shear zones [7, 13, 14]. In comparison, the lithospheric delamination model suggests that the overthickened crust formed during plateau convergence foundered, stimulating plateau uplift as a response to the removal of the convective lithosphere (e.g., [2, 6]). Finally, the channel flow model suggests that middle–lower crustal material flowed towards the periphery of the plateau from the interior, causing crustal deformation and generating the topography observed in the eastern plateau (e.g., [9, 15]).

Understanding the growth of the Tibetan Plateau requires knowledge of the structure and composition of the lithosphere in this region. Plateau growth was focused along the peripheral parts of this region, making this an ideal location to test the models for plateau growth outlined above. The igneous rocks in the northern, southern, and eastern margins of the Tibetan Plateau were all sourced from the middle to lower crust and the lithospheric mantle, meaning they provide evidence of the structure and composition of the lithosphere (e.g., [1, 3, 16–18]). Previously published research indicates that Eocene to early Oligocene alkaline igneous rocks crop out along the Jinshajiang River in the eastern Tibetan Plateau [1]. These rocks consist of nepheline gabbro, alkali basalt, latite, trachyte, nepheline syenite, diopside syenite, hornblende syenite, amphibole monzonite, and alkali granite [1, 8, 19], all of which can provide insights into the dynamic evolution of the lithosphere of the eastern Tibetan Plateau [1]. The tectonic implications of the presence of these alkaline igneous rocks remain controversial despite previous research outlining several potential petrogenetic models. These geodynamic models suggest the alkaline igneous rocks in the eastern Tibetan Plateau were formed as a result of (1) activation of the ASRR shear zone [16, 20], (2) subduction of the continental lithosphere [21], or (3) convective removal of the lithospheric mantle [6, 22–24]. This uncertainty indicates the need for further geochronological and geochemical research on these alkaline igneous rocks to more comprehensively determine the petrogenesis of the magmas that formed these rocks, the geodynamic setting in which the magmatism occurred, and the tectonic implications for the evolution of the Tibetan Plateau.

This study presents new whole-rock geochemical and zircon U–Pb and Hf–O isotopic data for Cenozoic potassic granitoid samples from 15 intrusions along or near the Jinshajiang–Ailaoshan suture zone in the eastern Tibetan Plateau. Combining these data with the results of previous research [19, 23–30] indicates that the potassic magmatism in this area occurred between the late Eocene to early Oligocene (38–32 Ma) and involved magmas derived from either thickened lower crust or enriched lithospheric mantle sources. This magmatism was a response to the orogenic assemblage and subsequent collapse of the eastern Tibetan Plateau, with analysis of the temporal and spatial progression of this magmatism allowing the construction of a thickened crustal collapse model for the formation and growth of the eastern Tibetan Plateau.

Neo-Tethyan slab subduction and the subsequent continuous convergence of the Indian and Asian continents led to the formation of the Tibetan Plateau from the Late Cretaceous onwards (e.g., [1, 4, 17, 31, 32]. The interior of the Tibetan Plateau contains the Lhasa, Qiangtang, and Songpan–Ganzi blocks, which are separated by the Bangong–Nujiang (BNS) and Jinshajiang (JS) suture zones, respectively (Figure 1(a); e.g., [1, 4, 8]). The Yangtze and Simao/Indochina blocks along the eastern margin of the Tibetan Plateau are separated by the Ailaoshan–Song Ma zone, an area that represents either a Paleotethyan back-arc basin or a branch of the Paleo-Tethys Ocean (Figure 1(a); e.g., [33, 34]).

The Yangtze Block to north of the Ailaoshan–Song Ma suture zone contains crystallized basement material of the Kongling Complex, which is dominated by Archean–Paleoproterozoic amphibolite, trondhjemite–tonalite–granodiorite, and metasedimentary units [35, 36], all of which are variably overlain by a Silurian–Triassic cover sequence [37]. The southern margin of the Yangtze Block contains the Neoarchean–Neoproterozoic Cavinh and Ailaoshan–Hoang Lien Son complexes (e.g., [38, 39]). Numerous Neoproterozoic igneous rocks are present in the Yangtze basement material in the western and southern parts of the Yangtze Block (Song Chay and Po Sen; e.g., [40, 41]). The high grade Kontum metamorphic complex is thought to represent the ancient crystalline basement of the Indochina Block (e.g., [42]). The Indochina Block also contains a similar assemblage of early to middle Devonian fish fossils and late Paleozoic equatorial to tropical Cathaysia-type fauna and flora to that found within middle Paleozoic and later units of the South China Block (e.g., [33, 43, 44]). The Indochina block is thought to record Ordovician–Silurian and Triassic tectonothermal events that again are similar to those identified in the South China Block [45, 46].

The ASRR tectonic or shear zone is a Cenozoic strike-slip fault zone that partly overlaps the Ailaoshan–Song Ma tectonic zone in the Ailaoshan area and hosts Paleo-Tethys-related Permian mafic volcanic rocks and Triassic mafic–ultramafic intrusions [32]. However, this tectonic or shear zone is located to the north of the Ailaoshan–Song Ma tectonic zone in the Red River area (e.g., [22, 47]). This zone extends ~1000 km from the syntaxis of eastern Tibet into the South China Sea across the Xuelong, Diancang, Ailao, and Day Nui Con Voi metamorphic domains [47]. The shear zone also records Oligocene–Miocene sinistral shearing that transitioned to dextral shearing sometime after the Miocene–Pliocene (e.g., [47–49]).

The study area contains small volumes of alkaline igneous rocks, including both potassic alkaline and sodium alkaline rocks. They crop out mainly along the western margin of the Yangtze and eastern syntaxis of Tibet and are generally emplaced into or have been erupted onto Paleozoic sedimentary units. These alkaline igneous rocks include nepheline syenite, nepheline gabbro, shonkinite, diopside syenite, hornblende syenite, amphibole monzonite, granophyric granodiorite, olivine alkali lamprophyre, alkaline pyroxenite, alkaline basalt, latite, and trachyte, as well as porphyritic granite plutons, some of which host porphyry Cu (Au–Mo) mineralization (e.g., the Yao’an Au, Machangqing Cu–Mo, and Beiya Au deposits [25, 50, 51]).

