To investigate the direct evidence for a number of physico-chemical processes related to pluton construction and growth, we examine the Buya pluton of West Kunlun in Northwestern China, which emplaced within the 455–460 Ma time frame. Field observations, geochemical data, and thermodynamic modeling show that mafic dikes of the Buya pluton were conduits for magma chamber replenishment during pluton construction. These mafic inputs, and the enclaves that resulted from them, induced compaction of the semi-consolidated, crystal-rich, felsic mushes below them. The accumulation of highly silicic, fine-grained granite at the top of the Buya pluton is the result of episodic melt segregation events from these mushes. This sequence of events may reflect a common process that promotes compositional variation in granite suites. Combined geochemical and Hf- and Nd-isotopic data suggest that parental magmas of the mafic sheet and enclave are similar to sanukitoid, which is potentially consistent with a mantle peridotitic source metasomatized by slab melts. These mafic magmas intruded the lower crust where the original magma was modified by mafic lower-crust melt. Following emplacement at shallow crustal levels of the mafic inputs (~3.7 kbar, ~5.3 km, constrained by amphibole geobarometry), the felsic mush evolved through the extraction of interstitial melts driven by hybridization with episodic inputs of mafic magmas as well as crystal consequent accumulation and fractional crystallization of plagioclase, hornblende, and accessory phases such as allanite, apatite, and zircon. This fractional crystallization process may also provide an explanation for the apparently high Sr/Y features in some silicic high-K, calc-alkaline magmas.

Understanding the magmatic processes through which felsic plutons are constructed is fundamental to the production and differentiation of continental crust at convergent plate boundaries. It is widely recognized that the formation and growth of plutons result from crustal-scale magmatic systems (transcrustal magmatic systems [TCMS]; Cashman et al., 2017) that are controlled by injection processes (incremental growth and magma flow), magma chamber processes (hydrodynamic processes and crystal-melt separation), and tectonic processes (Richards, 2003; Annen et al., 2006, 2015; Barbey, 2009; de Saint Blanquat et al., 2011). The replenishment of dense mafic magmas can lead to an interplay or magma mingling and local mixing (Hildreth, 1981; Huppert and Sparks, 1988; Streck and Grunder, 1999; Sparks and Sigurdsson, 1977; Barnes et al., 1986; Metcalf et al., 1995; Jellinek and Kerr, 1999; Bergantz et al., 2015; Wiebe and Hawkins, 2015), incremental assembly (Glazner et al., 2004; Coleman et al., 2004; Miller et al., 2011), crystal (mush) resorption, magma convection and homogenization, and crystal-liquid segregation in the magma reservoirs (e.g., DePaolo, 1981; Wiebe and Collins, 1998; Bachl et al., 2001; Miller and Miller, 2002; Weinberg, 2006; Collins et al., 2006; Walker et al., 2007; Pupier et al., 2008; Deering and Bachmann, 2010; Turnbull et al., 2010; Gelman et al., 2014; Lee and Morton, 2015; Barnes et al., 2016; Schaen et al., 2017; Bachmann and Huber, 2019). Discussing the relative timing of intrusion and geometry of mafic magmatic inclusions/dikes/sheets hosted in granites (Kumar, 2020) is important. Many of these processes may individually play a dominant role in specific plutons, parts of plutons, or periods of plutonic growth (Pupier et al., 2008). Yet, these processes, and how they operate in governing the growth of plutons and chemical variation in granitic magmas, either at the level of emplacement or deeper, are still debated (e.g., Hildreth and Moorbath, 1988; Annen et al., 2006; Clemens and Stevens, 2012; Glazner, 2014; Lee and Morton, 2015; Barnes et al., 2016). Further studies are needed to understand the connection between these complex magmatic physical processes and the bulk-rock compositional evolution of growing granitic plutons.

To contribute to this topic, we examine a well-exposed felsic pluton located in the southwestern margin of the Tarim craton of Northwestern China (Fig. 1), which provides excellent exposures of interlayered mafic, hybrid, and felsic magmatic rocks. We combine detailed field relationships, zircon U-Pb-Hf isotopic data, whole-rock geochemistry, and thermodynamic modeling to elucidate the processes through which the pluton was constructed and evolved driven by intruded mafic magmas. This contribution provides a detailed understanding of how mafic magma replenishment, magma hybridization, and subsequent geochemical differentiation played a significant role in individual plutons and their growth in a convergent margin setting.

