The processes by which lamprophyres and associated carbonatites are generated remain subject to debate. The Wase Basin on the SE Tibetan Plateau contains trachytes, rhyolites, and minor carbonatites that were emplaced at 37–36 Ma. Coeval lamprophyre dikes are widespread in the adjacent regions. Geochemically and petrographically, both the extrusive trachytes and dikes can be classified as lamprophyre. The numerous millimete r-sized SiO2-rich and calcite-rich ocelli that occur within the trachytes and dikes are solidified pseudomorphs of felsic and calcic melt drops, respectively. These ocelli combined with inherited granitic zircons suggest magma mixing between the calcic melt or felsic melt and basaltic melt. Petrographic evidence, mineral compositions, and zircon textures and U-Pb ages (827–682 Ma) suggest that the felsic melt and the calcic melt were generated by rapid H2O-saturated partial melting of Neoproterozoic granite and limestone, respectively, at >800 °C and <2 kbar, according to available experimental data. These melts were separated from each other and were stored at upper crustal depths, forming a zone of transient magma lenses or parcels. Melts mixed when an ascending basaltic magma intersected this zone. Reaction between the basaltic melt and the calcic melt occurred when they mixed, which formed clinopyroxene of predominantly diop side-hedenbergite solid solution. The observed heterogeneity of the upper crust combined with the variable degrees of magma mixing account for the great chemical diversity of the lamprophyres of SE Tibet. This model sheds new light on the petrogenesis of other lamprophyre-carbonatite associat ions elsewhere.

Lamprophyres have been identified throughout the geological record, from ca. 2.7 Ga to the present day and in a wide range of tectonic regimes (Rock, 1987). These volatile-rich rocks are characterized by extreme enrichment in K, Rb, Ba, Cs, U, and Th (Wyman and Kerrich, 1993); a wide range of 87Sr/86Sr ratios; and a restricted range of 143Nd/144Nd ratios (Lustrino et al., 2016, and references therein), which are believed to reflect their origins by low-degree partial melting of ancient metasomatised mantle at depths of 50–150 km (Jaques et al., 1986; Rock, 1987, 1991). Given their origins, these rocks have long been thought of as a unique window into mantle-crust interactions and the evolution of the deep mantle throughout Earth’s history, despite their small volumes (e.g., Menzies and Halliday, 1988).

The petrogenesis of lamprophyres is, however, more poorly understood than that of any other type of igneous rock (Prelević et al., 2004), especially given the great diversity in their bulk compositions and mineral assemblages (Bell et al., 1998; Gill, 2010; Yoder, 1986). Each lamprophyre appears to be unique, and no single petrogenetic model can be applied universally (Mitchell and Bergman, 1991). The lack of understanding of lamprophyres is also manifested in the difficulties in defining, classifying, and naming these rocks (Le Maitre et al., 2002; Lustrino et al., 2016). Surveying the literature reveals the following two characteristics:

  1. Lamprophyres are commonly a volumetrically minor member of highly lithologically diverse magmatic associations in magmatic provinces with areas of >10,000 km2, where each igneous group occurs in several tectonic units with different lithospheric structures and compositions. For example, the Early Cretaceous lamprophyres in S to SE Brazil (Florisbal et al., 2018) and Namibia (Le Roex and Lanyon, 1998; Owen-Smith et al., 2017) are cognate with numerous other magmatic rocks related to continental rifting and the opening of the Atlantic Ocean. Late Cretaceous lamprophyres in India are part of the last stage of magmatism in the Deccan large igneous province and are associated with scattered alkaline complexes and lava flows (Chandra et al., 2019; Pandey et al., 2019; Sahoo et al., 2020; Vanderkluysen et al., 2011). This spatial and temporal distribution suggests that lamprophyres and other associated rocks were generated at mantle depths by lithospheric-scale processes such as a mantle plume (e.g., those in Siberia; Doroshkevich et al., 2019) or the break-up of a supercontinent (e.g., those in Namibia and Brazil; Florisbal et al., 2018). Thus, it is difficult to disentangle the origin of lamprophyres from the origin of other associated rocks.

  2. Local equilibrium, both chemical and textural, should be readily archived in a large magmatic system; however, disequilibrium textures and microstructures as well as isotopic disequilibrium are widespread in lamprophyres, which indicates incomplete reactions, even where a volatile phase is present (Rezeau et al., 2018; Streck, 2008; Yoder, 1986). These textures are generally considered indicative of open-system processes (e.g., Yoder, 1986), which are not seen in other mantle-derived melts of magmatic associations, including mid-ocean-ridge basalt (MORB) and oceanic island basalt (OIB).

Reconciling these two apparently inconsistent traits may provide new insights into the petrogenesis of lamprophyres and crust-mantle interactions.

We recently identified an Eocene basin—the Wase Basin—in the west Yunnan and west Sichuan lamprophyre provinces (Guo et al., 2005), SE Tibet. The basin fill contains a volcanic succession of chemically and mineralogically diverse lamprophyres with minor carbonatite. Numerous coeval lamprophyre dikes are also emplaced in adjacent regions (Zhu et al., 1992). These rocks contain abundant disequilibrium textures, some of which have not been reported previously; this paper focuses on these textures. Our objectives include: (1) determina tion of the emplacement age of the Wase volcanic rocks and dikes, (2) detailed descriptions and explanations of the disequilibrium textures, (3) reconstr uction of the magmatic evolution of the lamprophyres using the disequilibrium microtextures, and (4) creation of a petrogenetic model for the lamprophyres and calcic rocks as well as for other alkaline rocks including the potassic high–Ba-Sr granitoids (Wang et al., 2018) of SE Tibet. This model may be useful for interpreting similar disequilibrium textures and chemical diversity of lamprophyre-carbonatite associatio ns worldwide.

The northward indentation of Eurasia by India generated an E-W–trending orthogonal collision belt (the Tibet Himalaya) and an N-S–trending oblique collision belt (the southeastern Tibetan Plateau) along the frontal and eastern edges of the Indian indenter, respectively (Fig. 1A; Burchfiel and Chen, 2012; Mattauer et al., 1999; Molnar and Stock, 2009). A comprehensive compilation of the published geological, geochronological, and geophysical data by Yang et al. (2021) enabled the identification of an ENE-WSW–trending, ~250-km-wide, ~550-km-long lithospheric transitional zone between the orthogonal and oblique collision belts. A well-known geophysical lithospheric boundary—the latitude 26°N line (Lev et al., 2006; Sol et al., 2007)—forms its southern boundary. Abundant late Eocene mantle-derived plutons (Chen et al., 2017; Ding et al., 2016; Lu et al., 2013; Xu et al., 2007; Zhao et al., 2016) and a few Eocene basins, including the Jianchuan (Gourbet et al., 2017; Yang et al., 2014b), Yongping (Yang et al., 2021), and Wase (this study) Basins, mark this transitional zone, in which lamprophyre dikes are widespread (Guo et al., 2005). Tomographic imaging using P-wave arrival times (e.g., Li et al., 2008) has revealed a prominent ENE-WSW–trending high-velocity zone at depths of 100–300 km in the mantle beneath this lithospheric transitional zone. Similarly, the P-wave velocity of the middle crust in this zone is higher than that to the south (Lin et al., 2019).

The ENE-WSW–trending transitional zone is developed upon the NNW-SSE–trending Palaeo-Tethyan orogenic belt, which records the eastward (in present-day coordinates) subduction of the Palaeo-Tethyan oceanic crust beneath Indochina, an event that formed the large Permian–Triassic Jomda-Weixi-Yunxian continental arc (Xin et al., 2018; Yang et al., 2014a). The Permian–Triassic igneous rocks roughly define the eastern boundary of the Jomda-Weixi-Yunxian arc, which is coincident with the eastern marginal faults of the Diancangshan and Ailaoshan complexes (Fig. 1). The ENE-WSW–trending transitional zone cuts across the Jomda-Weixi-Yunxian arc and extends eastward into the interior of the Yangtze block (Guo et al., 2005; Lu et al., 2013). The eastern and western terminations of this zone are the Yimen and Sagaing faults, respectively (Fig. 1A; Yang et al., 2021).

The upper crust of the Yangtze block consists of two parts: Neoproterozoic basement and upper Neoproterozoic–Palaeozoic sediment ary cover (Yunnan Bureau of Geology and Mineral Resources, 1990). The basement of the Yangtze block consists of Neoproterozoic (850–650 Ma; e.g., Geng et al., 2008, and references therein) magmatic rocks that intrude sedimentary rocks including limestone, sandstone, and mudstone (Yunnan Bureau of Geology and Mineral Resources, 1990).

The recently identified Wase Basin is ellipsoidal and elongated E-W, with long and short axes of 10 km and 2 km, respectively. This basin is located in the eastern central part of the transitional zone, a few kilometers east of the Jomda-Weixi-Yunxian arc (the Diancangshan complex; Lin et al., 2012; Liu et al., 2013; Xu et al., 2021; Fig. 1). The base of the Wase Basin is the upper Palaeozoic platform-facies sedimentary succession of the Yangtze block, which is intercalated with late Permian Emeishan OIB-like basalts (First Geological Survey of Yunnan Province, 1973).

