Basalts and trachybasalts are mainly distributed in the eastern, western and central areas of the rift basin. A total of 13 samples were collected at the following specific locations: Mangbang, Tuantian, Mengnong, Sudian and Qushi (Figures 1B, Figures 2A–C). All samples were collected from quarries and road cuts, and the samples were of fresh quality. Under microscopic observation, the basalts and trachybasalts are characterized by weak to moderate porphyritic texture (Figures 2D–F), with ∼4–6 vol.% olivine, ∼3–8 vol.% clinopyroxene, and ∼6–10 vol.% plagioclase. The diameters of most olivines range from 100 to 500 μm.

Whole-rock major and trace element analyses were carried out at the Key Laboratory of Crustal Dynamics, China Earthquake Administration. Sr and Nd isotopic ratios were analyzed at the Institute of Geochemistry, Chinese Academy of Sciences. The procedure to obtain the whole-rock major and trace element and Sr–Nd isotope data followed Guo et al. (2015). Before conducting the three types of experiments, the samples were first powdered. The major element contents of basalts were obtained by measuring their loss on ignition (LOI) and oxides. The LOI of the samples was determined after heating the basalt powders at a temperature of 1,000°C. Subsequently, the powders were transformed into fused disks, and major element oxides were analyzed by using Axios-Minerals sequential X-ray fluorescence (XRF) on fused disks. The trace element contents of basalt were examined using inductively coupled plasma–mass spectrometry (ICP–MS) after 4 days of HF+HNO₃ digestion. Analytical precisions are generally better than 1% relative for major elements and 5% relative for trace elements. Sr and Nd isotopic ratios were analyzed by using a Neptune Plus mass spectrometer after HF + HNO₃ + HClO₄ digestion. The mass fractionation corrections for Sr and Nd isotope ratios were based on ⁸⁶Sr/⁸⁸Sr = 0.1194 and ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219, respectively. The analyzed international standard NBS987 had an ⁸⁶Sr/⁸⁸Sr ratio of 0.710245 ± 0.000016 (n = 5), and standard JNdi-1 had a¹⁴³Nd/¹⁴⁴Nd ratio of 0.512117 ± 0.000012 (n = 5), which implies good instrument status and accurate testing results.

Major element analysis of olivines was conducted at the East China Institute of Technology. The olivine phenocrysts were manually picked, embedded in epoxy resin and polished on one side for subsequent major and trace element analyses. The major elements of olivine were measured by using a JEOL JXA-8100 electron probe microanalyzer (EPMA). The parameter settings for the experiment were as follows: an acceleration voltage of 15 kV, a beam current of 20 nA, and a beam diameter of 5 μm. During the measurements, the peak counting time was set to 20 s for major elements (Si, Fe and Mg) and 90 s for minor elements (Mn, Ni, Cr, Ca and Al). After the measurement of every five olivine phenocrysts, an internal olivine standard (MongOl; Batanova et al., 2019) was measured to monitor instrumental drift. According to the analyzed standard and recommended values, the analytical uncertainty is less than 1% for major elements and 5% for minor elements. Olivine major elements are presented in Supplementary Table S2 in Supplementary Information (S1).

Trace elements of olivine were measured by an ELEMENT XR (Thermo Fisher Scientific) inductively coupled plasma–sector field–mass spectrometer coupled with a 193-nm (ArF) Resonetics RESOlution M-50 laser ablation system in the State Key Laboratory of Isotope Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences. The detailed experimental procedure and data reduction strategy are described in Zhang et al. (2019). To analyze olivine trace elements, we set the machine parameters to a laser beam size of 45 μm with a repetition rate of 5 Hz and an energy density of ∼4 J cm⁻². In addition, the time for the blank collection was 20 s, and the time for signal detection was 30 s. Si was selected as the internal standard element before EPMA analysis. After the measurement of every seven olivine phenocrysts, three United States Geological Survey (USGS) reference glasses, namely, BCR-2G, BHVO-2G and GSD-1G, were examined to correct for time-dependent drift. The measured values of reference glasses agree with the recommended values or the values of previous studies, and the analytical precision was better than 10% for most elements. Olivine trace elements are presented in Supplementary Table S2 in the Supplementary Information (S1).