The West Kunlun study area lies along the northwestern margin of the Tibetan plateau (Fig. 1A) and can be divided into four main tectonic units. From north to south, these include: the Tiekelike belt, the West Kunlun belt, the Taxkogan–Tianshuihai terrane, and the Karakorum–Qiangtang terrane (Fig. 1B; Wang et al., 2016). The Tiekelike belt, previously named the Northern Kunlun belt, comprises Precambrian amphibolite facies gneisses, schists, migmatites, and sedimentary and volcanic rocks, which represent the basement of the Tarim block (Wang et al., 2015). The West Kunlun belt constitutes the complex continental Eurasia–India accretionary orogenic belt (Matte et al., 1996; Xiao et al., 2005). The Taxkogan–Tianshuihai terrane is considered to be a large accretionary prism that formed due to northward subduction of the Paleo-Tethys Ocean (Xiao et al., 2005, and references therein). The Karakorum–Qiangtang terrane has been in the continental interior since the closure of the Tethyan Ocean during the Early Cretaceous (Dewey et al., 1988). Voluminous granitoids record a period of magmatism and tectonic activity that spans the early Paleozoic to early Mesozoic, and they are exposed along several NW–SE-trending faults that extend from the Tiekelike belt to the Taxkogan–Tianshuihai terrane (Fig. 1B). Early Paleozoic felsic magmatism occurs within an ~1000-km-long belt parallel to the West Kunlun belt itself. Here, magmatism has been dated at 530–500 Ma, 470–468 Ma, 450–430 Ma, and 408–404 Ma (Fig. 1B), and it is suggested that the formation of these granitoids was closely related to the subduction of Proto-Tethys oceanic lithosphere (e.g., Mattern and Schneider, 2000; Wang and Fang, 1987; Zhang and Xie, 1989; Xu et al., 1994; Pan, 1996; Jiang et al., 2002, 2013; Yuan et al., 1999, 2003; Xiao et al., 2005; Wang et al., 2013, 2016).

The Buya pluton forms a 3.5-km-long axis within an ~0.5-km-wide, E–W strip intrusion within Neoproterozoic quartz schist of the Ailiankate Group (Fig. 1C) in the southeast of the Taxkogan–Tianshuihai terrane, ~100 km south of Hetian City (Fig. 1B). This pluton is a relatively small but typical example of plutons of the early Paleozoic granitic suites exposed along the southwestern margin of the Tarim Craton and shows high Ba and Sr concentrations typical to the appinite-granite suite of that region (e.g., Ye et al., 2008). Li et al. (2007) reported a sensitive high-resolution ion microprobe (SHRIMP) zircon U-Pb age of 459 ± 23 Ma for porphyritic granite of the Buya pluton, and Ye et el. (2008) obtained an age of ca. 430 ± 12 Ma for alkali-feldspar granite using the same method. Ye et al. (2008) suggested that the Buya pluton represents the final stage of an early Paleozoic crustal thickening event, following terrane accretion onto the southern Tarim craton, and the beginning of a post-orogenic collapse phase in the Paleozoic West Kunlun orogenic belt.

The Buya pluton is well exposed and easily accessible for detailed observation and sampling along the Yulongkashi River (Fig. 1C). Fine-grained granite (FG, the grain size varies from 0.1 mm to 2 mm) forms much of the eastern part of the pluton and a narrow band on the west, and it likely forms the outer zone of the entire pluton (Fig. 1D). Contacts between FG and the country rocks are typically sharp, and xenoliths of quartz schist occur in FG near the contact (Fig. 2A). Dark gray, aplitic dikes, around ~1 m thick, intrude both the FG and quartz schist (Fig. 2B). Scarce, small mafic enclaves are found within FG (Fig. 2C). The sample of FG shows evidence of solid-state recrystallization (Fig. 3A). It contains K-feldspar (40 vol%), plagioclase (15 vol%), quartz (40 vol%), biotite (5 vol%), and accessory apatite, zircon, allanite, and Fe-Ti oxides (Fig. 3A).

The FG zone grades inward to medium-to coarse-grained porphyritic quartz syenite (PQS; Fig. 1D). In the western part of the pluton, the transition between FG and PQS is narrow (~100 m; Fig. 1D). However, in the east, FG shows a gradual westward transition into medium- and then coarse-grained PQS with a progressive increase in both size and abundance of K-feldspar phenocrysts. The PQS is the predominant rock type in the central and western parts of the Buya pluton and is in contact with coarse K-feldspar porphyritic granodiorite (GD) in the west (Fig. 1D). A notable characteristic of PQS is its abundant cargo of mafic enclaves (MEs; Fig. 2D) of various sizes and textures. It is also cut by fine-grained granitic veins and mafic dikes and schlieren. The PQS is characterized by abundant K-feldspar megacrysts. These typically form crystal mush zones with 30–70% densely packed euhedral to subhedral megacrysts of up to 4 cm in length and enclose earlier crystallizing plagioclase, quartz, and biotite. This PQS also contains plagioclase (5–15 vol%), subhedral hornblende (5–15 vol%), anhedral quartz (5–15 vol%), clinopyroxene (1–2 vol%), biotite (2 vol%), and accessory apatite, zircon, allanite, and titanite (Fig. 3B). The GD is locally intermingled with PQS and has quite a mesocratic, mafic mineralogy with a color index of ~40; it includes abundant amphibole and biotite and contains frequent MEs (Fig. 2E). Large, subhedral to euhedral, white plagioclase crystal is mantled by pink alkali feldspar, and alkali feldspar is mantled by plagioclase, which suggests the possibility of magma mixing (Fig. 2E).

