Multistage Fractional Crystallization in the Continental Arc Magmatic System: Constraints from the Appinites in Tengchong Block, Southeastern Extension of Tibet

The petrogenesis and evolution process of continental arc magmatism provide insight into discovering the formation and di ﬀ erentiation of continental crust. Therefore, the geochemical, isotopic, and mineralogical analyses were conducted for coeval continental arc igneous rocks in the Tengchong Block to clarify their evolution process in the continental arc magmatic systems. The Middle Triassic appinites in the Tengchong Block, southeastern extension of Tibet, were generated at the subduction setting with zircon U-Pb age of ca. 243Ma. The Nb/Yb , Zr/Yb , and Ta/Yb ratios along with depleted zircon Hf isotopic compositions indicate a source with an N-MORB a ﬃ nity for the appinites. However, relatively enriched whole-rock Sr-Nd isotopic compositions with the characteristic of high Sr/Nd , Ba/Th , Th/La , and Th/Nd ratios suggest the source was metasomatized by ~ 2% subducted sediment-derived ﬂ uid. According to the REE ratios modeling, the primary magma of Nabang appinites was due to 5-10% partial melting of such metasomatized mantle source. The appinites are characterized by variable compositions, such as SiO 2 contents of 47.82-61.74wt.% and MgO of 10.61-2.61wt.%, which resulted from the polybaric and multistage fractional crystallization of a slightly hydrous primary magma in a thick crust. At lower crustal pressures, clinopyroxene was the main fractionating phase, and at middle crustal pressures, amphibole+magnetite were the dominant fractionating phases; predominant plagioclase fractionation occurred at the magma emplacement level. This process could be an e ﬀ ective mechanism to induce the di ﬀ erentiation of continental crust. The fractionation of clinopyroxene and amphibole, accompanied by suppressing plagioclase at lower-middle crustal pressures, induces the high alumina in the evolved melt and forms high-alumina basaltic to andesitic magma.


Introduction
The subduction zone magmatism has long been considered an essential role in continental crustal formation and differentiation based on the trace element similarities [1]. However, there is a "crust paradox" that the bulk composition of arc magmatism is basalt, too basaltic relative to the "andesitic" bulk composition of continental crust [2]. Therefore, the evolution of magmatism in the arc crust is a key to understanding the formation and differentiation of continental crust. The fractionation of mantle-derived hydrous primary magma at different crustal levels within arc crust is proposed as an effective mechanism to explain this paradox [3][4][5]. The pyroxenite and hornblendite cumulates found in the arc magmatism expand the recognition of fractionation products [5][6][7][8]. In addition, the transcrustal magmatic system proposed by Cashman et al. [9] provided a new concept to understand the process of segregation, differentiation, ascend, and storage or final erupt through the deep to the shallow crust. Therefore, the coeval magmatic rocks in continental arc supply an opportunity to further discover the evolution of magmatic system from depth to shallow.
Appinites are a group of coeval plutonic or hypabyssal rocks, ranging from ultramafic to felsic in compositions [10]. Generally, they occur as subordinate components associated with a large pluton complex in a region, such as a satellite around the pluton [11]. They are important water or fluid donor to induce migmatization and anatexis of crustal compositions [10,12]. The common feature in appinites is that hornblende is the dominant mafic mineral, occurring both as a large phenocryst and fine-grained matrix, representing the high water content in mafic magma [10]. The primitive magma for appinites has been considered to be derived from a hydrated mantle previously metasomatized by fluid/melt released from subducted oceanic slab or/and sediment [10,12]. Therefore, the appinites shed light on the generation, emplacement, and crystallization of mafic to felsic magma under water-rich conditions [10,13]. In the Tengchong Block, southeastern extension of Tibet, the Middle Triassic appinites are exposed in the Nabang area with variable compositions, representing continental arc magmatic products related to the Paleo-Tethys. This study is aimed at evaluating the petrogenesis and evolution of the appinites in the continental arc. The geochemical and mineralogical data indicate that the appinites were derived from N-MORB-type mantle metasomatized by subducted sedimentderived fluid and experienced the multistage fractionation in a thick crust.