This study focuses on samples collected from 15 potassic granite and syenite intrusions in the Dali, Heqing, Jianchuan, Yulong, Ninglang, Yongsheng, and Yao’an areas of the eastern syntaxis of Tibet (Figure 1(b)). These intrusions are undeformed massive stocks or dikes that were emplaced into Cenozoic, late Paleozoic, and Mesozoic units and crop out sporadically within the northwestern part of Yunnan Province. The granitic samples contain ~35% K-feldspar, ~25% plagioclase, ~20% quartz, and ~15% hornblende + biotite in addition to accessory zircon, apatite, and titanite (Figure 2(a)). The porphyritic granite units contain subhedral–euhedral K-feldspar, plagioclase, and biotite and anhedral quartz phenocrysts hosted by a quartz and plagioclase dominated groundmass (Figures 2(b) and 2(c)). The syenite samples contain 60%–65% euhedral K-feldspar, 10%–15% hornblende, 5%–10% plagioclase, and 5%–10% quartz in addition to accessory epidote, zircon, apatite, and titanite (Figure 2(d)). These units also contain subhedral–euhedral K-feldspar and biotite phenocrysts, rare quartz and pyroxene phenocrysts, and a K-feldspar dominated groundmass (Figures 2(e) and 2(f)).

Samples for whole-rock elemental and Sr–Nd isotope analyses were crushed to pass a 200-mesh using an agate mill. Major element concentrations were determined using wavelength-dispersive X-ray fluorescence spectrometry at Sun Yat-sen University (SYSU), Zhuhai, China. Details of the sample preparation and analytical procedures used are given by Wang et al. [52], and the relative standard deviations of repeat analyses were <5%. Trace element concentrations were determined using inductively coupled plasma–mass spectrometry (ICP–MS) employing an iCAP-RQ instrument at the SYSU and the State Key Laboratory of Ore Deposit Geochemistry, Chinese Academy of Sciences, Guiyang, China. Details of the analytical approaches used are given by Qi et al. [53]. Sr–Nd isotope analysis was undertaken using Neptune Plus multicollector–ICP–MS (MC–ICP–MS) instrument employing nine Faraday cup collectors and eight ion counters at SYSU. Separation and purification procedures prior to analysis were undertaken at Guizhou Tongwei Analytical Technology Co. Ltd., Guizhou, China, with details of the sample preparation and analytical approaches used during Sr–Nd isotope analysis given by Wang et al. [52]. Mass fractionation corrections are based on $86Sr/88Sr=0.1194$ and $146Nd/144Nd=0.7219$ with a within-run precision better than 0.000015 for $146Nd/144Nd$ at the 95% confidence level. Analysis of the (NIST) NBS 987 and ALFA-Nd standards yielded mean $86Sr/88Sr$ and $146Nd/144Nd$ ratios of $0.710269±10$ (⁠$2σ$⁠) and $0.512419±8$ (⁠$2σ$⁠), respectively.

Prior to U–Pb dating, zircons from representative samples were separated using conventional heavy liquid and magnetic separation techniques. The internal structures of these zircons were imaged using cathodoluminescence (CL) approaches employing a scanning electron microprobe. Zircon U–Pb and in situ Lu–Hf isotope analysis was undertaken using iCAP RQ ICP–MS and Neptune Plus MC–ICP–MS instruments, respectively, both of which were coupled with a Geolas HD excimer ArF laser ablation system at SYSU. The analytical procedures used follow those outlined by Xia et al. [54] and Wang et al. [52]. 91500 (⁠$1065±5Ma$⁠) and Plešovice (⁠$337±0.4Ma$⁠) standard zircons were used for U–Th–Pb ratio calibrations and the determination of absolute U concentrations with data processing undertaken using the SQUID 1.03 and Isoplot/Ex 2.49 programs of Ludwig [55]. Details of the zircon Lu–Hf isotope analytical approaches used are given by Li et al. [56]. A 91500 standard zircon was used for external standardization during in situ zircon Lu–Hf isotope analysis, and the resulting data were normalized to $176Hf/177Hf=0.7325$⁠. The $εHft$ values were calculated using present-day chondrite values of $176Hf/177Hf=0.282772$ and $176Lu/177Hf=0.0332$ [57]. The Hf model ages (⁠$TDM1$⁠) were calculated relative to the depleted mantle using $176Hf/177Hf=0.283250$ and $176Lu/177Hf=0.0384$ values with two-stage Hf model ages (⁠$TDM2$⁠) calculated with a mean continental crust value of $176Lu/177Hf=0.015$ [58]. The in situ determination of zircon oxygen isotopic compositions was undertaken using secondary ion mass spectrometry (SIMS) employing a Cameca IMS1280-HR instrument at the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou, China. Oxygen isotopes were measured in multicollector mode using two off-axis Faraday cups with a Penglai zircon standard (⁠$δ18OVSMOW=5.3‰$⁠) used for instrumental mass fractionation corrections of the resulting O isotopic data.

The zircon U–Pb and Hf–O isotopic compositions of 19 samples were determined during this study. The zircons from these samples are generally transparent, light brown to colorless, and euhedral and prismatic. They have concentric oscillatory zonation that is visible during CL imaging, suggesting they are magmatic (Figure S1). The results of the zircon U–Pb and in situ Hf–O isotope analyses are given in Supplementary 1 and Supplementary 2 with sampling locations and dating results shown in Figure 1(b) and summarized in Table 1. Concordia diagrams for representative samples are shown in Figure S2.