3.1. Fieldwork and Sampling

Our field observations in an abandoned quarry adjacent to the Dali-Lijiang highway revealed an ~17-m-thick volcanic succession in the Wase Basin (location DL007; Fig. 1). This succession (Fig. 2A) consists of, in ascending stratigraphic order: (1) an ~2.5-m-thick trachyte (the lower trachyte), (2) an ~0.5-m-thick massive rhyolite (the lower rhyolite; Fig. 2B), (3) another ~2.5-m-thick trachyte (the middle trachyte), (4) an ~1.5-m-thick massive rhyolite (the middle rhyolite), (5) an ~5-m-thick trachyte (the upper trachyte), and (6) an ~5-m-thick massive rhyolite (the upper rhyolite). This volcanic succession is overlain by poorly consolidated terrigenous sediments. High-K granite intrudes the succession in some places. The thicknesses of the rhyolitic units are highly variable along their strike. Carbonatite is emplaced alongside the rhyolitic units in some places (Fig. 2C). We collected seven samples from the ~5-m-thick upper trachyte, a total of four samples from the lower and middle trachytes (Fig. 2), a total of six samples from the rhyolites, and a total of six samples of carbonatite for petrographic, geochemical, and geochronological analyses. The GPS locations, lithological names, and ages are listed in Table S1 in the Supplemental Material.1

Numerous mafic-ultramafic dikes (Fig. 3A) are emplaced in the regions around the Wase Basin (Fig. 1; Zhu et al., 1992). We conducted fieldwork and sampling in a quarry adjacent to Fengyi Airport (location DL009; Fig. 1). The main rocks in the quarry are Devonian–Carboniferous conglo merate consisting of limestone breccias cemented by fine-grained sands (Fig. 3B). Numerous E-W–trending lamprophyre dikes cut across the conglomerate beds (Fig. 3A). The lamprophyre dikes are porphyritic with large vermiculite crystals. Some of the vermiculite crystals are clumped together to form roughly spherical aggregates (Fig. 3C). The lamprophyre dikes contain basaltic xenoliths with irregular outlines (Fig. 3D), ranging from a few to tens of centimeters in diameter. Seven lamprophyres and five basaltic xenoliths were sampled for petrological, geochemical, and geochronological analyses. The GPS locations, lithologies, and a summary of the analytical results are listed in Table S1.

3.2. Microtextures

At least one thin section was made of each specimen for petrographic study using a polarized light microscope (Leica DM2700P) equipped with a camera adapter (Leica DFC450). Based on detailed microscopic observations, 10 well-polished thin sections were selected for backscattered electron (BSE) imaging and energy dispersive X-ray spectroscopy (EDS) using a scanning electronic microscope (Oxford INCA) at the Institute of Geology, Chinese Academy of Geological Sciences (CAGS), Beijing, China, with an accelerating voltage of 15 kV and a probe current of 20 nA. The EDS results are listed in Table S2.

3.3. Mineral Compositions

Five thin sections were selected for analysis using an electron probe micro-analyzer (EPMA; JEOL JXA-8100) with an accelerating voltage of 15 kV, probe current of 20 nA, and beam diameter of 1 μm at the Institute of Geology, CAGS. Fifty-three natural and synthetic mineral standards (Structure Probe, Inc.) were used for standardisation, and ZAF corrections taking atomic number (Z) effect, absorption (A) effect, and fluorescence excitation (F) effect into account were carried out. The EPMA results are listed in Table S3.

3.4. Bulk-Rock Geochemistry

Twenty-four samples—five basalts, nine trachytes, four rhyolites, and six lamprophyres—were selected for bulk-rock geochemical analyses at the National Research Centre for Geoanalysis, CAGS, using a Rigaku 3080E wavelength-dispersive X-ray fluorescence spectrometer. Analytical errors were <2%. Trace element analyses were conducted at the same center using a Thermo Scientific X-series inductively coupled plasma–mass spectrometer (ICP-MS). The analytical errors on the ICP-MS analyses were <5% for elements with contents of >10 ppm and <8% for elements with contents of <10 ppm, except for the transition metals which had analytical errors of ~10%. The major and trace element contents are listed in Table S4.

3.5. Zircon U-Pb Dating

We separated zircon grains from four samples for cathodoluminescence (CL) imaging and U-Pb dating. Zircon grains were obtained using standard crushing and heavy liquid and magnetic separation techniques. We handpicked 150–200 grains from the >25 μm non-magnetic fraction of each sample. The grains were mounted in epoxy resin for reflected and transmitted light photography, CL imaging, and laser ablation–ICP–MS (LA-ICP-MS) U-Pb dating. CL images were obtained at the Beijing SHRIMP Center, CAGS, using a Hitachi S-3000N scanning electron microscope fitted with a GEOL Gatan Chroma CL detector.

In situ LA-ICP-MS determination of the U-Pb ages of the zircons was conducted using an Agilent 7500x ICP-MS coupled with a GeoLas Pro 193 nm ArF excimer laser at the State Key Laboratory of Ore Deposit Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, China. Helium was used as the carrier gas and was mixed with Ar via a T-connector before entering the ICP-MS. Each analysis incorporated a background acquisition of ~30 s (gas blank) followed by 60 s of data acquisition from the sample. Offline selection and integration of the background and analytical signals, drift corrections, and quantitative calibration for U-Pb dating were performed using ICPMSDataCal (Liu et al., 2008). The 91500, Plešovice, and GJ-1 zircons were selected as primary and secondary standards for U-Pb dating and were analyzed twice every eight sample analyses. Uncertainty in the measured values of the external standard (91500) was propagated through to the reported ages of the samples. Final statistics and age calculations were made using Isoplot/Ex version 4.15 (Ludwig, 2003).

To correct for time-dependent elemental fractionation (the matrix effect), we tried to select as similar a duration as possible for the integration of the background and analytical signals from both the sample and the standards. The results of U-Pb zircon dating, including the standards, are listed in Table S5.

3.6. Bulk-Rock 40Ar/39Ar Dating

One trachyte and one basaltic xenolith sample were selected for bulk-rock 40Ar/39Ar analyses. The samples were crushed into fragments ~1 mm in diameter, and the freshest fragments with no amygdales or phenocrysts were then handpicked under a binocular microscope. The selected fragments (~1.0 g) were further crushed to 80 mesh for 40Ar/39Ar analyses.

The rock powders, along with Fish Canyon Tuff sanidine (Lanphere and Baadsgaard, 1999) and ZBH biotite (a Chinese standard from the Fangshan granite; Wang, 1983) flux monitors, were irradiated in a nuclear reactor belonging to the Research Institute of Atomic Energy of China, Beijing, China, and set in an H8 hole of the nuclear reactor for fast neutron irradiation. The irradiation duration and neutron dose for the analyzed samples were 7.8 h and 1.83 × 1017 n/cm2, respectively. The samples were analyzed after low-temperature (250–300 °C) degassing for 20–30 min. The J factor was estimated using repeated analyses of Fish Canyon Tuff sanidine (27.55 ± 0.08 Ma) and the ZBH biotite standard (133.3 ± 0.24 Ma), using one standard deviation (1σ). The J values for individual samples were determined using a second-order polynomial interpolation.

40Ar/39Ar dating was conducted on the powdered samples, and the age spectra were recorded during step heating. Interference due to nuclear reactions of K and Ca was calculated using co-irradiated pure K2SO4 and CaF2 salts, assuming an 40Ar/39ArK ratio of 0.004782, an 39Ar/37ArCa ratio of 0.00081, and an 36Ar/37ArCa ratio of 0.0002398. Samples were loaded into a Christmas-type double-vacuum furnace in aluminum packets and step heated from 550 to 1500 °C. The gas was purified using Ti and Al-Zr getters. Once cleaned, the gas was introduced into an MM-5400 mass spectrometer (Micromass UK Ltd.) at the 40Ar/39Ar laboratory of the Chinese University of Geosciences, Beijing, China, and allowed to equilibrate for 4–5 min before being subjected to static analyses. The measured isotopic ratios were extrapolated to time zero, normalized to the atmospheric 40Ar/36Ar ratio, and corrected for neutron-induced 40Ar from K and neutron-induced 39Ar and 36Ar from Ca.

Ages and uncertainties were calculated using Isoplot version 3.0 (Ludwig, 2003), which was also used to generate age spectra and isotope correlation plots. We used the decay constant of 40K suggested by Steiger and Jäger (1977). The uncertainty in the plateau age is given as 1σ. Weighted mean plateau ages (WMPAs) are reported as the value where >50% of the 39Ar released in contiguous steps is within 1σ. The analytical results are listed in Table S6.