Close to the western GD–PQS transition, a prominent sheet of mafic rocks occurs within the PQS (Fig. 2F). This sheet is ~10 m thick and east-dipping. The mafic rock is typically of monzodioritic–dioritic composition, and the sheet is locally situated against the overlying PQS granite unit. The mafic sheet does not contain K-feldspar megacrysts. However, the associated MEs have abundant K-feldspar megacrysts similar to those of PQS. Mafic dikes also cut PQS and represent the last pulses of mafic magma intruding the pluton (Fig. 2G). These mafic dikes typically contain slightly more K-feldspar and less hornblende, plagioclase, and biotite than the MEs, but they also contain up to 20% euhedral to subhedral clinopyroxene. In addition, K-feldspar forms large poikilitic grains enclosing clinopyroxene and hornblende (Fig. 3C). The dikes also locally contain pockets with high concentrations of K-feldspar megacrysts that are interpreted to have been incorporated from the PQS host rock. Both the MEs and the dike rocks contain accessory titanite, apatite, ilmenite, magnetite, and zircon. Although the late mafic dikes contain a slightly more anhydrous mineralogy, all of the mafic rocks show broad petrographic similarities. Sub-horizontal channels of schlieren (likely pipes in three-dimensional space, e.g., Weinberg et al., 2001), up to tens of meters wide, cut the PQS and are truncated by late granite veins (Fig. 2H). Mafic schlieren are also locally associated with sinuous enclave swarms (likely disaggregating dikes) that are parallel to mafic dikes. The intrusive contact between the schlieren and PQS, and the alignment of the schlieren themselves, is commonly discordant to the flow-alignment of feldspar in the host PQS (Fig. 2H). The foliation is highly oblique to the aligned enclaves and schlieren, and it is probable that more than one strain field is recorded. There are late granitic veins cutting MEs (Fig. 2I).

The presence of mafic enclaves is a distinctive feature of PQS. The MEs show a wide range in size of up to 0.5 m. They occur as large, isolated masses or swarms of smaller (0.5–50 cm in size) MEs. The MEs consist mainly of a fine-to medium-grained assemblage of K-feldspar (up to 35 vol%), euhedral–subhedral hornblende (up to 30 vol%), subhedral plagioclase (up to 25 vol%), biotite (~5 vol%), and quartz (~5 vol%). Xenocrysts of quartz mantled by hornblende and biotite show the characteristics of ocelli texture (Fig. 3D). ME-rich, tongue-like bodies also occur (Fig. 5A). MEs range from angular, rounded to oval, or flattened in shape (Figs. 4A4C). Some are composite MEs (i.e., enclaves in enclaves; Fig. 4D) that reflect several generations of mafic magma. MEs in high-density enclave swarms are also sometimes associated with K-feldspar, crystal-rich “clusters,” and both the MEs and the K-feldspar clusters show a preferred alignment that is consistent with movement of an inhomogeneous layer of semi-consolidated mafic and crystal-rich felsic magmas (Fig. 4A). In addition, within these swarms, there appears to be a decrease in the size and abundance of enclaves toward the northeast (Fig. 1D). A magmatic foliation in host PQS granite mush in the vicinity of one mafic dike defines arc-like patterns within a lobe pattern (Fig. 4E).

Four samples—18YL-1 (PQS), 07HT-38 (PQS), 07HT-39 (ME), and (12TK-16 (FG)—were collected from the Buya pluton for geochronological analysis. Zircons from samples 07HT-38 and 07HT-39 were collected for Lu-Hf isotope analyses. All analytical methods and full data sets are presented in Item A and Tables S1–S2 in the Supplemental Material.1

Zircon U-Pb Geochronology

Zircon grains from PQS (samples 18YL-1 and 07HT-38) are subhedral and dark in cathodoluminescence (CL) images (Fig. S1, footnote 1). Some grains consist of unzoned, resorbed, bright cores and wide overgrowths with dark, fine, magmatic oscillatory zoning. Twelve spots of sample 18YL-1 were analyzed. Two cores give Mesoproterozoic to Neoproterozoic ages (751 Ma and 1151 Ma; Fig. 5A), which suggests an inherited component. Five of seven spots from magmatic zircons yielded younger concordant results that correspond to a 206Pb/238U weighted mean age of 457 ± 11 Ma (Fig. 5B; MSWD = 3.4, 1σ; Fig. 5B). Two of seven spots give older 206Pb/238U ages of 475 Ma and 469 Ma (possible antecrysts and/or xenocrysts) and were not taken into account. The remaining three analyses were discordant and may relate to lead loss. Ages from the two inherited cores also match specific age components of the detrital zircon age spectrum from Tiekelike metasedimentary rocks (Wang et al., 2015). Thirteen of 18 spots of sample 07HT-38 yielded concordant results that correspond to a 206Pb/238U weighted mean age of 456 ± 6 Ma (Fig. 5C; MSWD = 1, 1σ). There are five data that are discordant and were not considered in weighted mean age calculation. The ca. 456–457 Ma crystallization age is consistent with that determined by Li et al. (2007).