A total of 12 analyses of zircons from porphyritic granite sample 16DL-2H from the Machangqing area with $Th/U$ values of 0.45–0.92 yield a weighted mean age of $37.5±0.4Ma$ (⁠$MSWD=1.2$⁠), with a further zircon yielding an older apparent age of $807±11Ma$⁠. Nineteen analyses of zircons from syenite 16DL-10A₁ from the Diancangshan area with $Th/U$ values of 0.26–0.81 yield a $206Pb/238U$ age of $35.4±0.4Ma$ (⁠$MSWD=0.3$⁠). These zircons yield corresponding $εHft=35Ma$ and $TDM2$ values from −1.5 to +0.8, and from 1.2 to 1.1 Ga, respectively. Thirteen analyses of zircons from porphyritic quartz monzonite samples 16DL-60A₁ from the Yao’an area have $Th/U$ values of 0.65–1.52 and yield an apparent age of 33.0 Ma and a weighted mean age of $33.1±0.3Ma$ (⁠$MSWD=1.2$⁠). These zircons yield corresponding $εHft$ and $TDM2$ values from −9.8 to −6.6, and from 1.7 to 1.5 Ga, respectively.

Porphyritic quartz syenite samples 16DL-23A₁ and -25A₁ from the Heqing area yield $206Pb/238U$ ages of $36.4±0.6$ (⁠$n=8$⁠, $MSWD=1.4$⁠, $Th/U=0.01–2.13$⁠) and $36.8±0.5Ma$ (⁠$n=10$⁠, $MSWD=1.0$⁠, $Th/U=0.04–3.06$⁠), respectively. The remaining ten analyses of zircons from samples 16DL-23A₁ and -25A₁ yield ages that cluster at 78 (⁠$n=1$⁠), 263–198 (⁠$n=3$⁠), 468 (⁠$n=1$⁠), 699–696 (⁠$n=3$⁠), and 795–781 (⁠$n=2$⁠) Ma. The ten analyses of zircons from sample 16DL-25A₁ with crystallization ages of 36 Ma yield $εHft$ values from −4.3 to +5.2 and $TDM2$ ages of 1.4–0.8 Ga.

Syenite samples 16DL-31A₁, -46A₁, and -46B₁ yield ²⁰⁶Pb/²³⁸U-weighted mean ages of $34.5±0.3$ (⁠$n=11$⁠, $MSWD=0.6$⁠), $36.0±0.5$ (⁠$n=15$⁠, $MSWD=0.2$⁠), and $36.8±0.5$ (⁠$n=13$⁠, $MSWD=0.6$⁠) Ma. Fourteen further analyses of zircons from these samples yield Neoproterozoic xenocryst ages of 842–735 Ma. Zircons from sample 16DL-46B₁ yield $εHft$ values from −1.4 to +1.1 and $TDM2$ model ages from 1.2 to 1.0 Ga. A further 23 analyses of zircons from porphyritic granite sample 16DL-35A₁ from the Laojunshan area have $Th/U$ values of 0.09–0.28. Of these analyses, 8 yield a coherent age cluster with a weighted mean age of $33.5±0.5Ma$ (⁠$MSWD=0.2$⁠), and the remaining 15 yield older ages that range from 1402 to 53 Ma.

Samples from the Lijiang area analyzed during this study include porphyritic quartz monzonite samples 16DL-38A₁ and 16DL-42A₁ from the Shigu intrusion and quartz syenite samples 16DL-41A₁ and -44B₂ from the Taoyuan intrusion. Seven analyses of zircons from sample 16DL-38A₁ yield a mean age of $37.0±0.4Ma$ (⁠$MSWD=0.1$⁠) with a further seven analyses yielding older ages of 833–135 Ma that reflect the presence of xenocrysts. A total of eight, ten, and seven analyses of zircons from samples 16DL-41A₁, -42A₁, and -44B₂ yield weighted mean ages of $37.4±0.4$(⁠$MSWD=0.3$⁠, $Th/U=0.18–0.43$⁠), $34.4±0.4$ (⁠$MSWD=0.2$⁠, $Th/U=0.08–0.22$⁠), and $35.1±0.8$ (⁠$MSWD=1.6$⁠, $Th/U=0.09–0.33$⁠) Ma, respectively, reflecting crystallization ages of 34.0–34.0 Ma. The remaining 35 analyses of zircons from these samples yield older apparent ages of 1856–57 Ma that are interpreted to be the ages of xenocrysts within these samples. Eight analyses of zircons from sample 16DL-41A₁ with a mean age of 37 Ma yield $εHft$ and $TDM2$ values from +1.2 to +3.7 and from 1.0 to 0.9 Ga, respectively.

A total of 27 analyses of zircons with $Th/U$ values of 0.33–1.31 from biotite granite sample 16DL-47A₁ from the Ninglang intrusion yield a weighted mean age of $34.0±0.2Ma$ (⁠$MSWD=0.3$⁠). Porphyritic biotite granite samples 16DL-49A₁ and 16DL-51A₁ from the Zhanhe pluton yield weighted mean ages of $33.6±0.3$ (⁠$n=24$⁠, $MSWD=0.8$⁠, $Th/U=0.56–1.50$⁠) and $33.6±0.3$ (⁠$n=26$⁠, $MSWD=1.0$⁠, $Th/U=0.50–0.99$⁠) Ma, respectively. Zircons from these samples have corresponding $εHft$ values that range from −1.0 to +4.4, $TDM2$ ages that range from 1.2 to 0.8 Ga, and in situ $δ18O$ values that range from 6.78‰ to 7.36‰. Finally, 18 analyses of zircons from porphyritic biotite granite sample 16DL-54A₁ yield a weighted mean age of $33.0±0.3Ma$ (⁠$MSWD=1.1$⁠, $Th/U=0.13–3.22$⁠).