3.7. Raman Shift Spectra Determination

In order to determine the calcic phases, we conducted Raman shift spectra analyses. The Raman spectra were determined at the Laser Micro-Raman Lab in the Institute of Geology, CAGS, using a confocal Raman microscopy system. This system consists of a Horiba spectrometer LabRAM HR EVOLUTION equipped with an Olympus BX41 polarized light microscope with a 100-fold objective (Olympus MPlan N, Numerical Aperature 0.90, Working Distance 0.21 mm). Samples were excited using a 532 nm (room temperature of 23 ± 1 °C) frequency-doubled Nd:YAG laser beam with 100 mW beam power and 1 µm beam diameter. The scattered light is guided to the spectrograph through a confocal 80 μm pinhole and received by an 600 groves/mm optical grating. A charge-coupled device (CCD) array detector (Syncerity), thermoelectrically cooled to −60 °C, detects the scattered light. The Raman signal was collected in a spectral interval of 100–1400 cm−1 with a spectral resolution of 0.59 cm−1 and calibrated using a silicon wafer with a Raman main shift at 520.7 ± 0.5 cm−1. The acquisition of the Raman spectra comprised two accumulations of 1 s each. The data were processed with LabSpec 6 software. For analytical details, see Zhang et al. (2022).

4.1. Petrography

4.1.1. Upper Trachyte

The upper trachyte is porphyritic (Fig. 4). The phenocrysts are mostly clinopyroxene, set in a groundmass of fine laths of alkali feldspar (30–100 μm in length), biotite, minor Fe-oxides, apatite, and interstitial glass (Figs. 4A and 4B). The phenocrysts exhibit oscillatory zoning, and some show simple twinning (Figs. 4C and 4D). Most phenocrysts contain inclusions of apatite and a silicate glass phase, and some have microfractures that are partly filled with glass (Fig. 4D).

There are two types of vesicles in the upper trachyte. The larger vesicles are mostly >100 μm in length with angular, faceted outlines. They occur only along the faces of clinopyroxene phenocrysts as shadows (Fig. 4B). The other type of vesicle is small, with most having diameters of <5 μm. They are found pervasively throughout the glass and form small bulb pits on the surface of the thin sections (Figs. 4B and 4D).

4.1.2. Middle Trachyte

The middle trachyte samples contain small alkali feldspar laths (~70 vol%) and biotite crystals (~10 vol%) with lengths of <0.2 mm. Gradational compositional zoning is well preserved in the feldspar laths. These microlites are randomly oriented and interlocked with the interstitials filled with silicate glass. Small Fe-oxide crystals (<20 μm) with irregular outlines are pervasive. These four phases form a groundmass that hosts a few euhedral clinopyroxene phenocrysts (0.5–1.0 mm in diameter) and numerous millimeter-sized ocelli with highly irregular outlines (Fig. 5A). The clinopyroxene phenocrysts show oscillatory zoning, simple twinning, and internal microfractures.

Two types of ocelli occur in the middle trachyte (Fig. 5A). The less common type consists of a groundmass of plagioclase that hosts vermicular clinopyroxene, calcite, quartz, and glass and a few F-bearing apatite needles and Fe oxides (Fig. 5B). The plagioclase exhibits patchy compositional zoning, which is apparent as patches of variable brightness in BSE images.

The more common ocelli have irregular outlines and are composed predominantly of non-crystalline calcite (the dense mass of CaCO3 shown in Figs. 5C5D and 6A6C). The polarized light microscopy observations (Fig. 6A), BSE images (Fig. 5C5D and 6B6C), EDS, and microprobe analyses reveal two phases intergrown with or hosted within the non-crystalline calcite groundmass, respectively. A more common phase is silicate glass, which is intergrown with the dense mass of CaCO3 with an irregular, fuzzy boundary (Figs. 5D and 6B). The silicate glass has the same composition as that filling the interstitials among feldspar or biotite laths. Another less common phase is a dense mass of fine quartz needles set in the non-crystalline CaCO3 gro undmass with a sharp boundary. Most dense masses of quartz needles contain a calcite crystal in the center (Fig. 6C). Locally, this core-and-mantle texture is well developed, consisting of a core of single calcite crystal surrounded by quartz (Figs. 6A and 6B). The quartz that surrounds the calcite crystals forms a series of abutting half spherules consisting of a dense mass of fine needles radiating from a single nucleus. We interpret the quartz needles with core-and-mantle texture as representing a filled vesicle with all of the quartz nuclei aligned along the wall of the vesicle (Fig. 6A). These dense masses of quartz needles are smooth at the surface of the thin section and thus are easily differentiated from the silicate glass as well as from the calcic glass with small bulb pits (Figs. 5D and 6B6C). In some calcic ocelli, arcs of Fe-oxide dense masses (curved Fe-oxide septa) outline second-stage spheroids in host ocellus (Figs. 6A and 6B), although no variation in composition between them is identified (Fig. 6B).

4.1.3. Lower Trachyte

The lower trachyte contains a few clinopyroxene phenocrysts (<1 mm in diameter) and abundant felsic ocelli set in a homogeneous groundmass (Figs. 7A and 7B). The groundmass consists mainly of alkali feldspar laths (~70 vol%), biotite (~10 vol%), minor clinopyroxene (~5 vol%), and interstitial non-crystalline quartz (~5 vol%; Figs. 7C and 7D). The ocelli are SiO2 rich, contain a few alkali feldspar microlites, and have sizes of <10 μm to ~5 mm with sharp, faceted outlines. All of the ocelli contain vesicles, with vesicularities of ~5% (Fig. 7C) to >80% (Figs. 7A and 7B). Elongated vesicles are developed along the edges of some ocelli (Fig. 8C). A few large vesicles (<5 mm in length) are partly filled by a mixture of opal and Fe-oxides with pervasive small voids (Fig. 8) or by fine-grained dendritic or acicular opal and Fe-oxides (Fig. 8C). This indicates that the vesicles were previously filled by a fluid phase from which the opal and Fe-oxides precipitated during cooling.

4.1.4. Lamprophyre Dikes

The lamprophyre dikes are porphyritic, with clinopyroxene (0.1–2 mm in diameter; Fig. 9A), biotite (<1 mm in length), and vermiculite phenocrysts (<5 mm in length; Fig. 9B) and calcic ocelli (Fig. 9C) set in a fine-grained groundmass. The clinopyroxene phenocrysts show oscillatory zoning and simple twinning. Each vermiculite phenocryst comprises several flakes separated by flake-parallel fractures. These fractures do not extend into the groundmass, indicating that they formed during crystallization of the vermiculite crystals. Given the roughly spherical vermiculite aggregates (Fig. 3C), we suggest that the vermiculite crystals are a deuteric hydrothermal phase that filled the larger vesicles.

The groundmass consists of randomly oriented 100–200-μm-long and 20–100-μm-wide alkali feldspar laths with interstitial glass. Randomly oriented acicular clinopyroxene microlites (30–100 μm long) occur as inclusions in the feldspar laths as well as in the glass (Figs. 9D and 9E). This relationship indicates that the clinopyroxene microlites crystallized before the feldspar laths. Both the feldspar laths and the acicular clinopyroxene show compositional zoning.

Most of the ocelli are calcic and have similar compositions and textures to those in the middle trachyte (e.g., Figs. 5 and 6) and comprise a dense mass of non-crystalline calcite and silicate glass (Fig. 9B). Some ocelli preserve magmatic flow banding (Fig. 9C). A few non-crystalline calcite microveins cut across the feldspar laths, clinopyroxene microlites, interstitial glass (Fig. 9B), and clinopyroxene phenocrysts (Figs. 9F and 9G). This suggests that the calcic melt had not solidified when the clinopyroxene and feldspar crystallized and that it flowed along microfractures in these crystals. F-bearing apatite crystals commonly grew within the calcite patches or microveins (e.g., Fig. 9F).

There are clinopyroxene-rich zones surrounding some silicate glass patches, which consist mainly of interlocking clinopyroxene laths and minor isolated vermicular alkali feldspar crystals and glass (Figs. 9G and 9H). These clinopyroxene-rich zones suggest that the silicate glass patches are pseudomorphs of water-bearing melt drops. The drops became H2O saturated when the clinopyroxene crystallized from the host melt and were transferred to the H2O-saturated phase. Final degassing generated pervasive irregular vesicles within the pseudomorphs (Fig. 9G). This type of phase fractionation is similar to the industrial ore refining process of flotation (Ballhaus et al., 2015) and indicates that the clinopyroxene microlites crystallized during magma ascent.

4.1.5. Basaltic Xenoliths

The basaltic xenoliths can be divided into two types on the basis of their geochemistry, which we term type I (enriched MORB [E-MORB]-like) and type II (OIB-like) xenoliths. BSE images and EPMA analyses (Table S3 [see footnote 1]) show that the type I xenoliths are porphyritic (Fig. 10A) with augite and biotite phenocrysts (Fig. 10) set in a glassy groundmass. The glassy groundmass is filled with tiny vermicular vesicles <5 μm long that form small bulb pits on the surface of the thin sections. The groundmass contains numerous irregular patches of SiO2-rich glass that hosts fine-grained anhedral alkali feldspar microlites and pervasive Fe-oxides. These patches are smooth at the surface of the thin section and thus are easily differentiated from the basaltic glass with small bulb pits (Fig. 10A). The proportions of feldspar vary among patches. Some patches are vein-like and cut across the augite and biotite phenocrysts (Fig. 10B). A few smaller patches are arranged linearly (Fig. 10A). This suggests that these patches are a solidified felsic melt that was injected into the basaltic melt. The felsic melt was still fluid after the augite and biotite crystallized from the basaltic melt.