Tengchong basalts show relatively low SiO₂ (47.6–52.1 wt.%) and moderate MgO (4.6–9.1 wt.%) and K₂O+Na₂O (4.2–6.3 wt.%) (Supplementary Table S1 in Supplementary Information (S1)), which plot in the fields of basalts and trachybasalts on the total alkali-silica diagram (Figure 3A). The basalts from the western and eastern areas are slightly more alkaline than those from the central area (Figure 3).

As shown in whole-rock major and trace element covariation diagrams (Figure 4), the Tengchong basalts show negative correlations between MgO and SiO₂, Al₂O₃, and TiO₂, positive correlations between MgO and the Mg number and Ni content, and constant correlations between MgO and CaO contents and CaO/Al₂O₃ ratios.

The pattern of primitive mantle-normalized incompatible trace elements shows positive anomalies in large ion lithophile elements (LILEs, e.g., K, Rb and Ba) and negative anomalies in high field strength elements (HFSEs, Nb, Ta, Zr, Hf and P) (Supplementary Table S1 in Supplementary Information (S1)). The negative Nb-Ta-Ti anomalies and positive Pb anomalies shown in Figure 5A indicate that the Tengchong volcanism is subduction-related. The Tengchong basalts are enriched in light rare earth elements (LREEs). Furthermore, basalts from the central region show higher heavy rare earth elements (HREEs) and lower LREEs than those from the eastern and western regions (Figure 5B).

The Sr–Nd isotope compositions of the Tengchong basalts are shown in Supplementary Table S1 in the Supplementary Information (S1). The ⁸⁷Sr/⁸⁶Sr ratios range from 0.705269 to 0.708690, and the ¹⁴³Nd/¹⁴⁴Nd ratios range from 0.512359 to 0.512683. In the ¹⁴³Nd/¹⁴⁴Nd versus ⁸⁷Sr/⁸⁶Sr diagram (Figure 6), the samples are located between the EM-I and EM-II endmembers. The ⁸⁷Sr/⁸⁶Sr ratios of basalts from the central area are similar to those from the western area, and they are lower than those from the eastern area. Integrating previous studies, Figure 6 shows that basalts from the central area generally have relatively lower ⁸⁷Sr/⁸⁶Sr ratios in the Tengchong volcanic field.

Olivine in the Tengchong basalts exhibits three main features. First, olivines were generated from the crystallization of magmas. The major and trace elements of olivine are displayed in Supplementary Table S3 in the Supplementary Information (S1). Previous studies have shown that olivine derived from peridotite has low CaO contents (<0.1 wt.%) and that the crystallized from magmas has high CaO contents (>0.1 wt.%) (Simkin and Smith, 1970; Kamenetsky et al., 2006; Wu et al., 2022). Olivine samples from the Tengchong basalts display high CaO (0.14–0.35 wt.%) and moderate Fo (75.61–86.21) contents (Supplementary Table S3 in Supplementary Information (S1)). Second, olivine is in equilibrium with its host magmas. Based on the Fe-Mg exchange coefficient K_D(Fe-Mg)^ol-liq=0.30 ± 0.03 (Roeder and Emslie, 1970), we examine olivine-liquid equilibrium for olivines in the Tengchong basalts (Supplementary Figure S1 in Supplementary Information S2). Although some olivines show variable degrees of olivine crystallization, the data are close to the K_D, indicating olivine-melt equilibrium (Supplementary Figure S1). Third, most olivine phenocrysts in the Tengchong basalts are homogenous. The olivines do not show significant compositional differences from core to rim (Figures 7D–F). In addition, backscattered electron images display consistent color (Figures 7A–C). In this study, we analyze the compositions of olivines to investigate the nature of the mantle source. Previous studies found some olivines showing “bright-white-rim” in backscattered electron images, indicating composition change from core to rim (Li and Zhang, 2011; Duan et al., 2019). Considering this feature, the cores of olivine contain more information on primary melts, and the compositions of cores in olivines can be used to study the nature of mantle sources (Zhang et al., 2016). Therefore, we only analyze the composition of cores in olivine.