A total of 21 analyses of zircons from biotite quartz syenite sample 16DL-56A₁ from the Yongsheng yield a ²⁰⁶Pb/²³⁸U mean age of $34.3±0.4Ma$ (⁠$MSWD=0.7$⁠, $Th/U=0.98–3.97$⁠). Their $εHft$ and $TDM2$ values range from −6.1 to −0.6 and from 1.5 to 1.2 Ga, respectively. Ten analyses with a crystallization age of 34 Ma from sample 16DL-56A₁ yield in situ $δ18O$ values of 6.37‰–6.89‰. A total of 21 analyses of zircons from quartz monzonite sample 16DL-58A₁ from the Yongsheng area yield a weighted mean age of $32.9±0.3Ma$ (⁠$MSWD=0.6$⁠, $Th/U=0.73–2.75$⁠) with significantly negative $εHft$ values (−11.4 to −7.7), old $TDM2$ model ages (1.8–1.6 Ga), and $δ18O$ values of 6.31‰–6.82‰.

The whole-rock major, trace, and Sr–Nd isotopic compositions of representative examples of the samples analyzed in this study are given in Table 2. The samples are split into three groups using mineralogical, geochemical, and Sr–Nd–Hf isotopic characteristics combined with previously published data (e.g., [23–30]): i.e., Group 1, adakite-like granites; Group 2, syenites; and Group 3, low-$εNdt$ granitoids.

These samples contain high SiO₂ (63.31–73.62 wt.%) and K₂O (3.97–8.78 wt.%) contents, with $Al2O3=13.53–16.06wt.%$⁠, $FeOt=0.73–4.13wt.%$⁠, $MgO=0.28–3.06wt.%$⁠, $TiO2=0.19–0.54wt.%$⁠, $K2O+Na2O=8.30–11.02wt.%$⁠, and $K2O/Na2O$ ratios of 0.81–3.93. Their $A/CNK$ (molar Al₂O₃/(CaO+Na₂O+K₂O) and $A/NK$ (molar Al₂O₃/(Na₂O+K₂O) ratios range from 0.83 to 1.16 and from 1.10 to 1.36, respectively, and they are classified as high-K calc-alkaline to shoshonitic granites and quartz monzonites (Figure 3). They contain high concentrations of Sr (495–1568 ppm) and Ba (714–2735 ppm), low concentrations of Y (<20 ppm) and Yb (<2 ppm), and have $Sr/Y$ (42.3–193.1) and $La/Yb$ (7.5–72.7) ratios that indicate they may have adakitic affinities [59] and are similar to the compositions of Eocene–Oligocene adakite-like rocks in the Qiangtang Block and the eastern syntaxis of Tibet (Figure 4).

The samples have chondrite-normalized REE patterns that dip steeply to the right and have $La/Ybn$ values of 12.2–86.7, $Gd/Ybn$ values of 2.00–4.42, and $Eu/Eu∗$ values of 0.72–1.00. Their primitive mantle-normalized multielement variation diagram patterns show enrichment in the large-ion lithophile elements (LILEs) and depletion in the high field strength elements (HFSEs), and they have $Nb/La$ values of 0.11–2.01, and positive Sr anomalies (Figures 5(a) and 5(b)). Their measured $87Sr/86Sr$ and $143Nd/144Nd$ ratios range from 0.70580 to 0.70845 and from 0.51231 to 0.51259, respectively, with corresponding initial $87Sr/86Sr$ (i) and $εNdt$ values of 0.7057–0.7079 and from −5.8 to −0.6, respectively (Figure 6(a)). They also have $TDM2$ ages of 1.1–0.8 Ga. These characteristics are again similar to the Eocene–Oligocene adakite-like granites in the Qiangtang Block and the eastern syntaxis of Tibet.

These samples contain $SiO2=56.36–63.86wt.%$⁠, $K2O=5.40–7.69wt.%$⁠, $Al2O3=13.55–16.65wt.%$⁠, $FeOt=4.15–6.32wt.%$⁠, $MgO=0.41–5.31wt.%$⁠, and $TiO2=0.45–0.89wt.%$⁠. They have $K2O+Na2O$ values of 9.46–11.47 wt.% and $K2O/Na2O$ ratios of 1.29–2.67 (Figure 3(a)). They contain lower SiO₂ concentrations than the Group 1 samples and have lower $A/CNK$ (0.57–0.93), $Sr/Y$ (36.72–75.74), and $La/Yb$ (12.97–66.59) values but contain higher concentrations of K₂O, Y (>20 ppm), and Yb (generally >2 ppm; Figures 3 and 4). These syenites are enriched in LILEs and light rare earth elements (LREEs) and have $La/Ybn$ values of 9.30–45.76, $Gd/Ybn$ values of 2.19–4.20, $Eu/Eu∗$ values of 0.77–0.99, and positive Sr anomalies but are depleted in Nb, Ta, and Ti (Figures 5(c) and 5(d)). The analysis of six representative samples yields initial $87Sr/86Sr$ (i) ratios that range from 0.7065 to 0.7075 and $εNdt$ values that range from −8.4 to −2.4 with $TDM2$ ages of 1.3–0.9 Ga (Figure 6(a)). These characteristics are identical to those of contemporaneous syenites and mafic rocks within the eastern Tibetan Plateau, but contrast with the compositions of the Group 1 samples.

The Group 3 samples contain 62.27–64.59 wt.% SiO₂, 5.37–6.11 wt.% K₂O, 15.80–16.88 wt.% Al₂O₃, 0.47–0.64 wt.% TiO₂, and 9.63–10.34 wt.% $K2O+Na2O$⁠, with $K2O/Na2O$ ratios of 1.2–1.4. They contain 3.52–4.78 wt.% $FeOt$ and 1.18–2.34 wt.% MgO and are classified as shoshonitic syenites and quartz monzonites (Figure 3). These granitoids have $A/CNK$ values of 0.87–0.99 and $A/NK$ values of 1.20–1.24. They have relatively high $Sr/Y$ ratios (74.9–92.9; Figure 4(a)) and record LREE and heavy REE (HREE) fractionation with $La/Ybn$ values of 53.92–89.15 and $Gd/Ybn$ values of 4.00–5.91. These granitoids are enriched in LILEs but depleted in HFSEs and have insignificant Sr anomalies (Figures 5(e) and 5(f)). They have initial $87Sr/86Sr$ (i), $εNdt$⁠, and $TDM2$ values from 0.7092 to 0.7098, from −12.6 to −10.9, and from 1.6 to 1.5 Ga, respectively (Figure 6(a)). These samples have the lowest $εNdt$ values of any of these groups and are compositionally similar to the Yao’an lamprophyres within the eastern syntaxis of Tibet [16, 23, 25].