Geological Setting and Petrography
The Tengchong Block is considered a continental arc related to Neo-Tethyan subduction and Indian-Asian continental collision during the Late Mesozoic to Cenozoic [14,15]. It is bounded by the Gaoligong Belt from the Baoshan Block to the east and the Sagaing Fault and Mogok metamorphic belt from the west Burma Block to the west ( [16]; Figures 1(a) and 1(b)). The Tengchong Block is mainly composed of Paleo-Proterozoic basement, named Gaoligongshang Formation, Late-Paleozoic to Mesozoic sedimentary rocks and Early Paleozoic, Late Permian to Triassic, and Early Paleocene to Eocene intrusions with Late Miocene to Holocene volcanic rocks ( [15] and references therein), which record the multistage oceanic subduction and microcontinental collision [17]. The Tengchong Block is considered to have an affinity with the Lhasa Block since Early Paleozoic [14,18] and experienced similar tectonic evolution histories. The Lhasa Block is subdivided into northern, central, and southern subterranes by the Shiquan River-Nam Tso Melange zone, Luobadui-Milashan Fault [19]. The Permian-Triassic arc-type magmatism, eclogites, and Carboniferous-Triassic ophiolites were discovered in the Sumdo area, suggesting the existence of a branch of Paleo-Tethyan Ocean within the Lhasa Block [20][21][22][23][24]. The arc-type, syn-collision, and within-plate magmatism from Late Permian to Early Jurassic were also reported in the Tengchong Block, southeastern extension of Lhasa Block [25]. Thus, a Permian to Triassic subduction setting was related to a branch of the Paleo-Tethyan Ocean within the Lhasa-Tengchong Block. The Middle Triassic appinites occurred in the Nabang area, Tengchong Block (Figure 1(c)), formed in a subduction setting consistent with the consensus of the tectonic setting for appinites [10].
The lithologies of appinites in the Nabang area are mainly composed of hornblende gabbros and diorites in the outcrop (Figures 2(a) and 2(b)). The hornblende gabbros and diorites are transitional contacts, and they have similar mineral assemblage but different volume contents, representing the different units in the same magmatic system. The hornblende gabbros selected in this study show porphyritic texture for whole rock and equigranular texture within the matrix, indicating the products of magmatic crystallization. The subhedral to euhedral amphiboles are dominant mafic minerals in the hornblende gabbro (modal volume of 65-75%), and they are subdivided into two groups: the first group shows large and short columnar crystal with 1.5-3 mm in length (phenocryst), and another group is characterized by relatively small and granular crystal with <0.2 mm in diameter (matrix, Figures 2(c) and 2(d)). Parts of amphibole phenocrysts display apparent zoning texture (Figure 2(d)), enclose numerous minor plagioclase, quartz, and biotite, and show poikilitic texture (Figure 2(c)). Several amphibole phenocrysts contain the clinopyroxene, implying the replacement reaction between these minerals. The plagioclases are subordinate mineral phases in the hornblende gabbro with the modal volume of 20-25% and show prominent Carlsbad-albite compound twin with subhedral to anhedral crystal. A few opaque minerals and apatites are enclosed in plagioclase (Figures 2(e) and 2(f)). The biotite, quartz, and K-feldspar occur as interstitial phases, and they are 5-10% in modal volume (Figures 2(e) and 2(f)). The accessory minerals mainly include apatite, zircon, and opaque minerals, such as Fe-Ti oxide.