The type II (OIB-like) xenoliths are also porphyritic and have a similar mineral assemblage (Figs. 11A and 11B) to the type I xenoliths, with the following distinctive differences: (1) some clinopyroxene microlites occur in the glassy groundmass (Fig. 11B), (2) titanite and epidote phenocrysts occur with vermiculite (Figs. 11C and 11D), and (3) the opaque phase is mainly ilmenite.

4.1.6. Rhyolite

The massive rhyolite has the typical mineral assemblage and texture of a fine-grained leucogranite; however, the thinner rhyolitic beds consist of large plagioclase (~100% albite) fragments cemented by devitrified glass (Figs. 12A and 12B). The albite fragments generally show undulose extinction and kink bands (Fig. 12B), indicating ductile deformation before the solidification of the glass, given that the latter shows no evidence of ductile strain. Microfractures within the fragments and the matrix glass are widespread. The matrix consists of SiO2-rich glass that hosts numerous fine-grained, anhedral alkali feldspar microlites (Figs. 12A and 12B). Irregularly shaped vesicles are common and range in size from a few micrometers to a few millimeters in length; most have been filled with epidote fibers.

4.1.7. Carbonatite

The carbonatite has a granular texture consisting of irregularly shaped bands of dark brown calcic dense mass with fuzzy boundaries set in a groundmass of fine-grained calcite crystals (<0.05 mm) (Fig. 13A). These two phases form a matrix hosting numerous irregularly shaped quartz patches with sharp boundaries, commonly marked by a thin film (curved septa) of dense masses of calcite (Fig. 13A). The quartz patches have spherulite structure consisting of fine needles of quartz radiating from the centers (Fig. 13B). The dark brown band has the same brightness as the fine-grained calcite in BSE images (Fig. 13C), indicating that these bands represent quenched calcic melt (feathery quench dendrites). Obviously, the quartz patches are gas bulbs (vesicles) filled with tiny quartz crystals (amygdales). Dendritic microfractures most likely being tension gashes due to melt flow (e.g., Bullock et al., 2018) are widespread and have been filled by coarse calcite crystals (as large as 1 mm) alternating with dense masses of tiny quartz crystals with dendritic growth textures (Fig. 13D). The BSE imaging reveals a few small, rounded, unfilled vesicles in the fine-grained matrix (Fig. 13C).

4.2. Phase Compositions

4.2.1. Major Igneous Minerals

All of the clinopyroxene crystals from the trachyte and lamprophyre samples are a diopside-hedenbergite solid solution with minor Al (~1.5 at%), Na (0.2–0.5 at%), and Cr (0.2–0.5 at%; Table S2-1 [see footnote 1]). They yield atomic Ca/(Mg + Fe) ratios of 0.55–1.1 (Fig. 14A) and atomic Mg/(Mg + Fe) ratios of 0.4–0.9 (Figs. 14B and 14C). The clinopyroxene crystals in the basaltic xenoliths are augite with <10 mol% tschermakite. All feldspar crystals are solid solutions of albite and orthoclase (Table S2-2; Fig. 14D) with atomic Ca/(Na + K) ratios of <0.06. The feldspar crystals from the trachytes have higher albite contents (atomic Na/[Na + K] > 0.4) than those from the lamprophyre and rhyolite (<0.31). A few plagioclase crystals that are in contact with the calcic ocelli yield higher anorthite contents, with atomic Ca/(Na + K) ratios >0.09 and <1.02. Nearly all feldspar crystals contain small amounts of Fe and P. This, combined with the widespread gradational compositional zoning, suggests that they formed by rapid crystallization. Biotite crystals yield high TiO2 contents (>4 wt%; 1–2 at%), and the apatite crystals are all F bearing (4–6 at%; Table S2-3).

4.2.2. Phases in Calcic Ocelli

Forty analyses of calcic phases from the trachyte and lamprophyre samples (Table S2-4) reveal that they comprise predominantly Ca and O with minor Mg contents (total >99%). On an atomic Ca-O diagram (Fig. 15A), these analyses define a straight line with an R2 value of 0.9833 and with the atomic O/Ca ratios ranging from 1 to 3. This suggests that the calcic phases are probably mixtures of CaO, Ca(OH)2 (portlandite), and CaCO3 (calcite). However, a total of 66 spot analyses of the calcic phase from three samples using a Raman spectroscope (532 nm laser) combined with a confocal microscope reveal 3 siderites, 3 dolomites, and 60 calcites with a main Raman shift at 1085–1088 cm−1 (spectra not shown). Here we interpret these calcic phases as glassy calcite with very minor dolomite and siderite. Further study is required to reveal the reason(s) for the different results of these two techniques.

4.2.3. Opaque Phases

All of the samples contain opaque phases, most of which are Fe-oxides, although a few yield high Cr2O3 contents (Table S2-5). They yield variable Si (<8.4 at%) and low Al (<1.29 at%), Ca (<1.62 at%), Mg (<0.8 at%), and Ti (<3.45 at%) contents. On an atomic O-Fe diagram (Fig. 15B), 11 low-Si analyses plot along a straight line (R2 = 0.8767) with atomic O/Fe ratios of 1.23–2.28, suggesting that these Fe-oxides form a series from magnetite to H2O-bearing hematite. The opaque phases in the OIB-like xenolith are ilmenite.

4.2.4. SiO2-Rich Glassy Phase

A total of 64 analyses (Table S2-6) of the SiO2-rich glassy phases yield highly variable compositions. On an atomic O/Si-Si diagram, these analyses form a straight line with an R2 value of 0.9354 (Fig. 15C). This line intersects the region with Si contents of 30–33 at% and atomic O/Si ratios of ~2, indicating that the glassy phases are different mixtures of SiO2 with other oxides, including Al2O3, FeO, MgO, and CaO. About half of the analyses have atomic O/Si ratios of 2.1–3.5 and high Al (<11 at%), Mg (<12 at%), and Fe (<2.7 at%) contents; they are obviously silicate glass. Nine analyses yield atomic O/Si ratios of 1.6–1.9, suggesting the presence of natural Si or simply reflecting the analytical uncertainty of EDS. These glassy phases correspond to silicate melts with SiO2 contents ranging from 47 to 98 wt%. Some SiO2-rich glasses, particularly those from the lamprophyre (e.g., Figs. 9F and 9G), are hard to differentiate from vermiculite because of their similar brightness in BSE images. All of the glassy phases and quartz contain minor P and trace In, Sn, and Hg contents.

4.3. Geochemistry

On a total alkali–SiO2 diagram (Fig. 16A), the samples from the volcanic succession (Fig. 2A) plot in the trachyandesite, trachyte, and rhyolite fields, whereas all samples from the Fengyi Airport quarry are basaltic (45.24–49.50 wt% SiO2, values normalized to total 100%, i.e., H2O free). The contents of most compatible elements, including Co, are lower in the basalts given their lower SiO2 contents. On a Th-Co diagram (Fig. 16B), all of the basaltic samples plot in the calc-alkaline andesite field. The basalts can be divided into three groups, and the trachyte samples into two groups. Each group is distinguished from the others by their compositions and incompatible element ratios, which yield limited ranges for each group.

4.3.1. Trachyte

The bulk geochemical analyses of nine trachyte samples (Table S4 [see footnote 1]) show that they are trachydacite, with consistent SiO2 (60.0–63.5 wt%), Al2O3 (15.7–17.0 wt%), and total FeO (4.7–5.2 wt%) contents but highly variable CaO contents (2.5–6.7 wt%). They can be divided into two groups using their K2O and Na2O contents and Mg# [Mg number, molar Mg2+/(Fe2+ + Fe3+ + Mg2+)]. Six samples from the upper trachyte (Fig. 2A) have higher contents of K2O (>6.3 wt%) and Na2O (>4.6 wt%) but lower MgO (<1.6 wt%) than the other three samples (<4.2 wt%, <4.2 wt%, and >1.6 wt%, respectively) from the middle and lower trachytes. The upper trachyte yields different Mg# (0.26–0.37), total alkali contents (11.0–11.4 wt%), and K2O/Na2O ratios (1.37–1.41) from the middle and lower trachytes (0.37–0.53, 8.1–8.4 wt%, and 0.96–1.03, respectively).