The samples have low loss on ignition (LOI) values (<3.0 wt.%) that do not correlate with the concentrations of immobile elements such as Rb, Sr, and Ba [60]. This, combined with the fresh appearance of these samples in hand specimen and during optical microscopy and the poor correlation present between $Nb/La$ values and SiO₂ contents, suggests that the sampled units have undergone negligible postmagmatic alteration (Figure 2; [61]).

The Group 1 samples are geochemically similar to high-SiO₂ adakitic rocks. Adakitic magmas were originally thought to form as a result of partial melting of subducted oceanic slab material, a process that involved a garnet-dominant residual phase that either lacked or contained small amounts of plagioclase (e.g., [59]). However, more recent research has determined that adakitic magmas can form as a result of the fractional crystallization of mafic magmas, a process that involves amphibole ± titanite plagioclase segregation (e.g., [62]), or by the low-pressure partial melting of suitable source rocks [63, 64]. Adakitic magmas have also been identified in continental settings and are thought to be the result of partial melting of delaminated (e.g., [65, 66]), subducted (e.g., [67]), or thickened regions of the lower continental crust (e.g., [68, 69]).

Group 1 samples yield zircon ages of 37.5–33.0 Ma that indicates the formation after the onset of the India–Asia collision. This suggests that oceanic or continental subduction did not occur along the Ailaoshan zone during the Eocene (e.g., [1, 4, 31]). Their Sr–Nd isotopic compositions are distinct from those expected for oceanic slab-derived adakites, which typically have positive $εNdt$ values (+6 to +10 [70]). It is also likely that the Ailaoshan zone and the western margin of the Yangtze Block were in an intracontinental setting at 38–33 Ma, suggesting that these adakites cannot have been derived from subducted oceanic slab or lower continental crustal sources.

Plotting the adakite-like granites in Harker diagrams (Figure S3) suggests that these intrusions have undergone differentiation. This generated lower concentrations of Al₂O₃ (Figure S3(a)), CaO (Figure S3(b)), and Na₂O (Figure S3(c)) with increasing SiO₂, which indicates fractional crystallization of plagioclase. In addition, the presence of negative correlations between $FeOt$ (Figure S3(d)) and MgO (Figure S3(e)) and SiO₂ most likely reflects the separation of ferromagnesian minerals (biotite ± hornblende) during the evolution of the magmas that formed these granites, which is consistent with the trends for the samples evident in Rb and Ba vs. Sr diagrams (Figures 7(c) and 7(d)). Decreases in TiO₂ (Figure S3(f)) and P₂O₅ (Figure S3(g)) concentrations with increasing SiO₂ also reflect the fractionation of Ti-bearing phases (e.g., ilmenite and titanite) and apatite. The adakite-like granites in the study area also have Sr–Nd isotopic compositions that are similar to contemporaneous mafic rocks in the eastern Tibetan Plateau (Figure 6; [16, 71, 72]). However, it is unlikely that the magmas that formed these granites formed as a result of the fractional crystallization of mafic magmas. For one, the adakite-like granites have similar K₂O contents to those of the contemporaneous mafic rocks (Figure 3(b)), which is contrary to the differences expected when comparing genetically associated magmas where more evolved magmas would typically contain higher concentrations of incompatible elements such as K. High $Sr/Y$ granitoids formed as a result of fractional crystallization also generally have positive correlations between $Sr/Y$ and $Dy/Yb$ and SiO₂ contents, and La concentrations that negatively correlate with SiO₂ [73]. This is not the case for the adakite-like granites in the study area (Figures 7(b), S3(h), and S3(i)). Finally, poor correlations between the major element concentrations (relative to SiO₂) of the Group 1 adakite-like granites and the contemporaneous mafic rocks (Figure S3) strongly suggest that the former cannot have formed as a result of the fractional crystallization of the latter

High-$Sr/Y$ signature of some adakitic granitoids is thought to have been inherited from their source regions as a result of low-pressure partial melting (<40 km; e.g., [63, 64]). This possibility can be examined using $La/Yb$ ratios that can be used as crustal thickness proxies for low magnesium intermediate calc-alkaline rocks (55–68 wt.% SiO₂). This means that the high $La/Yb$ ratios of the adakite-like granites in the study area reflect the formation within a region of the crust thicker than 50 km, precluding this as a petrogenetic model and indicating that the source for these adakitic granitoids was located at significant depth [74].

The positive correlation between the $La/Sm$ ratios and the La concentrations of the Group 1 samples suggests they are linked by partial melting (Figure S3(j)). The formation of adakitic melts by the partial melting of a delaminated region of the lower continental crust would generate magmas containing low concentrations of SiO₂, MgO, and TiO₂. These magmas would also have compatible element abundance (e.g., Ni and Cr) that would increase as a result of interaction with mantle peridotites located above the delaminated and melting region of the lower crust during the ascent of the adakitic magmas [67]. Group 1 samples have high SiO₂ contents but low MgO, TiO₂, Ni, and Cr contents, contrasting with adakites derived from delaminated regions of the lower crust but indicating they are compositionally similar to adakites derived from thickened regions of the lower crust (Figure S3(e), (f), (k), and (l)). Adakitic magmas can be generated by the melting of mafic rocks at high pressure within thickened regions of the crust with residual garnet ± rutile (e.g., [75, 76]). The Group 1 samples have high $Sr/Y$ and $La/Yb$ ratios and low HREE concentrations. This, combined with the presence of positive correlations between SiO₂ and $La/Yb$ and $Dy/Yb$ (Figures 4, 7(a), and 7(b)), suggests that the magmas that formed these adakites were derived from a source containing residual garnet (e.g., [77, 78]). The insignificant Eu and Sr anomalies also indicate that this residue was free of plagioclase as both Eu and Sr are released during high-pressure melting as a result of the instability of plagioclase under these conditions (Figure 5(a); [75]). The Neoproterozoic mafic lower crust of the western Yangtze Block also has negative $εNd$ (35 Ma) values (−5.3 to −3.9; [79]) that overlap with the values for the Group 1 adakite-like granites. This suggests that the Group 1 adakite-like granites in the eastern Tibetan Plateau were formed from magmas generated by partial melting of a thickened region of the lower crust under the western margin of the Yangtze Block. These adakitic magmas underwent variable amounts of fractional crystallization prior to their emplacement.