During magma ascent in continental crust, crustal contamination usually affects the elemental and isotopic compositions of continental intraplate volcanic rocks (Li et al., 2016). However, the Tengchong basalts do not display a positive correlation between ⁸⁷Sr/⁸⁶Sr and SiO₂ (Figure 6B), suggesting that basalts have not been significantly modified by crustal contamination. Previous studies have also found negligible effects of crustal contamination on the Tengchong basalts (Fan et al., 1999; Zou et al., 2017).

Because the Mg# values range from 50.3 to 67.7, lower than the values of primary magma (Mg# values of 68–75; Frey et al., 1978), the Tengchong magmas experienced evolution. During magma evolution, olivine, clinopyroxene, plagioclase or Fe-Ti oxides may have developed from fractional crystallization in basalts. Olivine crystallization can be verified by two pieces of evidence. First, SiO₂ and Al₂O₃ contents increase and Ni contents decrease with decreasing MgO (Figures 4A, B, H), indicating fractional crystallization of olivine (Lee et al., 2021). Second, constant Fe₂O₃ ^T with varying MgO confirms an olivine-controlled fraction in basalts (Zhang et al., 2016) (Figure 4C). The fractional crystallization of the other three minerals is insignificant based on three pieces of evidence. First, crystallization of clinopyroxene is negligible according to nearly constant CaO with decreasing MgO (Figure 4E). Second, crystallization of plagioclase is limited because the positive correlation between Al₂O₃ and MgO contents is absent, and most basalts have Euʹ values >0.88 (Supplementary Table S1 in Supplementary Information (S1)) (Euʹ values ˃0.88 represent little plagioclase fractionation in the Tengchong volcanic field, as suggested by Zou et al. (2017)). Third, Fe₂O₃ and TiO₂ contents do not decrease with decreasing MgO (Figures 4C, D), suggesting little fractional crystallization of Fe-Ti oxides in the Tengchong basalts.

Basalts from the central area show low MgO and Ni contents, suggesting that they are relatively evolved. However, we consider that the olivines in the central area can reflect the information of primary magmas based on three pieces of evidence. First, the Fo values of olivine in the central area range from 80 to 84, which are relatively close to the values of primitive olivine (88–92; Foley et al., 2013). Second, previous studies have found that ³He/⁴He ratios in olivines from the central area are in the range of MORB, indicating that magmas were generated from asthenospheric melts (Chen et al., 2022). Third, melt inclusions from the central area show lower Th/Yb and higher Nb/U ratios than other melt inclusions from the western and central areas (Supplementary Figure S4), indicating that magmas were generated from a primary mantle source.

According to previous studies, olivines generated from fractional crystallization generally exhibit a trend of decreasing Ni content with decreasing Fo, and mafic magmas that suffer sulfide saturation display sequestration of Ni at the same Fo (e.g., Herzberg, 2011). As shown in Figure 10, olivines from the central area are characterized by lower Ni content than those from the western and the eastern areas at the same Fo. Although this phenomenon can be caused by sulfide saturation, sulfide saturation is negligible. Microscope observations from previous (Zhou et al., 2012) and our studies have not found sulfide in volcanic rocks, suggesting very low amounts of sulfide in magmas. This large variation in Ni contents at the same Fo is likely caused by fractional crystallization. The partition coefficient of Ni is high in silicate melt (Rollision, 1993; Li and Audetat, 2012), and silicate melt is able to affect the Ni content under different melting conditions (Niu et al., 2011). Thus, we consider that the Ni content in olivine can represent the melting conditions and was not affected by sulfide saturation. We conclude that the basaltic elemental and isotopic compositions indicate the nature of the mantle source.

It is necessary to investigate the asthenosphere–lithosphere interaction and source lithology for Tengchong magma source before studying the variation in lithospheric thickness. Except for the lithospheric thickness, the lithospheric composition and petrological properties of mantle sources may also affect the chemical composition of basalt and olivine.