The trachyte samples are enriched in Sr (890–1490 ppm) and Ba (1790–2049 ppm) with low Yb (1.0–2.5 ppm) and Y (14.3–27.6 ppm) contents, leading to high Sr/Y ratios (42–73). The high-Mg# samples show a higher degree of rare earth element (REE) fractionation than the low-Mg# samples on a primitive mantle–normalized spider diagram (Fig. 17A), which is also shown by their higher (La/Yb)PN (where the subscript PN indicates primitive mantle–normalized values; 32.2–45.7 versus 12.5–13.8, respectively), (La/Sm)PN (5.5–6.0 versus 2.9–3.3), and (Gd/Yb)PN (3.5–4.3 versus 2.5–2.7) ratios. The trachyte samples do not yield negative Eu anomalies, but they have negative Ti, Nb, and Ta anomalies with (Nb/La)PN ratios of 0.2–0.3 and (Nb/Th)PN ratios of 0.1 (Fig. 17A).

4.3.2. Lamprophyre Dikes and Basaltic Xenoliths

Six analyses of the lamprophyre dike indicate that it has high contents of CaO (17.75–20.08 wt%), MgO (7.35–9.13 wt%), and CO2 (3.24–6.47 wt%) and low Al2O3 contents (8.67–10.12 wt%). The lamprophyre samples have K2O + Na2O contents of 2.49–3.81 wt% (K2O/Na2O ratio >>1.0) and Mg# of 0.67–0.70. On a primitive mantle–normalized spider diagram (Fig. 17B), the dike samples exhibit similar patterns to the trachyte, with negative Nb, Ta, and Ti anomalies and slightly positive Sr and Sm anomalies (Fig. 17B).

The geochemical analyses reveal two types of basaltic xenolith with distinctive major and trace element contents. Two samples (type II; Table S4) are high-TiO2 (4.65 wt%; Ti/Y > 720), low-MgO (3.50–3.79 wt%) alkaline basalts with K2O + Na2O contents of >4.0 wt%, and the other three (type I; Table S4) have lower TiO2 (1.8–1.9 wt%) and higher MgO (5.6–6.5 wt%) contents. The Mg# of these two groups of basalt are 0.31 and 0.46, respectively. The most distinctive feature is seen on a primitive mantle–normalized spider diagram: The type II high-TiO2 basalts yield an OIB-like pattern with negative K, Pb, and Sr anomalies, whereas the type I low-TiO2 basalts yield an E-MORB-like pattern but with positive water-soluble element anomalies (Rb, Ba, K, Pb, and Sr; Fig. 17C).

4.4. Geochronology and Thermochronology

A total of 144, 147, 99, and 153 zircon grains from samples DL007-01 (the massive rhyolite), DL007-03 (upper trachyte), DL007-09 (a thin rhyolite bed), and DL007-14 (another thin rhyolite bed), respectively, were handpicked, mounted, and cast in epoxy resin for photographing under reflected and transmitted light and obtaining CL images. Most of the zircon grains were small (<100 μm long) and euhedral and had aspect ratios of 1.5–3. More than 54% of the zircon grains from the three rhyolitic samples (48 of 144 from DL007-01, 47 of 99 from DL007-09, and 120 of 153 from DL007-14) were intensely cracked (e.g., those indicated by white arrows in Fig. 18), with the randomly oriented microfractures filled with low-luminescence material.

Analytical spots were located in the larger fragments of the fractured zircon grains or in the unfractured zircon grains. The analytical results are listed in Table S5 (see footnote 1). We selected 32 magmatic zircon grains from sample DL007-01 for analysis, with these yielding 27 ages with >95% concordance. The concordant analyses yield apparent 206Pb/238U ages of 827–682 Ma, 19 of which form a cluster with a weighted mean age of 772 ± 6 Ma (mean square of weighted deviates [MSWD] = 3.6). All analyses plot on an isochron with an upper intercept age of 823 ± 62 Ma (MSWD = 9.2; Fig. 19A). Thirty analyses were conducted on grains from sample DL007-09, all of which yielded concordant results with apparent ages of 797–736 Ma. Twenty-eight analyses form a tight cluster with a weighted mean age of 782 ± 3 Ma (MSWD = 1.4; Fig. 19B). Thirty zircon grains were selected for analysis from sample DL007-14. Twenty-four of the analyses yielded concordant results, with ages of 798–709 Ma. Seventeen analyses form a tight cluster with a weighted mean age of 774 ± 4 Ma (MSWD = 1.4; Fig. 19C). On a concordia diagram, these analyses of sample DL007-14 plot on an isochron with an upper intercept age of 792 ± 11 Ma (MSWD = 1.7).

A total of 34 zircon grains were selected for analysis from the trachyte sample (DL007-03), with these yielding 24 results with >95% concordance. The concordant analyses fall into two clusters with weighted mean ages of 767 ± 10 Ma (n = 10; MSWD = 3.7) and 35.8 ± 0.6 Ma (n = 14; MSWD = 2.7). The two clusters and the less-concordant analyses plot on an isochron, which intercepts the concordant curve at 770 ± 22 Ma and 45 ± 28 Ma (MSWD = 6.5; Fig. 19D).

Step heating (10–13 steps) of a trachyte sample (DL007-10) released ~100% of the 39Ar, yielding an acceptable WMPA of 37.37 ± 0.39 Ma (Fig. 20A) and an identical isochron age (Fig. 20B). Step heating of a type I basaltic xenolith sample (DL009-03; Figs. 20C and 20D) yielded an acceptable WMPA of 35 ± 1 Ma (including 45.9% of the 39Ar released) and an isochron age of 35.8 ± 4.6 Ma (MSWD = 5.0). The analytical data are listed in Table S6.

Our new data demonstrate that both the trachytes of the Wase Basin and the lamprophyre dikes in the wider region are lamprophyres in terms of their geochemistry and petrography (Le Maitre et al., 2002; Rock, 1991). They share the following characteristics: (1) hi gh loss on ignition (mostly >1.0 wt%, see Table S4 [see footnote 1]), indicating high volatile contents (mainly H2O and CO2); (2) phenocrysts of clinopyroxene (diopside-hedenbergite solid solution) and biotite; (3) alkali feldspar laths occurring only in the groundmass; (4) high K, Rb, Sr, and Ba contents; (5) an arc trace element signature (i.e., negative Nb, Ta, and Ti anomalies; see, e.g., Wilson, 2007); (6) formation over a short period (37–35 Ma); (7) numerous spherulites; and (8) common disequilibrium textures, including compositional zoning of clinopyroxene and feldspar crystals. These similarities imply a petrogenic connection between the trachytes and the lamprophyre dikes, despite the compositional gaps between the two. The widespread disequilibrium textures record the evolution of the magmas.

5.1. Calcic and Silica-Rich Ocelli: Drops of Carbonite and Felsic Melt

There are two types of ocelli within the Wase igneous rocks: Some are SiO2 rich, and some are calcic. The origin of ocelli, particularly the calcic ocelli, remains poorly understood (Ballhaus et al., 2015; Cooper, 1979; Ferguson and Currie, 1971; Le Roex and Lanyon, 1998). Several contrasting models have been proposed to explain ocelli and other globular or orbicular structures in igneous rocks, including melt segregation caused by melt immiscibility (Ferguson and Currie, 1971) or fluid saturation (Ballhaus et al., 2015; Cooper, 1979); late-stage alteration, i.e., devitrification of glass or solid-state diffusional alteration (Hughes, 1977; Philpotts, 1977) or fluid infiltration (Hanski, 1993); filling of free spaces (i.e., vesicles) with liquid minerals (amygdales) (Anderson et al., 1984; Cooper, 1979; Smith, 1967); and mingling or incomplete mixing of two melts (incomplete assimilation) (Goldie, 1980; McNaughton et al., 1988; Vogel and Wilband, 1978).

The total absence of hydroxyl silicate mineral, the tabular segregations of phases (Figs. 5D, 6C, 7D, and 9C), and the generally glassy texture of these phases discredit the possibility that the Wase ocelli are amygdales or gas vesicles infilled with late-stage residual liquid. Any conjugate immiscible liquids, which are in equilibrium, should crystallize phases in equilibrium, i.e., rocks related through liquid immiscibility should have similar mineral assemblages (Bowen, 1928) although the modal proportions of minerals formed in each phase may be quite different because of the differences in liquid compositions (Hanski, 1993). The distinctive mineral assemblage and crystal size between ocelli and their black matrix make melt segregation or late-stage alteration equally improbable as formation mechanisms of the Wase ocelli. On the contrary, these features suggest that they may be incompletely assimilated inclusions, i.e., the result of mingling and incomplete mixing of three melts from three different sources. Widespread tiny vesicles in the ocelli (e.g., Figs. 6B and 7A) are indicative of the presence of vapor or liquid in the melt. The Wase ocelli probably represent quenched drops of melts.