The variable $Rb/Ba$ and $Rb/Sr$ ratios of the Group 1 adakite-like granites are similar to those expected for basalt-derived melts, although a few of these samples plot close to greywacke compositions along mixing trends between metabasalt and metapelite end-members (Figure 8), suggesting the source region for these rocks was most likely dominated by metabasalt material. However, the variable isotopic compositions of these samples suggest that the formation of the Group 1 granites most likely involved mantle-derived melts. This is consistent with the following observations. (1) The variable SiO₂ contents (63.31–73.62 wt.%) and A/CNK ratios (0.83–1.16) of the adakitic granites reflect derivation from a hybrid source region. (2) Some Group 1 samples contain elevated MgO, TiO₂, Ni, and Cr contents, reflecting the involvement of enriched mantle melts in their petrogenesis. (3) The Sr–Nd isotopic compositions of the samples plot along a mixing curve defined by Neoproterozoic lower crustal material and enriched mantle-derived melts in the westernYangtze Block (Figure 6). (4) These granites have zircon $εHft$ and $δ18O$ values that range from −4.3 to +5.2 and from 6.78‰ to 7.36‰, respectively, plotting between values expected for crust- and mantle-derived magmas (Figure 6; e.g., [80]). This evidence in support of a binary mixing model for these adakite-like granites is supported by the presence of deep-derived mafic enclaves in these units [24]. All of this suggests that the Group 1 adakite-like granites in the eastern Tibetan Plateau formed from magmas derived from a hybrid source region dominated by overthickened lower crustal material with lesser amounts of mantle-derived melt.

Syenites are thought to form as a result of partial melting of metasomatized mantle wedge or basement material in the lower continental crust (e.g., [81]), or by the fractional crystallization of mantle-derived basaltic magmas (e.g., [82]). The Group 2 syenites have lower SiO₂ and higher MgO, Cr, and Ni contents than the Group 1 adakite-like granites. Their negative $εNdt$⁠, variable $εHft$⁠, and high $δ18O$ values are also indicative of derivation from an enriched region of the lithospheric mantle (Figures 6; [58, 83, 84]), rather than from the lower continental crust. The syenites are enriched in La and have nearly constant $La/Sm$ ratios, suggesting their compositions reflect fractional crystallization processes (Figure S3(j)). The Group 2 syenites and contemporaneous mafic rocks define linked trends in the Harker diagrams (Figure 3(b) and S3), again consistent with a petrogenesis involving fractional crystallization. The syenites have similar Sr–Nd–Hf isotopic compositions to these contemporaneous mafic rocks, which are thought to have formed from magmas derived from the partial melting of enriched lithospheric mantle (Figures 6; e.g., [16, 26, 71, 72, 85]). This suggests that the Group 2 syenites might have formed from magmas derived from mafic melts sourced from a mantle wedge that was previously enriched by interaction with subduction-related fluids and/or melts [16, 28, 71, 72, 86].

The Group 2 syenites have SiO₂ contents that negatively correlate with CaO (Figure S3(b)), FeO_t (Figure S3(d)), and MgO (Figure S3(e)) contents. Combined with their moderately elevated $La/Yb$ ratios (Figure 7(a)), this indicates differentiation of olivine and clinopyroxene. The syenites also define trends in Rb vs. Sr and Ba vs. Sr diagrams that are indicative of amphibole crystallization (Figures 7(c) and 7(d)). Their TiO₂ and SiO₂ contents negatively correlate, suggesting that the magmas that formed these units underwent titanite fractionation. The presence of slightly negative Eu anomalies (Figure 5(c)) and positive correlations of the Al₂O₃ (Figure S3(a)) and Na₂O (Figure S3(c)) contents with SiO₂ also indicate they underwent insignificant plagioclase crystallization. All of this suggests that the magmas that formed the Group 2 syenites in the eastern Tibetan Plateau formed as a result of fractional crystallization of mafic minerals (olivine, clinopyroxene, and amphibole) from contemporaneous metasomatized mantle-derived magmas.