Several studies have revealed that the lithospheric mantle and its interaction with the asthenosphere affect the origin of lavas in the Tengchong volcanic field (Zhao and Fan, 2010; Chen et al., 2022). To investigate the contributions of lithospheric mantle and asthenosphere to the magma, we conduct a two-component mixing model in terms of Sr-Nd isotopes. Two endmembers, metasomatized subcontinental lithospheric mantle (SCLM) and enriched asthenospheric mantle, are proposed to form the primary components of the Tengchong volcanic rocks (Chen et al., 2022). According to the mixing model shown in Figure 6, most of the Tengchong basalts plot around the mixing lines and reflect less than 15% of the SCLM contribution (Figure 6). This phenomenon indicates that asthenospheric melts play a dominant role in forming the magmas. Parameters, such as SiO₂ and TiO₂, Na₂O contents and (La/Sm)_N, Hf/Lu and Ba/Zr ratios, are used to study the lithosphere thickness. Generally, these parameters in the metasomatized SCLM and enriched asthenospheric mantle exhibit small differences (Supplementary Table S4). Combining the contribution proportion and compositions of two endmembers, the parameters used in the following sections can be used to analyze the nature of the asthenosphere and can be further applied to infer the lithosphere thickness.

In general, basaltic magmas are formed from the melting of peridotite and pyroxenite components. The pyroxenite component is considered to be related to recycled crustal materials from subducted slab (Sobolev et al., 2007). The element (Mn, Ni, and Ca) contents and ratios (Mn/Fe, Ni/Mg, Ca/Fe and Ni/(Mg/Fe)/1,000) in olivine phenocrysts are used to distinguish the melting of pyroxenite or peridotite (Sobolev et al., 2007; Straub et al., 2008; Rasmussen et al., 2020). Two reasons indicate that the Tengchong basaltic magmas are from the partial melting of a peridotite source. First, Supplementary Figure S2 represents the chemical compositions of olivines in the Tengchong basalts and basalts from other representative zones (Herzberg, 2011). The olivines from the Tengchong basalts display low Ni contents and Fe/Mn ratios and relatively high Mn contents. Their compositions are similar to those of olivines from the Kamchatka and Central American regions, where magmas are from peridotite sources (Ruprecht and Plank 2013; Li et al., 2020). Thus, the Tengchong volcanic rocks may be have been generated from a peridotite source. Second, Mn/Fe does not rely on the process of olivine fractionation, but is mainly controlled by mantle source rocks (Sobolev et al., 2007). Olivines generated from peridotite are generally characterized by 100×Mn/Fe with values higher than 1.4, which is distinct from the values of pyroxenite ranging from 1.1 to 1.3 (Sobolev et al., 2007). Specifically, olivines from Tengchong basalts show that the 100×Mn/Fe values are mainly larger than 1.4, suggesting the dominant contribution from the peridotite source. In the diagram of Ni/(Mg/Fe)/1,000 versus 100×Mn/Fe, the samples mostly plot close to and overlap with those of olivine crystallized from peridotite-derived melts (Supplementary Figure S3A). In Supplementary Figures S3B, C, the 100×Ni/Mg and 100×Ca/Fe ratios are lower than the peridotitic values owing to fractional crystallization (Kim et al., 2021). Based on the features of olivine compositions, Tengchong basaltic magmas are formed from the partial melting of a peridotite source.

Variations in lithospheric thickness can be investigated by the major and trace elements of basalt and olivine (Niu et al., 2011; Gale et al., 2014; Zhang et al., 2016; Liu et al., 2016; Matzen et al., 2017). Based on the estimated primary magmas and measured olivine elemental compositions, we consider that the lithospheric thickness is greater in the western and eastern areas than in the central area of the Tengchong volcanic field.

Primary magmas and olivines can reflect lithospheric thickness. The primary magmas were estimated based on the addition of olivine proposed by Tamura et al. (2014). The composition of the estimated primary magmas of the Tengchong volcanic rocks and the olivine abundance added to the samples are shown in Supplementary Table S2 in the Supplementary Information (S1). Based on the elements and isotopes of primary magmas and olivines shown in Figures 8–9, the lithosphere in the western and eastern areas is probably thicker than that in the central area. This conclusion is drawn according to the following five pieces of evidence.