As mentioned earlier, there are two types of ocelli within the Wase igneous rocks: Some are SiO2 rich, and some are calcic. They do not occur together in any individual sample. The silica-rich ocelli comprise a cryptocrystalline groundmass of mostly SiO2 that hosts a few feldspar microlites. Their microtextures (Fig. 7D) are comparable to those of the devitrified glass in the rhyolitic lava (Figs. 12C and 12D). Moreover, the feldspar microlites, both in the ocelli and in the rhyolite, contain minor P and Fe (Table S2-2 [see footnote 1]), indicating feldspar in both locations results from rapid crystallization. Combined with the sharp, faceted boundaries of the silica-rich ocelli and abundant vesicles (Figs. 7, 8, and 12), these similarities suggest that the silica-rich ocelli are solidified pseudomorphs of drops of felsic melt that were mixed with a basaltic melt (see Section 5.4 for further discussion). Some felsic melts reached the surface and were rapidly quenched to form the rhyolite horizons (Fig. 2A).

The calcic-rich ocelli in the trachytes and lamprophyre are made predominantly of non-crystalline calcite and silicate glass with highly variable proportion between the two phases even in an individual thin section (Figs. 6B and 6C). The similarity in composition between the silicate glass in the calcic-rich ocelli and the silicate glass in the groundmass of the ocelli, combined with fuzzy boundary between the silica-rich and calcic phases of the ocelli (Figs. 5D, 6B, and 9B), indicates that the calcic ocelli are solidified pseudomorphs of drops of a mixed melt consisting of a calcic melt and a silicate melt. Here the silicate melt represents the latest-stage melt of a partially crystallized basaltic magma. A few calcic melts reached the surface and were rapidly quenched to form the carbonatite (Figs. 2A and 2C). Both the calcic ocelli (Fig. 6A) and the carbonatite (Fig. 13) contain amygdales consisting of dense masses of quartz needles that are commonly cored by a calcite crystal.

5.2. Generation of the Felsic and Calcic Melts: Rapid H2O-Saturated Partial Melting of the Yangtze Block Basement

Our 40Ar/39Ar and zircon U-Pb ages show that the middle trachyte erupted at ca. 37 Ma and the upper trachyte erupted at ca. 36 Ma. The consistency between the ages and the stratigraphic positions of the samples (Fig. 2) demonstrate that the volcanic succession formed over a short period (37–36 Ma). The rhyolitic and carbonatitic volcanic rocks that are intercalated with the trachytes should have formed over the same period; however, our U-Pb ages for 92 zircon grains from 3 rhyolitic samples all yield concordant ages much older than 37 Ma, suggesting they were inherited from their source rocks and dating of them does not constrain the timing of emplacement. In addition, the U-Pb analyses from each sample form one tight cluster with weighted mean ages of 782, 774, and 772 Ma (Fig. 19), which are within analytical uncertainty of each other. The geochronological data and the zircon zoning patterns (Fig. 18) suggest that the source rock of the rhyolite is a granitic pluton with a zircon U-Pb age of ca. 770 Ma. Petrographic evidence supports this suggestion. The rhyolite consists of devitrified glass and numerous angular plagioclase fragments (Fig. 12). The latter experienced ductile deformation and subsequent fracturing, suggesting they are a remnant phase. The devitrified glass contains pervasive irregularly shaped vesicles that have been rapidly filled by epidote fibers (Fig. 12D), suggesting that the melt was rich in H2O and that the melting took place under H2O-saturated conditions.

Acosta-Vigil et al. (2006) conducted melting experiments on an H2O-saturated leucogranit e at 200 MPa and 690–800 °C. The leucogranite consisted of plagioclase (~90 mol% albite) + K-feldspar + quartz + minor biotite. Their experiments reproduced similar microtextures and phase assemblages to those of the Wase rhyolites. These experiments showed that quartz, plagioclase, biotite, and K-feldspar were remnant phases at 740 °C, whereas biotite and feldspar disappeared at temperatures ≥800 °C. Therefore, we suggest that the Wase rhyolite was likely generated by H2O-saturated partial melting of a ca. 770 Ma granite pluton at a temperature of ≥800 °C. During melting, no zircon and little plagioclase from the source rocks was resorbed, but they were fragmented (Figs. 12A, 12B, and 18). As a result, no new zircon or plagioclase recrystallized when the melt solidified. This indicates that the partial melting was rapid, most likely comparable to that in the melting experiments of Acosta-Vigil et al. (2006).

The basement of the Yangtze block consists of Neoproterozoic plutons and upper Proterozoic strata comprising predominantly intercalated sandstone and limestone (Geng et al., 2008; Yunnan Bureau of Geology and Mineral Resources, 1990). When the granitic plutons were partially melted at temperatures of ≥800 °C (<2 kbar, see next section) under H2O-saturated conditions, the sedimentary country rocks would have inevitably melted as well, and the coeval H2O-saturated partial melting of the limestone (Durand et al., 2015; Foustoukos and Mysen, 2015; Wyllie and Tuttle, 1960) may have generated calcic melt (Özkan et al., 2021; Schumann et al., 2019).

Wyllie and Tuttle (1960) conducted a series of melting experiments on the CaO-CO2-H2O system at 1 kbar between 600 and 1320 °C. These experiments revealed that melting begins at 740 °C on the calcite-water phase boundary, isobaric invariant equilibrium between CaO + calcite + portlandite + liquid occurs at 683 °C, and equilibrium between calcite + portlandite + liquid + vapor occurs at 675 °C. Using a similar experimental protocol, Durand et al. (2015) showed that partial melting of a calcite + H2O system begins at a lower temperature, below 650 °C. It is possible that the original melt of the calcic ocelli in the trachytes and lamprophyre had the same phase assemblage as the quenched melts produced by the experiments of Wyllie and Tuttle (1960), i.e., CaO + calcite + portlandite, some phases of which likely have been consumed by melt reactions during melt mixing (see Sections 5.4 and 5.5 for further discussion). We suggest that the calcic ocelli represent drops of calcic melt quenched at 650–680 °C, and the melts were derived from a calcite + H2O system at >740 °C.

5.3. What Caused the H2O-Saturated Upper Crustal Conditions?

The volcanics of the Wase Basin overlie Devonian sedimentary rocks. The post-Devonian strata were eroded before the eruption of the Wase volcanic rocks. Geological mapping shows that the total thickness of the sedimentary cover in the Wase region is ~11,250 m, including ~5440 m of Permian Emeishan basalts and ~1000 m of Mesozoic strata (First Geological Survey of Yunnan Province, 1973). This means that the basement of the Yangtze block was at a depth of <5 km when the granite, sandstone, and limestone were partially melted under H2O-saturated conditions. This depth (<2 kbar lithostatic pressure) is consistent with the pressure of the experiments mentioned above.

High-temperature, fluid-dominated upper crustal conditions are difficult to achieve in nature (Clemens et al., 2020). The Wase Basin is in the eastern central part of the ENE-WSW–trending magmatic zone (the lithospheric transitional zone) in SE Tibet (Fig. 1A; Yang et al., 2021), where numerous mantle-derived igneous rocks were emplaced at 37–34 Ma (Guo et al., 2005; Lu et al., 2013). A compilation of geophysical, geochronological, sedimentary, and structural data shows that the magmatic zone likely resulted from NNW-SSE–directed lithospheric stretching at 50–25 Ma (Liang et al., 2022; Yang et al., 2021), which caused asthenospheric upwelling that triggered partial melting of the mantle. We suggest that most of the mantle-derived magmas accumulated in the middle crust. The crystallized basaltic magmas cause the higher P-wave velocities beneath the magmatic zone than in the normal middle crust beneath the regions to the south of the magmatic zone that were revealed by the deep seismic sounding results of Lin et al. (2019). Crystallization of the accumulated basaltic magma would have released high-temperature (900–950 °C), H2O-rich fluids (Huppert and Sparks, 1988), a process that has been numerically simulated by Lamy-Chappuis et al. (2020). These fluids rapidly heated the basement of the Yangtze block and led to H2O-saturated partial melting of the Neoproterozoic granite and its sedimentary country rocks.

The Wase rhyolites have highly variable and decoupled Nb/U (2.88–12.29) and Ce/Pb (0.58–21.27) ratios (Fig. 21). Although the Nb/U ratios are broadly similar to average continental crust values (Nb/U = 4.3–12.1, Ce/Pb = 3.3–4.1; Rudnick and Gao, 2003), some of the Ce/Pb ratios are the same as that of MORB (Ce/Pb = 24.35; Gale et al., 2013) or OIB (Ce/Pb = 17.5–53; Kawabata et al., 2011). This indicates the addition of a mantle-derived H2O-rich fluid, given that both Ce and Pb are fluid mobile. Indeed, the contents of most fluid-mobile elements in the Wase igneous rocks, including K, Sr, Ba, Rb, Th, and U, are positively correlated, but their contents decrease with increasing contents of the most compatible elements, including Co (Fig. 22).

5.4. Magma Mixing

New zircon U-Pb geochronological data (Fig. 19D) and textural evidence suggest that the trachyte and lamprophyre in the Wase Basin resulted from different degrees of mixing among felsic, calcic, and basaltic melts. As discussed earlier, the felsic melt was derived from partial melting of Neoproterozoic granite, while the calcic melt resulted from partial melting of limestone. This triple magma mixing accounts for the distinct difference in composition between the two rock types (Fig. 16) and the geochemical heterogeneity of the lower and middle trachytes shown by their different REE patterns (Fig. 17A).