The Group 3 granitoids record significant fractional crystallization that is evidenced by their generally constant $La/Sm$ ratios relative to increasing concentrations of La (Figure S3(j)). Their CaO and MgO contents negatively correlate with increasing SiO₂ contents, which is indicative of fractional crystallization of amphibole (Figure S3(b) and S3(e)). The negative correlation between SiO₂ and TiO₂ (Figure S3(f)) and the negative Ti and Ba anomalies (Figure 5(f)) suggest fractional crystallization of Ti-oxides and biotite, which is consistent with the trends defined by these samples in Figures 7(c) and 7(d). They also have Sr–Nd isotopic compositions that are similar to the contemporaneous Yao’an lamprophyres, suggesting that the granitoids were formed from magmas derived from a source containing ancient enriched mantle material (Figure 6; [16]). However, the Group 3 samples have $La/Sm$⁠, $La/Yb$⁠, and $Zr/Nb$ ratios that contrast with those of the Yao’an lamprophyres and define distinct trends compared with these lamprophyres in diagrams of SiO₂ vs. CaO (Figure S3(b)), FeO_t (Figure S3(d)), MgO (Figure S3(e)), and TiO₂ (Figure S3(f)). This indicates that the Group 3 samples cannot have formed by the fractional crystallization of the Yao’an lamprophyre melts. Their distinctive Sr–Nd isotopic compositions, low $εHft$ values, and extremely high $La/Yb$ ratios relative to Groups 1 and 2 [23, 25] suggest that the Group 3 granitoids also were formed from a distinct source compared with the Groups 1 and 2 samples. The extremely negative $εNdt$ and $εHft$ values of the Group 3 granitoids are indicative of derivation from ancient crustal material (e.g., [80]). The weakly negative Eu anomalies in these samples are also indicative of a lower crustal source region, as granitoids derived from upper and middle continental crustal sources typically have very well-developed negative Eu anomalies [87]. This is consistent with the following observations. (1) The relatively low $Rb/Sr$ and $Rb/Ba$ ratios of the Group 3 samples are similar to those of basalt-derived melts, suggesting they were derived from a metamafic source (Figure 8). (2) The high $La/Yb$ and $Dy/Yb$ ratios and low HREE concentrations are indicative of a source region containing garnet. (3) The $δ18O$ values (6.31‰–6.82‰) are also generally lower than those of supracrustal materials (e.g., [84]). The Sr–Nd isotopic compositions of the Group 3 samples are also compositionally similar to Neoproterozoic lower crust material in the western margin of the Yangtze Block [88, 89]. They also have the same range of two-stage Hf model ages as granitoids and diorites in the lower crust of the western Yangtze Block [89–91]. All of this suggests that the Group 3 granitoids of the eastern Tibetan Plateau formed as a result of fractional crystallization of magmas derived from the lower crust of the Yangtze Block, which in turn was derived from an ancient enriched mantle source.

The results of the dating analyses indicate that the Group 1 adakite-like granites and Group 2 syenites were formed at 37.6–33.0 Ma and 36.8–34.3 Ma, respectively. The Group 3 low-$εNdt$ granitoids within the Yongsheng and Yao’an areas of the interior of the Yangtze Block yield a crystallization age of ca. 33 Ma, which is slightly younger than the Groups 1 and 2 magmatism in the eastern Tibetan Plateau (Figure 2). Previously published researches indicate that Eocene–Oligocene igneous rocks are not only present in the northwestern part of Yunnan Province but also in the Qiangtang Block and its southeastward extension (e.g., [1]). These rocks include 46–32 Ma shoshonitic rhyolite, andesite, trachyte, and dacite of the central Qiangtang Block (e.g., in the Suyingdi, Bandaohu, Duogecuoren, Zhentouya, and Fenghuoshan areas; [67, 92, 93]). Zircon U–Pb dating of basalt, ivernite, syenite, and porphyritic monzonite in the Nangqian area indicate that these potassic alkalic rocks were formed at 40–35 Ma (e.g., [94]). Monzonitic porphyritic granite and porphyritic diorite in the Xiariduo area of the northern Yulong porphyry copper belt yielded crystallization ages of ca. 41 Ma [95], whereas porphyritic granitoids in the southern Yulong area yield slightly younger U–Pb ages (39–38 Ma; e.g., [96]). These data suggest that potassic igneous rocks in the central Tibetan Plateau were formed either contemporaneously with or earlier than 38–32 Ma potassic igneous rocks in the eastern Tibetan Plateau (e.g., [16, 19, 20, 27–29, 51, 71, 85, 97, 98]). This also indicates that the Cenozoic magmatism is young to the southeast (i.e., from the central to the eastern Tibetan Plateau). The potassic magmatism in the central and eastern Tibetan Plateau was indirectly triggered by the continental collision of India with Asia. All of this indicates that the magmatic response of this collisional event might have commenced within the central plateau before propagating to the eastern margin, resulting in the generation of an Eocene to early Oligocene potassic igneous belt extending from the central Tibetan Plateau to the eastern part of the region.

As discussed above, the Group 1 adakite-like granites and Group 2 syenites were derived from overthickened lower crustal and enriched lithospheric mantle sources, respectively. The Group 3 low-$εNdt$ granitoids were formed from magmas generated by the partial melting of ancient lower crustal material. The adakitic rocks in the central Qiangtang Block and its southeastward extension are thought to have been formed from magmas derived from heterogeneous lower crustal material (e.g., [67, 93, 96]). The geochemical similarities of the adakitic rocks in the Qiangtang Block and the eastern Tibetan Plateau and the temporal and spatial variations in potassic igneous rocks along the Jinshajiang River (e.g., [16, 67, 70, 72]) suggest that all of these magmas formed as a result of similar processes. Plotting the $Sr/Y$⁠, $εNdt$⁠, and $εHft$ values presented in this study with previously published data for the central and eastern Tibetan Plateau indicates regular spatial variations in the compositions of the adakite-like granites (Figure 9). Their $Sr/Y$ ratios decrease gradually from west to east in the Qiangtang Block and from northwest to southeast in the eastern Tibetan Plateau, with the highest $Sr/Y$ values present in the Duogecuoren and Lijiang areas (Figure 9(a)). The $Sr/Y$ ratios of adakite-like granites derived from thickened lower crust regions can be used as a proxy for crustal thicknesses, suggesting the data shown in Figure 9(a) indicates that the overthickened crust of central Tibet and the eastern Tibetan Plateau thins to the southeast [75]. In addition, their decreasing $εNdt$ and $εHft$ values (Figures 9(b) and 9(c)) indicate that these units record the increasing participation of ancient lower crustal or enriched lithospheric mantle sources from northwest to southeast in the eastern syntaxis of the Tibetan Plateau and the western margin of the Yangtze Block.