First, SiO₂ is sensitive to the pressure of melting, and Na₂O and TiO₂ are sensitive to the extent of melting (Langmuir et al., 1992; Gale et al., 2014). Therefore, a thick lithosphere leads to a decrease in SiO₂ and TiO₂ contents and an increase in Na₂O content (Langmuir et al., 1992; Gale et al., 2014). Primary magmas from central to western and eastern areas show a rough decrease in SiO₂ and TiO₂, and an increase in Na₂O (Figures 8A–C). Compared with the SiO₂ content, the correlation is more evident for the TiO₂ and Na₂O contents. These phenomena may suggest a change in lithospheric thickness under the Tengchong volcanic field.

Second, because the Tengchong basaltic magmas formed from the partial melting of a peridotite source, the trace element contents are mainly controlled by the degree of partial melting. As lithospheric thickness is inversely proportional to the degree of partial melting, and (La/Sm)_N, Hf/Lu and Ba/Zr ratios increase with decreasing degree of partial melting (Sun et al., 2017; Guo et al., 2020), these trace element ratios are adopted to investigate the change in lithospheric thickness. The basalts from the western and eastern areas show higher ratios of (La/Sm)_N, Hf/Lu and Ba/Zr than those from the central area (Figures 8D–F), suggesting thicker lithosphere in the western and eastern areas.

Third, the Sr-He isotopes can partially provide information on lithospheric thickness. Previous studies suggest that the metasomatized SCLM is characterized by high ⁸⁷Sr/⁸⁶Sr and low ³He/⁴He ratios, and that enriched asthenospheric mantle features relatively low ⁸⁷Sr/⁸⁶Sr and high ³He/⁴He ratios in the Tengchong volcanic field (Chen et al., 2022). In Figures 9A,B, basalts from the western and eastern areas show lower ³He/⁴He and higher ⁸⁷Sr/⁸⁶Sr ratios than those from the central area. Basalts from the western and eastern areas display more lithospheric signatures, and those from the central area have more asthenospheric signatures, which can be interpreted as thicker lithosphere in the western and eastern areas.

Fourth, elements in olivine can also be applied to study the variation in lithosphere thickness. According to Niu’s model (Niu et al., 2011), the composition of minor elements in olivine relies on the variation in lithosphere thickness, which affects the melting process. The lithosphere thickness affects the final depth of melt equilibration, determining the D^olivine/melt, such as the parameter Kd_Ni ^ol/melt (Niu et al., 2011; Zhang et al., 2016). Several practical studies have applied the Ni content in olivine to investigate the lithospheric thickness (e.g., Walter, 1998; Humphreys and Niu, 2009; Zhang et al., 2016). For the same Fo value, olivine generated under thick lithosphere often exhibits a high Ni concentration. Figure 10 shows that olivines from the western and eastern areas have higher Ni contents than those from the central area, indicating that the thickness of the lithosphere in the western and eastern areas is greater than that in the central area.

Fifth, the study of receiver functions provides direct evidence for the depth of the lithosphere-asthenosphere boundary (LAB) beneath the Tengchong volcanic field. The imaging results show that the depth of the LAB in the central area is 80 km and that in the western and eastern areas is 90 km (Hu et al., 2012; Zhang et al., 2015; Yang et al., 2017).

The basalts in the Tengchong volcanic field were generated from 7.2 Ma to 0.02 Ma (Guo et al., 2015; Zou et al., 2017; Tian et al., 2018). Figure 9 shows clear geochemical variations over the eruption episodes. The basalts that erupted from 2.8 to 7.2 Ma display more lithospheric signatures, and those that formed from 0.02 to 0.6 Ma have more asthenospheric signatures (Chen et al., 2022). The systematic geochemical variations over time provide evidence that the lithosphere become thinning from the late Miocene to the Holocene. The lithospheric thinning can be caused by lithospheric delamination or lithospheric extension. The lithospheric delamination is less likely, based on the geophysical results. High-resolution seismic tomographic results only show low-velocity anomalies in upper mantle beneath the Tengchong volcanic field, and do not provide any evidence for lithospheric delamination (Lei et al., 2009; Huang et al., 2019; Yao et al., 2021).