The analyses from high-U zones of sample 07HT-39 (ME) show variable discordance, and seven analyses yielded concordant 206Pb/238U ages from 439 Ma to 465 Ma (Fig. 5D), which defines an imprecise crystallization age for the enclave, but they have more or less the same concordant ages as the host PQS (Fig. 5C). The discordant data in MEs could be the result of common lead that cannot be precisely corrected without measuring 204Pb (e.g., Wu et al., 2020). From the CL, most of the zircons of MEs are subhedral crystals with disturbed inherited cores and are similar to those from the PQS (Fig. S1). Many K-feldspar megacrysts are found in the MEs (Fig. 2D), which provide evidence of the physical mixing of mafic and felsic magmas. It is likely that some zircons in MEs may be xenocrystic from the host PQS. CL images show that the majority of zircon grains from the FG (Sample 12TK-16) are dark and prismatic with a banded structure (Fig. S1). Some grains consist of unzoned, resorbed, and bright cores, with dark overgrowths that show weak magmatic oscillatory zoning. Analyses of three of four cores yielded concordant Neoproterozoic–Mesoproterozoic ages (207Pb/206Pb age of 577 Ma, 1112 Ma, and 1185 Ma), which suggests an inherited component (Fig. 5E). The remaining spot analyses are discordant, reflecting partial Pb loss. On the Concordia diagram (Fig. 5F), 13 analyses fit a discordia line with a lower Concordia intercept 207Pb/206Pb age of 452 ± 3 Ma (MSWD = 0.27), which is interpreted as the time of FG crystallization.

Zircon Hf Isotopic Composition

Magmatic zircons from PQS sample 07HT-38 with a 206Pb/238U age of 456 ± 6 Ma have 176Hf/177Hf ratios of 0.282200–0.282339 and εHf (456) values of –10.4 to –5.4 (Fig. S2, footnote 1). Two-stage crustal model ages (TDM2; see details in Table S2) range from 1520 Ma to 1769 Ma (Table S2). The main group of zircons from the ME sample (07HT-39) with a 206Pb/238U age of ca. 452 Ma shows a large range of 176Hf/177Hf ratios from 0.282219 to 0.282289 with εHf (452) values ranging from –7.4 to –10.0 (Figs. S2–S3). They yield two-stage crustal model ages of 1615–1748 Ma (Table S2).

Twenty-three samples taken from the Buya pluton were analyzed for major and trace elements (Item A and Table S3, footnote 1). Analytical methods and analytical data are given in Item A. Whole-rock major and trace element data are plotted on Harker variation diagrams in Figures 67.

Mafic Sheet, Dike, and Enclave Units

These units are characterized by rocks with 53.2–61.6 wt% SiO2 (Table S3). In a total alkali versus silica (TAS) diagram, they collectively define an alkaline trend (Fig. 6) from monzodiorite to monzonite compositions. In the K2O versus SiO2 diagram (Fig. 6), they largely plot in the shoshonitic field, with two of the lower silica samples plotting in the high-K field. Compared with the other rocks of the Buya pluton, the mafic rocks have high MgO (and Mg#), CaO, and P2O5 but similar or slightly lower Na2O and Al2O3 and similar or slightly higher TiO2. High MgO contents correlate with high Cr and Ni concentrations, but notably, the mafic rocks also have concentrations of Rb, Th, Zr, Nb, and rare earth elements (REEs) higher than those of all other rocks of the Buya pluton except for a few samples of PQS and GD (Figs. 7 and 8). Weak, negative Eu anomalies are developed in the mafic samples and are accompanied by negative anomalies in normalized abundances of Sr and Ti (Fig. 8).

The close compositional similarity among the various mafic groups (MEs, dikes, and sheets) provides strong evidence that they originated from the same, or similar, relatively primitive, trace element–enriched magma source (i.e., they are possibly comagmatic and cogenetic). The combination of high MgO (>6 wt%), Mg# (>60), Cr (>100 ppm), and Ni (mostly >60 ppm), with high Sr and Ba (each >500 ppm), K2O (up to 6.9 wt%), Rb, REE, and La/Yb, is characteristic of sanukitoids, as defined by Martin et al. (2005). Some MEs also exhibit geochemical features similar to those of high Nb basalts (Nb>20 ppm; Table S3).