Eocene–Oligocene magmatism in the central Qiangtang Block and the eastern Tibetan Plateau is thought to be a response to the India–Asia collision (e.g., [1, 7, 8, 92]). The timing of formation and petrogenesis of Cenozoic potassic igneous rocks in the eastern Tibetan Plateau has led to block extrusion, thickened lithospheric delamination, and lower-mid crustal flow models for the growth of the Tibetan Plateau. The block extrusion model assumes that rigid crustal blocks were extruded along major tectonic boundaries represented by the ASRR and the Gaoligong shear zone (e.g., [7, 13, 99, 100]). However, the potassic igneous rocks in these areas are not spatially restricted to these shear zones or adjacent areas. In addition, sinistral shearing along the ASRR shear zone occurred mainly during 32–20 Ma or even as late as 17 Ma, which is significantly younger than the timing of formation (38–32 Ma) of the potassic igneous rocks (e.g., [49, 97, 99–103]). This indicates that any block extrusion processes cannot be responsible for the formation of the potassic igneous rocks in this area.

A model involving the delamination of a thickened region of the lithosphere and subsequent asthenospheric upwelling has been proposed for the generation of the potassic igneous rocks in the eastern Tibetan Plateau (e.g., [24, 27, 28]). However, the 38–32 Ma potassic igneous rocks in this area are low volume and have a scattered distribution in the eastern Tibetan Plateau, contrasting with the granitoids in the Gangdese area that formed as a result of lithospheric foundering [104]. In addition, very few contemporaneous faults or rift basins have been identified in this region, which is inconsistent with the large-scale extension predicted by a delamination model.

The lower-mid crustal flow model suggests that low viscosity materials in the middle to lower crust flowed outward from the plateau interior as a response to topographic loading and gravity (e.g., [9, 12, 15]). The synchronous adakitic magmatism in the Qiangtang Block (e.g., in the Suyingdi, Duogecuoren, and Nangqian areas) and the eastern Tibetan Plateau formed intrusions with similar geochemical and Sr–Nd isotopic compositions (Figures 3–8; [67, 93]). The lower crust beneath the Qiangtang Block is thought to contain a heterogeneous mix of metaigneous and metasedimentary rocks (e.g., [93]). Combining this with the spatial variations present in the Cenozoic igneous rocks and the fact that this magmatism is young to the southeast (i.e., from the central to the eastern Tibetan Plateau) indicates that this area contains thickened lower crustal material [93]. However, the igneous activity in the central and eastern Tibetan Plateau occurred at 45–32 Ma, which is substantially earlier than the proposed onset of lower crustal flow (e.g., [15, 105, 106]). In addition, the lower crustal flow model remains a controversial model for the early growth of the eastern Tibetan Plateau (e.g., [15, 107, 108]).

Numerous geological and geophysical observations indicate that lithospheric shortening and crustal thickening in the eastern Tibetan Plateau are the result of continuous continental convergence since the Paleogene (e.g., [11, 109]). Numerous Eocene to early Oligocene NW–NE- to NNW–SSE-trending anticline–syncline pairs are present in the Gaoligong, Diancangshan, Xuelongshan, Lanping–Simao, and Chuxiong areas (Figure 1(c); [103, 110]). The NNW–SSE-trending Ludian–Zhonghejiang fold and thrust belt between Xuelong and Diancang Shan might also have formed at 50–39 Ma, as evidenced by the timing of deposition of the Baoxiangsi sedimentary sequence in the Jianchuan basin (e.g., [111]). These structural elements indicate that the eastern Tibetan Plateau underwent nearly E–W- or NE–SW-orientated compression between the Eocene and early Oligocene. Combining this with the 32–20Ma sinistral movement along the ASRR shear zone suggests that the 38–32 Ma potassic igneous rocks formed contemporaneously with the transition from compression to sinistral strike-slip movements and an associated relaxation of stress.

Synthesizing the information outlined above allows the generation of an Eocene–Oligocene thickening-collapse tectonic model for the generation of Cenozoic potassic igneous rocks and the associated growth of the eastern Tibetan Plateau. The 55–38 Ma shortening and thickening of the lithosphere beneath the eastern Tibetan Plateau involved NW–SE-orientated compression in response to the India–Asia collisional event. This caused the relatively cold lithospheric mantle to sink into the asthenosphere, rheologically weakening the middle–lower crust. The deformation at this time was dominated by the formation of NW–SE to NNW–SSE anticline–syncline pairs and fold and thrust belts that accommodated the regional strain (Figure 10(a)). Stress relaxation and mantle convective thinning at 38–32 Ma led to the weakening and collapse of the overthickened crust as a result of gravitational disequilibrium. This resulted in the simultaneous melting of enriched lithospheric mantle and thickened lower crustal material, generating the magmas that formed the potassic igneous rocks (Figure 10(b)). Sinistral ductile shearing along the ASRR shear zone occurred after ca. 32 Ma and was accompanied by the exhumation of metamorphic complexes (Figure 10(c)). All of this means that the potassic igneous rocks record the transition from thickened lithosphere during the early stages of the India–Asia collision to lithospheric extension in response to orogenic collapse in the eastern Tibetan Plateau.

New geochronological and geochemical data for Eocene potassic igneous rocks in the eastern Tibetan Plateau presented in this study outline the following conclusions. (1) The potassic rocks in this region are dominated by adakite-like granites, syenites, and low-$εNdt$ granitoids that were formed at 37.6–32.9 Ma. (2) The adakite-like granites, syenites, and low-$εNdt$ granitoids were derived from overthickened lower crust containing mantle-derived melts, enriched mantle, and ancient lower crustal sources, respectively. (3) The eastern Tibetan Plateau most likely underwent tectonic thickening (>38 Ma) and subsequent collapse (38–32 Ma) as a result of the India–Asia collision.

Data used to support the result of this study can be found in this manuscript text, the supplemental material file, and previous publications discussed in the text

The authors declare that they have no conflicts of interest.

This work was jointly supported by the Second Tibetan Plateau Scientific Expedition and Research Program (2019QZKK0703), National Natural Science Foundation of China (41830211 and U1701641), and Guangdong Basic and Applied Basic Research Foundation (2018B030312007, 2019B1515120019).