The variations in the CaO and alkali contents of the Wase igneous rocks should be caused mainly by additions of calcic and felsic melts, respectively. The K and Na in the pseudomorphs of the felsic melt are hosted predominantly in the feldspar microlites (Fig. 12C), meaning that the average Na2O/(Na2O + K2O) ratio (~0.222) of the feldspar microlites may approximate the ratio of the felsic melt. This means that the addition of felsic melt would lead to a larger change in K2O contents than Na2O contents. The trachyte samples plot on a straight line on a K2O-Na2O diagram (R2 = 0.9835; Fig. 23A). This line has a slope of ~0.227, similar to the average Na2O/(Na2O + K2O) ratio (~0.222) of the felsic melt.

On a CaO/(CaO + Na2O)–Na2O diagram (Fig. 23B), the basaltic xenoliths, lamprophyres, and trachytes define a straight line with an R2 value of 0.9546, with the basaltic xenoliths in the middle. This trend suggests that the lamprophyres were generated by the addition of mainly calcic melt to a basaltic melt, whereas the trachytes were generated by the addition of predominantly felsic melt. Indeed, the trachytes contain numerous Neoproterozoic zircons (sample DL007-03; Fig. 18), while we did not separate any zircon from a lamprophyric sample. In addition, all ocelli in the lamprophyre samples are calcic. The addition of calcic melt diluted the contents of some elements, most strikingly SiO2 (<47 wt%) and Na2O (Fig. 23C). The dilution of SiO2 would have been partly balanced by the addition of a small amount of felsic melt, as indicated by the higher K2O content of the lamprophyres than the basaltic xenoliths (Fig. 23A).

5.5. Melt Reactions and Crystallization of Phenocrysts of Diopside-Hedenbergite Solid Solution

While pyroxenes in other mafic rocks are augites containing some tschermakite, diopsid e-hedenbergite solid-solution phenocrysts occur in lamprophyric rocks worldwide. The processes leading to crystallization of the low-tschermakite clinopyr oxene are poorly understood (Rock, 1987; Xing and Wang, 2020). According to the experimental results of Wyllie and Tuttle (1960), H2O-satured partial melting of limestone would produce the phase assemblage CaO (lime) + calcite + Ca(OH)2 (portlandite). The reactions below would have occurred between the calcic and basaltic melts when they met (antiskarnification; Yaxley et al., 2022):


The CaO and Ca(OH)2 in the former two reactions are much more inconstant phases than the CaCO3 in the third reaction. As such, they would have been consumed firstly. Indeed, the remnant phase in calcic ocelli is predominantly CaCO3.

The clinopyroxene (diopside-hedenbergite solid solution) crystals yield variable atomic Ca/(Mg + Fe) (0.3–1.1) and Mg/(Mg + Fe) ratios (Figs. 14A14C), both of which vary with oscillatory zoning. As mentioned earlier, on the Ca/(Mg + Fe)–Ca diagram, all of the analyzed clinopyroxene crystals plot on a straight line with an R2 value of 0.9043 (Fig. 14A), whereas the variation in Mg/(Mg + Fe) ratios is independent of the Ca contents (Fig. 14B) but increases with Mg content (Fig. 14C). This indicates that increases in the Ca/(Mg + Fe) ratio correspond to increases in the Ca content caused by the addition of calcic melt.

5.6. When and Where Did Magma Mixing and Reactions Occur?

The isothermal time series experiments of Pontesilli et al. (2019) show that clinopyroxene microlites crystallize rapidly from a nominally H2O-bearing (2 wt%) melt with the composition of a trachybasalt from Mount Etna (Sicily, Italy) at 1100 °C and 400 MPa. Their experimental material (Armienti et al., 1989) is chemically similar to the Wase trachyte. Moreover, the experimentally produced clinopyroxene microlites have similar distribution within the melt, crystal morphology, grain size (~100 μm long), composition (slight Si depletion), and patchy compositional zoning to the clinopyroxene microlites in the Wase igneous rocks (Figs. 5B, 6B, and 9). These similarities suggest that the clinopyroxene microlites in the Wase igneous rocks likely crystallized at approximately the same pressure-temperature (P-T) conditions as the experiments (1100 °C and ~12 km depth; Pontesilli et al., 2019).

As well as the clinopyroxene microlites, most of the Wase igneous rocks contain clinopyroxene phenocrysts (Figs. 4, 5A, and 9) that have been fragmented, as shown by microfractures and the truncation of oscillatory zones (Figs. 9D and 9E). The microfractures have been partially filled by silicate glass (Fig. 4D). Voids with angular, faceted outlines are developed along the faces of some phenocrysts as pressure shadows (Figs. 4A and 4B). This suggests that the fragmentation was syn-eruptive and took place owing to decompression during eruption (van Zalinge et al., 2018). Similar textures are rare in the microlites, suggesting that the phenocrysts crystallized before the microlites at a temperature of >1100 °C.

Xing and Wang (2020) suggested that periodic mixing of magmas probably has led to oscillatory zoning of clinopyroxene phenocrysts in some lamprophyre rocks. For the Wase lamprophyres, it is possible that mixing of the felsic melt and the calcic melt with basaltic melt and the reactions among these melts likely took place during crystallization of clinopyroxene phenocrysts, which may account for oscillatory zoning of the phenocrysts (Fig. 4D). Meanwhile, addition of the relatively cooler melts probably cooled the basaltic melt, triggering crystallization of the clinopyroxene microlites.

Development of the clinopyroxene microlit e–concentrated zone surrounding some silicate glass patches (Figs. 9G and 9H) indicates the occurrence of phase fractionation (e.g., Ballhaus et al., 2015). This suggests that these glassy patches are H2O saturated. It is the existence of these H2O-saturated melt patches that cause the phase fractionation to occur, just like the industrial ore refining process of flotation.

The Wase lamprophyres have similar textures to the dense andesitic bombs from Galeras volcano, Colombia. Both have low vesicularity (<5%) and high crystallinity (Bain et al., 2019), indicating that the degassing and crystallization probably took place in the shallow conduit when the magmas had reached the surface. The lamprophyre dikes represent dense, degassed, and highly crystalline plugs formed in shallow conduits, whereas the trachytes are effusive extrusions of degassed, highly crystalline magma. The degassing reduced the H2O content of the magma and triggered the crystallization of feldspar (Mollo et al., 2010; Müntener et al., 2001). The late-stage feldspar laths contain numerous clinopyroxene microlites (Fig. 9D).

According to experimental results (Acosta-Vigil et al., 2006; Durand et al., 2015; Foustoukos and Mysen, 2015; Wyllie and Tuttle, 1960), the solidus temperatures of both the felsic (~740 °C) and calcic (~650 °C) melts are well below the crystallization temperature of clinopyroxene (~1100 °C; Pontesilli et al., 2019). This means that after crystallization of the clinopyroxene and probably feldspar laths, the calcic and felsic melts that were hosted by the lamprophyre magma would have remained fluid, forming isolated melt drops with angular, faceted outlines (Figs. 7 and 8) in a crystalline matrix. This explains why most calcic ocelli contain silicate glass of the latest-stage fluid of the basaltic melt. Some calcic melts flowed along microfractures in the lamprophyre dikes plugging the shallow conduit, as shown by the non-crystalline CaCO3 mic roveins that cut across the fragmented clinopyroxene phenocrysts (Fig. 9F) and the feldspar and clinopyroxene laths (Figs. 9G and 9H). Meanwhile, some felsic melt drops segregated into two phases: a H2O-rich fluid and a silica melt (Figs. 6 and 9) (Ballhaus et al., 2015). The latter solidified rapidly to form glass (e.g., Fig. 8B), and the segregated H2O-rich fluids were stored in vesicles where epidote (Fig. 11D), vermiculite (Figs. 9 and 11), Fe-oxides, and opal agate (Figs. 8C and 8D) precipitated, depending on the solutes in the final fluid.

In summary, rapid H2O-saturated partial melting of granite and limestone generated the felsic and calcic melts, respectively. They were separated from each other and stored at upper crustal depths as transient magma lenses or parcels. These magma parcels would have had different compositions depending on their source rocks and formed a magma zone along the roof of the accumulated and crystallized mantle-derived basaltic melt (Fig. 24). When an ascending hot basaltic melt intersected this magma zone, the melts mixed and the mixed melt started to crystallize, assisted by melt reactions. The chemistry of any individual hybrid magma would have been determined by the composition(s) of magma parcel(s) that had been intersected by the ascending basaltic melt. The crystal-bearing hybrid magma then continued ascending, during which time some phenocrysts were fragmented and the clinopyroxene microlites formed. Feldspar laths and biotite flakes most likely crystallized when the magma erupted (the trachyte) or rose close to the surface (the lamprophyre dikes), assisted by progressive degassing. Rapid cooling of the lava preserved the calcic and felsic ocelli and numerous other disequilibrium textures. The structure of the lithosphere in the magmatic zone, including the partially molten upper crust and the location of H2O-saturated partial melting, is shown in Figure 24.