Granodiorite (GD) and Medium-to Coarse-Grained Porphyritic Quartz Syenite (PQS) Units

These rocks have SiO2 content that ranges from 60.0 wt% to 67.6 wt%. In the TAS diagram, they mainly plot in the quartz monzonite field, and in the K2O versus SiO2 diagram (Fig. 6), they plot mainly in the high-K field. They are the most sodic rocks of the pluton (Na2O>4.5 wt%) and generally plot in the compositional gap that separates the mafic and strongly felsic rocks from all major and trace elements except for Al2O3, Na2O, and Zr, which are enriched in GD and PQS (Figs. 6 and 7). The three GD samples separate the PQS and ME in terms of silica content. The normalized trace element patterns for GD and PQS are also very similar to those of the mafic rocks, except for slightly more subdued anomalies for Sr, Zr, and Eu (Fig. 8). Ye et al. (2008) show that the Nd-isotopic composition of the PQS (εNd(T) of −8.4 to −10.4) is less radiogenic than that of the MEs (εNd(T) of −5.7 to −6.7). Estimated pressures of low-Altot amphibole crystallization have been calculated using the newly calibrated Al-in-amphibole barometer (applicable to granitic rocks) of Mutch et al. (2016). The outermost rims of unaltered amphibole from the PQS yield emplacement pressures of 3.7 kbar, and the corresponding depth range of 12 km (Table S4, footnote 1) reflects a mid-to upper-crustal emplacement level for the Buya Pluton.

Fine-Grained Granite (FG) and Granitic Vein Units

The granite veins are geochemically similar to the FG except for having slightly higher K2O and slightly lower Sr and Ba concentrations (Figs. 67). Compared with PQS, the granite veins and FG are typically depleted in REE, Th, Nb, and Zr (Fig. 8) with high Sr/Y and La/Yb (Fig. 7).

Magmatic Processes and Compositional Variation Related to Pluton Construction

The magmatic processes related to construction and modification of the Buya pluton can be interpreted from the structures and textures preserved in the rocks as well as from petrographic evidence. The fundamental control, or driver, of many of these processes appears to be mafic magma influx. This influx appears to have exerted both a direct (e.g., through magma volume added or mechanical or chemical mixing) and indirect influence (e.g., through compaction of granitic crystal mushes, melt segregation, heating, etc.) on compositional evolution.

Multiple Mafic Injections, Magmatic Flow, and Magma Mingling and Mixing

Several lines of textural evidence indicate that the mafic magmas forming the ME, mafic sheet, and some of the mafic dikes intruded semi-crystallized, felsic magma within the Buya pluton. The presence of angular MEs within large, rounded, pillow-like MEs, and of other multiple MEs (Figs. 4C4D), indicates that injection of mafic magmas into semi-crystallized, felsic crystal-mush occurred repeatedly (Barbarin, 2005; Paterson et al., 2016). The distinct variations of major and trace element compositions between the MEs, mafic sheet, and dikes (Figs. 67) may represent periodic mafic replenishment of the Buya mush. Thus, mafic injection into the Buya pluton, or contribution to pluton fill, was a dynamic multi-stage process. MEs locally contain pockets with high concentrations of K-feldspar megacrysts that are interpreted to have been incorporated from the PQS host rock. Thus, even at this late stage, pockets of PQS remained in a crystal-mush state, and the mafic dikes could be synplutonic dikes (e.g., Kumar, 2020). The amount and style of disruption of mafic magmas will depend on the specific mechanical properties of the mafic and felsic magmas, on the amount of added heat, and the strain rate (e.g., Andrews and Manga, 2014).

The tongue-like bodies comprising ME aggregates could be interpreted to have formed as higher density mafic magma rested on a granitic mush (Fig. 4A), which would indicate movement of crystal- and/or ME-laden magmas through mushy PQS. Magmatic foliation in host PQS granite mush defines different arc-like patterns, which also represent flow of the host magma mush produced by individual mafic replenishing magmas “rolled” across the chamber (Fig. 4E) and indicate that these megacrysts rotated within a liquid medium during magmatic flow without deforming plastically. The orientation of the mafic schlieren also possibly reflects local flow close to the interfaces between individual intraplutonic magma pulses on a magma chamber floor or intra-chamber layer (e.g., Fig. 2H; Weinberg et al., 2001; Vernon and Paterson, 2008).

The GD provides at least local evidence that dynamic processes within the Buya pluton have resulted in mechanical mixing and magma hybridization. This evidence includes the presence of mafic clots and rimmed alkali feldspar (anti-rapakivi structure), rimmed plagioclase (rapakivi structure), or mantled quartz grains (ocelli) (e.g., Vernon and Collins, 2011). The different crystallization histories suggest that the crystals formed in different parts of the magma chamber, or by mixing in separate magma batches, and were mechanically accumulated within GD. On all plots, GD lies in the lower end of the silica range of the PQS, separating PQS from ME along consistent trends that generally conform to linear mixing arrays and highlight the direct control mafic magma input has had on compositional evolution.

Given the observation that mafic magma influx was punctuated throughout the plutonic history, an additional influence that this influx likely had was in rejuvenating earlier partially to largely crystallized felsic crystal mushes. Large, corroded crystals of K-feldspar and quartz are within fine-grained, igneous-textured MEs (Fig. 2E). Also, zircons from MEs were xenocrysts that formed at an early stage in the granitic magma chamber, whereas those from PQS crystallized in magma source with a wide range of variation in Hf isotopic composition (Fig. S2). These reflect physical mingling of resident silicic crystal mush and the chamber-replenishing mafic magma (e.g., Wiebe, 1993). Alternatively (or additionally), they reflect mechanical incorporation (plucking) of crystals from the host granite mush by hybridized syn-magmatic dikes.