In the Tengchong volcanic field, lithospheric extension plays an important role in passive asthenospheric upwelling underground (Chen et al., 2002; Zhou et al., 2012; Guo et al., 2015). During the early stage, the magmas were produced by small degrees of partial melting at high pressures. During the late stage, the lithosphere became thinner, and magmas were produced by large degrees of partial melting at low pressures. The basalts became less alkaline and radiogenic along with lithospheric thinning in the Tengchong volcanic field. This process is similar to extension-related continental volcanism in other areas, where tectonism influences mantle melting and asthenosphere–lithosphere interactions, such as the Basin and Range province and Rio Grande Rift in the United States of America and the Ethiopian volcanic province in Africa (Hart et al., 1989; Thompson and Gibson, 1994; Depaolo and Daley, 2000).

Although lithospheric extension occurred in the Tengchong volcanic field, normal faults did not develop in this area (Wang et al., 2007; Wang et al., 2008). The lack of a normal fault may imply decoupling between the crust and lithospheric mantle. The extension of the lithosphere usually results in normal faults; for example, in the Basin and Range province and Ethiopian volcanic province, normal faults spread along the rift (Hart et al., 1989; Depaolo and Daley, 2000). However, the Tengchong volcanic field is dominated by strike-slip faults (Wang et al., 2007; Wang et al., 2008). This tectonic phenomenon probably indicates that a compressional regime is dominant in the crust. Therefore, the deformation mechanisms of the crust and lithospheric mantle are different.

Deformation is complex in SE margin of the Tibetan Plateau. It is generally accepted that the lithosphere experienced vertically coherent deformation north of ∼26°N, that is curst and lithospheric mantle undergo similar deformation process (Hu et al., 2012; Huang et al., 2015; Gao et al., 2020). However, lithospheric deformation south of ∼26°N is intensely debated (Wang et al., 2008; Huang et al., 2015; Shen et al., 2022). From the observation that the fast polarization directions (FPDs) of the XKS are consistent with the orientations of maximum extension in the crust and earthquake focal mechanism solutions, Huang et al. (2015) considered that the crust and lithospheric mantle suffered coherent deformation. That is crust and lithospheric mantle both experienced E–W extension. In addition, based on nearly the same azimuthal anisotropic distributions in the lower crust and uppermost mantle, Shen et al. (2022) inferred consistent deformation of the lithosphere. This means that the crust and lithospheric mantle experienced similar deformation. Many researchers consider the decoupling between the crust and lithospheric mantle south of ∼26°N based on the evidence that the directions of the GPS velocity field are generally perpendicular to the FPDs (Flesch et al., 2005; Lev et al., 2006; Sol et al., 2007; Hu et al., 2012; Gao et al., 2020) (Figure 11A). GPS velocity reveals that the crust south of ∼26°N undergoes NNE–NE-oriented deformation, while the FPDs (the parameter FPD mainly reflects the deformation of lithosphere) are generally WNW‒NW, and imply that lithospheric mantle south of ∼26°N must undergo WNW‒NW-oriented extension (Gao et al., 2020). Only this pattern of lithospheric deformation can lead to the observed FPDs. The difference between the directions of GPS velocity and FPDs indicates the different deformation patterns in shallow and deep lithosphere.

By integrating previous studies and this study, we constructed the dynamic model shown in Figure 11B to interpret the lithospheric deformation in the Tengchong field. The extrusion of crustal materials from the Tibetan Plateau (Gan et al., 2007) resulted in a compressional regime in SE margin of the Tibetan Plateau. In the Tengchong volcanic field, the crust is under NNE‒NE-oriented compression (Wang et al., 2007). The lithospheric mantle undergoes WNW‒NW-oriented extension (Wang et al., 2007), which results in the different thicknesses of the lithosphere in the Tengchong volcanic field revealed by our study. We consider that the extension is caused by corner flow in the mantle wedge because of the rollback of the subducting Indian slab in the Burma subduction zone (Hu et al., 2012; Lee et al., 2016). The rollback of the subducting Indian slab is also suggested by Shapiro et al. (2008) based on the mantle structure from seismic tomography.