5.7. New Insights into the Petrogenesis of Lamprophyre-Carbonatite Associations

The well-preserved disequilibri um microtextures in the Wase lamprophyres enable the reconstruction of their magmatic evolution. The microtextures and phases of each key stage of this process have approximate equivalents that have been experimentally reproduced under controlled physiochemical conditions in laboratories. The lamprophyres in SE Tibet are similar to other lamprophyre-carbonatite associatio ns around the world (Lustrino et al., 2016; Rock, 1987); therefore, our new data and interpretations may shed light on the petrogenesis of other lamprophyre-carbonatite associations.

Lamprophyres are usually distributed over a large area despite the fact that they are volumetrically minor in most regions. As seen in other lamprophyre provinces, including the Early Cretaceous provinces in the eastern North China Craton (Ma et al., 2016; Yang et al., 2019), S to SE Brazil (Florisbal et al., 2018), and Namibia (Le Roex and Lanyon, 1998; Owen-Smith et al., 2017), the late Eocene–Oligocene lamprophy res in SE Tibet are exposed over a large area (>550 × 250 km; Fig. 1A; Guo et al., 2005; Lu et al., 2013; Yang et al., 2021). All of these provinces are located in continental areas where lithospheric extension and related asthenospheric upwelling would have produced a large volume of mantle-derived basaltic magma. It is reasonable to suggest that some of the basaltic melt accumulated in the crust (Lamy-Chappuis et al., 2020) and induced H2O-saturated partial melting of the crustal rocks (Holness and Clemens, 1999; Huppert and Sparks, 1988).

The lamprophyres are cognate with other alkaline, shoshonitic, and adakitic rocks (Ding et al., 2016; Ma et al., 2016; Xin et al., 2020) and minor carbonatite (Hou and Cook, 2009; this study). These rocks are readily distinguishable from each other using incompatible element ratios, including Th/Nb, Th/La, Nb/U, and Ce/Pb ratios, as well as their Sr, Nd, Pb, and O isotopic compositions (Chen et al., 2017; Ding et al., 2016; Guo et al., 2005; Lu et al., 2013; Miao et al., 2021; Xu et al., 2007, 2019; Zhao et al., 2016). The alkaline rocks found within an individual complex have similar isotopic ratios, whereas different complexes in the same province yield a wide range of isotopic ratios (e.g., Chen et al., 2017; Guo et al., 2005; Lu et al., 2013; Xu et al., 2019). In addition, these isotopic ratios are not correlated with the associated SiO2 contents (Fig. 25) and other major element contents (see Lustrino et al. [2016] for other lamprophyric rocks elsewhere). This suggests that each complex has an independent origin, as also shown by the contrasting petrogenic hypotheses presented in previous studies.

Previously suggested hypotheses can be divided into two major groups: One group stresses the importance of chemically and isotopically anomalous upper mantle (Jaques et al., 1986; Rock, 1987), and the other places more emphasis on interactions with the crust (Chandra et al., 2019; van Bergen et al., 1983). The heterogeneous mantle model requires a large mantle fragment that contains numerous small parcels with different incompatible element and isotopic ratios from their wall rocks (e.g., the vein-plus-wall-rock model of Foley [1992]) in order to account for the small volume of each individual alkaline pluton across a large province. The Wase volcanic rocks and numerous other alkaline rocks worldwide (e.g., Rezeau et al., 2018; Leuthold et al., 2014; Streck, 2008) acquired their petrological and bulk-rock chemical characteristics during the final stages of their ascent, where rapid cooling preserved their heterogeneous compositions and widespread disequilibrium textures. These features are commonly considered to be indicative of open-system processes, which have no counterparts in typical mantle-derived melts (e.g., Yoder, 1986). Our new model (Fig. 24) provides a simpler explanation that emphasizes magmatic processes in the upper crust, where meter-scale geochemical heterogeneity is readily observable.

Quenched calcic melt drops are not common in nature (Schumann et al., 2019). Despite their small volume and outcrop area, identification of calcic melt drops demonstrates that calcic melt may be generated by H2O-saturated partial melting of upper crustal limestone or marble (Durand et al., 2015; Foustoukos and Mysen, 2015; Özkan et al., 2021; Schumann et al., 2019; Wyllie and Tuttle, 1960). Limestone and marble are widespread in the continental crust but rarely found in mature oceanic crust. Thus, “most carbonatites are restricted to continental areas” (Bell and Simonetti, 2009). Calcite glass is well preserved in the Wase igneous rocks because they were generated at upper crustal levels and then quenched rapidly at the surface. Although further study is required, we speculate that those carbonatites with more complex mineral assemblages and textures in other associations of alkaline rocks and carbonatites (Bell and Simonetti, 2009) were likely formed at middle or lower crustal levels by similar processes. The higher P-T conditions and longer distance to the surface would have led to more intensive magma mixing and melt reactions, which would have erased most of the textural evidence. In our model, an asthenosphere-derived H2O-rich fluid induced the partial melting. This may account for some of the geochemical features, particularly the isotopic ratios of noble gases that “support the generation of carbonated melts from sub-lithospheric mantle” (Bell and Simonetti, 2009).

The results of this work challenge much of what has been written about generation of lamprophyre and associated rocks not only in the SE Tibetan Plateau but also elsewhere in the word. Our field, petrographic, geochemical, and geochronological analyses of the lamprophyres from the Wase Basin form an internally consistent data set. This data set demonstrates that the lamprophyre magma formed at a shallow crustal level by magma mixing and reaction. Rapid cooling of the mixed magmas preserved numerous disequilibrium textures that record the magmatic evolution. Moreover, each stage of the process is supported by available experimental data. The new data and ideas presented in this paper address several long-debated issues about lamprophyres and provide new insights into mantle-crust interactions, although further studies are required. Our model suggests that the lamprophyres themselves provide little information about the evolution of the deep mantle. Our major conclusions are below.

  1. The Wase Basin, SE Tibet, comprises mainly late Eocene lamprophyric volcanic rocks with a few intercalated rhyolites and minor carbonatites.

  2. Numerous millim eter-sized ocelli comprising non-crystalline calcite (CaCO3) and a SiO2-rich phase in the lamprophyric volcanics and coeval lamprophyre dikes are quenched melt drops of calcic and felsic melts, respectively. This, combined with the widespread inherited zircons of granite, indicates that the lamprophyre magma was generated by triply mixing of calcic and felsic melts with basaltic melt.

  3. Rapid H2O-saturated partial melting of Neoproterozoic granite and the surrounding limestone in the upper crust generated the felsic and the calcic melts, respectively. The crystallization of the mantle-derived magmas that accumulated in the middle crust gave rise to high-temperature, fluid-dominated upper crustal conditions.

  4. The highly variable melt compositions were inherited from their heterogeneous upper crustal sources, and the different degrees of magma mixing account for the diversity in bulk-rock compositions as well as the diverse mineral assemblages in lamprophyre-carbonatite associa tions, not only in SE Tibet but also globally.

  5. The low-tschermakite-content clinopyroxe ne phenocrysts of lamprophyres probably formed during the earliest stage of evolution of the mixed melt through the reaction between calcic and basaltic melts.

  6. Zircons in the Wase rhyolite are all inherited, and zircon dating by conventional U-Pb techniques cannot always determine the actual emplacement age of such rocks.

1Supplemental Material. Table S1: List of samples for petrographic study from the Wase Basin and its adjacent regions, SE Tibetan Plateau. Table S2: Energy dispersive X-ray spectra data of six major phases of igneous rocks from the Wase Basin and its adjacent regions, SE Tibetan Plateau. In addition, the analytical results of each phase are listed in six separate sub-tables (Tables S1-1 to S1-6). Table S3: Microprobe analytical results of minerals of igneous rocks from the Wase Basin and its adjacent regions, SE Tibetan Plateau. Table S4: Bulk geochemistry of the rocks from the Wase Basin and its adjacent regions, SE Tibetan Plateau. Table S5: Zircon U-Pb dating results of the igneous rocks from the Wase Basin, SE Tibetan Plateau. Table S6: 40Ar/39Ar data of samples from the Wase Basin and its adjacent regions, SE Tibetan Plateau. Please visit to access the supplemental material, and contact with any questions.
Science Editor: Christopher J. Spencer
Associate Editor: Julie Roberge

This study is supported by the National Natural Science Foundation of China (92055206) and the Ministry of Sciences and Technology of China (2022YFF0800901). Constructive comments and suggestions by the journal editor (Christopher Spencer), editorial staff, and journal reviewers (Michael Anenburg, Ashutosh Pandey, and Pradip Kumar Singh) helped to improve the manuscript.

All geochemical, geochronological, and mineral compositional data generated or analyzed during this study are included in this published article and its Supplemental Material. Raman shift spectra data are available by contacting the corresponding author (

Gold Open Access: This paper is published under the terms of the CC-BY-NC license.