Compaction-Driven Crystal Accumulation and Melt Separation in Magma

Major mafic inputs occurred at various stages, and the textural evidence for processes such as magma injection and influx, flow, compaction, and magma redistribution indicate a much more complex magmatic history. Highly clustered K-feldspar in PQS forms a connected framework of coarse euhedral crystals that are indicative of cumulate processes (e.g., Clarke and Clarke, 1998; Weinberg et al., 2001; Paterson et al., 2005; Wiebe et al., 2002; Collins et al., 2006; Vernon and Paterson, 2008; Lee et al., 2015). Compaction of crystal mushes may lead to the segregation of fractionated interstitial melt (Philpotts et al., 1998; Weinberg, 2006; Collins et al., 2006; Bachmann and Bergantz, 2008), which may migrate into adjacent parts of a chamber (e.g., Wiebe and Collins, 1998). Under compaction, buoyant high-silica interstitial melt within the lower felsic layer can migrate upward by porous flow or via dikes through the crystal-rich mush and either pond beneath, or intrude into, the mafic sheet (e.g., Snyder and Tait, 1996; Wiebe and Hawkins, 2015). Also, rapidly crystallizing sheets of underplating magma can undergo second boiling, and gas-driven filter pressing may provide an additional driving force for the segregation of the residual melt (e.g., Sisson and Bacon, 1999). More generally, this felsic fractionated melt can ascend to the effective roof of the magma chamber.

In the Buya pluton, larger mafic enclaves typically have variable shapes and are strongly molded around feldspar megacrysts (Figs. 4B4C). This feature suggests that compaction occurred close to one of the periods when enclave rested on the Buya mush but was still at a stage when enclaves from an earlier event remained soft (partly liquid). Thin, fine-grained granite veins that penetrate the ME, PQS cumulate, and FG (Figs. 2I and 4A) may represent structures through which late felsic melt ascended. In the case of the Buya pluton, these veins are mineralogically and texturally similar to the FG, which suggests that FG was fed by the granitic veins. These form a layer of crystal-poor FG magma in the upper and outer parts of the Buya pluton. Accordingly, we suggest that K-feldspar–rich PQS shows an example of cumulate-textured PQS that was compacted when a dense ME plume settled (Fig. 4B), and the interstitial liquid was consequently evacuated. The episodic filter-pressing and melt segregation from the PQS mush yielded a thick cap of fractionated FG rock, and the observed veins may represent a component of the feeder system for this layer.

We suggest that physical processes, such as the compaction of granitic crystal mushes, also account for significant geochemical evolution, and as described above, are largely driven by mafic magma influx. This indirect influence of mafic magma influx is critical in terms of the compositional evolution of the FG. Textural evidence presented above suggests that some, if not all, FG represents compositionally evolved interstitial liquid extracted from crystal mushes at various stages of pluton construction. A consequence of mush-liquid extraction is distinct compositional gaps in SiO2 range (e.g., Dufek and Bachmann, 2010) as seen between PQS and FG. Over this silica gap, Sr concentrations decrease, and Eu anomalies become more negative, which reflects feldspar fractionation (Fig. 7). The retention of hornblende within the residual melt-depleted PQS mush would explain the trend from PQS to FG to higher Sr/Y and lower Dy/Yb, as well as to more peraluminous compositions (Figs. 67, e.g., Davidson et al., 2007; Reichardt and Weinberg, 2012; Nandedkar et al., 2014). Thus, this process may provide an explanation for the high Sr/Y ratios of FG and perhaps of other, silicic high-K, calc-alkaline magmas and may challenge the interpretation that high Sr/Y ratios in felsic rock invariably reflect high-pressure melting. Systematic depletions in REE, Th, P, and Zr from PQS to FG also suggest that the crystal mush was saturated in accessory phases such as allanite, apatite, and zircon prior to extraction of the FG interstitial melt. It is clear from both field and geochemical data that neither PQS nor FG represents primary liquid (i.e., melt) compositions. The primary composition of felsic magma inputs into the Buya pluton chamber are not preserved but almost certainly lie in or near the compositional gap separating PQS and FG.

Thermodynamic Modeling

Textural evidence from field observations and the combined geochronological, isotopic, and geochemical data allow us to identify at least three main modes of compositional evolution or modification for the Buya pluton. These include mafic and felsic magma influx, hybridization, and crystal-liquid separation. The actual direct contribution of mafic magma to the material that comprises the Buya pluton is significant but clearly understated if the contribution to hybrid magmas is not also considered. The physical processes of mafic injections into semi-consolidated felsic mush facilitated hybridization, and driven compositional change from ME toward PQS. These same physical processes have also acted on PQS crystal mushes, leading to the extraction of interstitial liquids and driving compositional trends further toward FG. Textural and outcrop evidence that we see for crystal-liquid separation relates mainly to the removal of liquid from a crystal mush and produces the same geochemical results. These processes did not form a continuous sequence of events; they were punctuated and repeated over the long magmatic history of the pluton.

Based on the observations made above, R2AFC (recharge–assimilation-fractional crystallization) and FC (fractional crystallization) are the only processes that seem to provide a potential mechanism to generate the Buya pluton. To evaluate these processes quantitatively, we conducted numerical modeling of R2AFC and FC processes using a Magma Chamber Simulator (MCS; Bohrson et al., 2014). All MCS output data and related illustrations can be collected from Item B (footnote 1).

The basaltic parent and the mafic recharge compositions are 07HT-38–4 and 13WKL-32(2), and the silicic contaminant is the most evolved PQS 07HT-37–3 (Table S3 and Item B, footnote 1). Crystallization conditions were chosen at P = 4.0 kbar. In the R2AFC model, assimilation occurs continuously during each fractionation and recharge cycle. The AFC process in the R2AFC model generates magmas with similar K2O, K2O+Na2O, and Ba and lower Sr than that of the GD, which implies that they may reflect mixing of PQS and mafic magma (Figs. 67). However, this model produces magmas with higher K2O+Na2O and Sr and lower K2O and Ba, which fail to reach the contents of the mafic rocks of the Buya pluton. That is, the excessively high K2O in the mafic rocks of the Buya pluton probably could not be obtained from the R2AFC model. Thus, it seems more likely that the compositional variation of the mafic rocks of the Buya pluton variably derived from the re-melting of lithospheric mantle sources previously modified through the addition of subduction-related K-fluids or melts (see next section).

For the crystal-liquid separation process, we made FC-only runs starting from the sample closest to the compositional gap separating PQS and FG (PQS 07HT-37–3) in the 4.0 kbar (Table S4, footnote 1). The modeled melt compositions produced the K2O and Ba and Sr and Ba compositions of the FG through a high degree of fractionation of the 07HT-37–3 (Figs. 67), whereas the models resulted in excessively high K2O in the melt. This is mainly because there is no crystallization of alkali feldspar in MCS. We also conducted numerical modeling of the AFC model for the felsic parental melt involving assimilation of the Ailiankate Group country rock. However, the model-generated melts are too high in K2O to be identical to FG (it doesn't show in figures). Xenoliths of Ailiankate Group country rock are found within FG but are not common. On this basis, we suggest that assimilation of crust at, or near, the level of intrusion has probably not had a significant effect on the compositional evolution of the Buya pluton.

Origin of the Mafic and Felsic Magmas in the Buya Pluton

The MEs from the Buya pluton have high Mg# (>60) and high Cr (up to 476 ppm) and Ni (up to 182 ppm) concentrations, which reflect equilibration/exchange with mantle peridotite. However, these rocks also show distinct enrichments in light rare earth elements (LREE), Th, K2O, and large ion lithophile elements (LILEs), and in some cases they are more abundant than in the co-magmatic felsic rocks. Concentrations of Zr, Hf, Nb, and Ta are also too high for most rocks of mantle origin but still produce distinct negative anomalies when plotted on mantle-normalized trace element diagrams. These features are in accord with negative εNd(T) values (−5.7 to −6.7; Ye et al., 2008), which indicates the presence of a crustal component within the rock. The PQS and ME samples have different whole-rock Nd isotopes (the PQS sample has εNd(T) values of −10.4 to −8.4 compared to −5.7 to −6.7 in the MEs; Ye et al., 2008) and geochemical (Figs. 67) compositions, which suggest little evidence that either the mafic or felsic magmatic contributions to the Buya pluton have been influenced by wall-rock assimilation. Hence, these observed compositions, and isotopic ranges, most likely reflect the bulk melting source. Among the possible explanations for these compositions are contamination of an asthenospheric melt by Proterozoic (or older) lower crust, melting of an ancient (perhaps Proterozoic?) “fossil” subduction-modified lithospheric mantle underplate, or melting of younger lithospheric mantle modified during subduction involving at least a component of Proterozoic (or older) crust, likely including K-rich sediment (e.g., Ye et al., 2008). We currently do not have the data to identify which of these possibilities is most likely. However, the West Kunlun region witnessed prolonged subduction of the Proto- or Paleo-Tethys Ocean throughout the Paleozoic (Wang et al., 2013, and references therein), and so the suggestion of a subduction-related source appears reasonable. The mafic rocks of the Buya pluton share geochemical characteristics with sanukitoids, high Nb basalts, and shoshonites, which are mafic to felsic rocks variably derived through re-melting of lithospheric mantle previously modified through the addition of subduction-related K-to Na-rich fluids or melts (e.g., Shirey and Hanson, 1984; Defant et al., 1992; Rapp et al., 1999; Smithies and Champion, 2000; Martin et al., 2005; Gómez-Tuena et al., 2018).

Magmatic zircons from PQS have εHf(t = 456) values of –10.4 to –5.4 and two-stage Hf model ages of 1520 to1769 Ma, and they are identical to MEs. Then the Hf isotopic data are readily explained by hybridization in the magma source of PQS. Available data (Ye et al., 2008), however, suggest that the Nd-isotopic composition of PQS is less radiogenic (εNd(T = 430 Ma) of −8.4 to −10.4) than that of the ME. According to Ye et al. (2008), the Nd-isotopic data preclude derivation of the Buya felsic magmas through melting of known, regionally available crustal sources. These data do not preclude the possibility that the source of PQS parental magma was similar to that of ME but included an additional component with less radiogenic Nd-isotopic composition and perhaps also a lower Hf/Nd ratio (e.g., an ancient subduction-modified source). However, it seems unlikely that magmas as silica-rich as those likely to be parental to the PQS-FG compositional array could be the melt product of the peridotitic source required to form the ME magma. In addition, the strongly potassic compositions of many of the ME mafic rocks reflect a source composition with a K2O/Na2O ratio that is much too high to also be the source of magmas that are parental to PQS-FG with a low K2O/Na2O ratio (<1). Thermodynamic modeling indicates that the strong K2O enrichment of the mafic rocks is not a result of alteration or hybridization within the Buya pluton at a shallow level. Rather, it suggests that the source feeding the mafic magmas themselves was compositionally inhomogeneous, at least with respect to K2O (Fig. 7). We suggest that the strongly enriched compositions of the mafic rocks within the Buya pluton reflect derivation from a mantle source that was previously enriched through the addition of melt components related to the earlier subduction of crust that included old and K-rich crust (likely Precambrian continental sediments, e.g., Ye et al., 2008). The resulting peridotite source was strongly but inhomogenously enriched in LILEs and K2O, as well as LREEs. The resulting magmas had ME-like compositions (Fig. 9A). They intruded mafic lower crust and probably established a lower crust in melting, assimilation, storage, and homogenization (MASH) or hot-zone (e.g., Hildreth and Moorbath, 1988; Annen et al., 2006). Castro (2020) proposed that sanukitoid magmas could act as water donors that trigger extensive melting of the lower crust, giving rise to granodioritic liquids.

The felsic magmas that were parental to the PQS-FG suite were possibly extracted from this MASH zone and emplaced within the Buya pluton (Fig. 9B). The parental magma of PQS-FG owes its high Sr and Ba concentrations and high Sr/Y ratios mainly to the enriched composition of the earlier ME-like magmas in the MASH zone at the base of the crust. Other ME-like magmas did not pond in the lower crust but were directly emplaced at higher crustal levels without significant interaction with mafic lower crust. Multistage injection of these mafic magmas into the Buya magma chamber formed mafic sheets, enclaves and dikes, and induced the convective flow of the crystal mush and formed composite enclaves and the hybrid mush of GD. The sinking or settling of the mafic sheets and enclaves led to compaction of the chamber and the extraction of interstitial melt, which formed PQS and high-K, Sr/Y ratio calc-alkaline FG (Fig. 9C).

The Buya pluton provides textural evidence of processes that form, fill, and modify a pluton. This preservation, to a large extent, may relate to the small size of the pluton itself, with many of the processes frozen before completion. Field and petrographic relationships, geochemical and geochronological data, and thermodynamic modeling indicate that sanukitoid mafic magmatism was a volumetrically significant component of pluton fill throughout the construction history and played an important role in the compositional evolution of the pluton. The physical processes of flow and the disaggregation of mafic injections into semi-consolidated felsic mush have facilitated crystal-liquid separation and driven the compositional evolution at shallow crustal levels. These processes did not form a continuous sequence of events but were punctuated and repeated over the lifetime of the pluton. Also, we suggest that assimilation of country crust at, or near, the level of intrusion has probably not played a significant role in the compositional evolution of the Buya pluton. Instead, much of the variation within the Buya pluton most likely reflects the features of the magma source. Our model also indicates that the crystal fractionation process may generate the apparently high Sr/Y trace element features of silicic high-K, calc-alkaline magmas, which calls into question the interpretation that high Sr/Y ratios in felsic rock invariably reflect high-pressure melting.

This work was supported by the National Natural Science Foundation of China (41672187 and 42030307) and the Ministry of Science and Technology of the People's Republic of China (MOST) Special Fund of the State Key Laboratory of Continental Dynamics, Xi'an, China. R.H. Smithies publishes with the approval of the executive director, Geological Survey of Western Australia. Calvin G. Barnes, R.A. Wiebe, Calvin F. Miller, and two anonymous reviewers provided extremely helpful comments on earlier versions of this paper.

1Supplemental Material. Item A: Methods and Figures S1 and S2. Item B: Magma Chamber Simulator output data and related illustrations. Table S1: LA-ICP-MS U-Th-Pb isotope data of zircon. Table S2: LA-MC-ICPMS Lu-Hf isotope data of zircon. Table S3: Major- and trace-element analytical data of the Buya pluton. Table S4: Amphibole compositions from the porphyritic quartz syenite of Buya pluton. Please visit to access the supplemental material, and contact with any questions.
Science Editor: David E. Fastovsky
Associate Editor: Julie Roberge
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.