Analytic Methods
The zircon U-Pb and Lu-Hf isotopic compositions for Nabang appinites were analyzed at the State Key Laboratory of Continental Dynamics, Northwest University in Xi'an, China. Zircon grains from three samples were separated by conventional heavy liquid and magnetic techniques and subsequently handpicked and mounted in epoxy resin disks, polished, and carbon-coated. The internal morphology and structure of representative zircons were examined by cathodoluminescence (CL). Laser ablation inductively coupled plasma-mass spectrometry (LA-ICP-MS) zircon U-Pb analyses were conducted on an Agilent 7500a ICP-MS equipped with a 193 nm laser, following the detailed methods of Yuan et al. [26]. The 207 Pb/ 235 U and 206 Pb/ 238 U ratios were calculated using the GLITTER program, and corrections were applied using Harvard zircon 91500 as an external calibration standard. The in situ zircon Lu-Hf isotopic compositions were analyzed using a Neptune MC-ICP-MS. The laser repetition rate was 6 Hz at 100 mJ, and the spot size was 30 μm. The analytical technique followed Yuan et al. [27]. During the analyses, the 176 Hf/ 177 Hf and 176 Lu/ 177 Hf ratios of the standard zircon (91500) were 0:282294 ± 15 (2σ) and 0.00031, respectively.

Lithosphere
Fresh whole-rock samples were chipped and powdered to 200 mesh using a tungsten carbide ball mill for the major and trace element analyses. Major and trace elements were analyzed by X-ray fluorescence (XRF; Rikagu RIX 2100) and ICP-MS (Agilent 7500a), respectively. Analyses of USGS and Chinese national rock standards (BCR-2, GSR-1, and GSR-3) indicated that the analytical precision and accuracy for major elements were generally better than 5%. Sample powders were digested using an HF + HNO3 mixture in high-pressure Teflon bombs at 190°C for 48 hours for the trace element analyses. For most trace elements, the analytical error was less than 2% and the precision better than 10% [28].
Whole-rock S--Nd isotopic data were conducted using a Nu Plasma HR multicollector mass spectrometer for four samples at State Key Laboratory of Continental Dynamics (SKLCD), Northwest University in Xi'an, China, and using a Neptune Plus MC-ICP-MS for another four samples at the Wuhan Sample Solution Analytical Technology Co.,   Major-element compositions of the minerals were determined using an electron microprobe (JXA-8230) at the Ministry of Education, Key Laboratory of Western China's Mineral Resources and Geological Engineering, Chang'an University. The operating conditions included an acceleration voltage of 15 kV, a beam current of 10 nA, and a beam diameter of 1 μm. Natural and synthetic microprobe standards were supplied by SPI, including jadeite for Si, Al, and Na, diopside for Ca, olivine for Mg, sanidine for K, hematite for Fe, rhodonite for Mn, and rutile for Ti.

Zircon U-Pb
Ages. Three hornblende gabbro samples were collected for LA-ICP-MS zircon U-Pb isotopic analyses. The results are listed in Table S1 and Figure 3.
The representative zircons selected from hornblende gabbros commonly show subhedral to euhedral, colorless, and transparent crystals, with widths of 50-120 μm, lengths of 50-200 μm, and aspect ratios from 1 : 1 to 1 : 4. The zircons show broad compositional zoning on the CL images, indicating that they are formed at high-temperature conditions [29]. Thirty-five spots have been analyzed for a sample of NB18-1 and yielded concordant 206 Pb/ 208 U ages ranging from 236 Ma to 245 Ma with a weighted mean age of 243 ± 1 Ma (MSWD = 0:6, 2σ), U contents of 117-617 ppm and Th contents of 62-470 ppm with Th/U ratios of 0.53 to 1.20. Twenty-eight spots have been conducted for NB18-2 and given concordant 206 Pb/ 208 U ages from 237 Ma to 249 Ma with a weighted mean age of 243 ± 2 Ma (MSWD = 0:3, 2σ), U contents of 102-   Figure 3: Zircon U-Pb Concordia diagrams, weighted mean age of 206 Pb/ 238 U and CL images of representative zircon grains for the Nabang appinites.
5 Lithosphere 649 ppm, Th contents of 55-444 ppm with Th/U ratios of 0.42 to 1.07. Twenty-seven spots have been analyzed for XM15 and yielded 206 Pb/ 208 U ages varying from 240 Ma to 246 Ma with a weighted mean age of 243 ± 2 Ma (MSWD = 0:1, 2σ), U contents of 96-965 ppm, Th contents of 53-852 ppm, and Th/U ratios of 0.50 to 1.11. Therefore, the hornblende gabbros were generation during the Middle Triassic with a crystallization age of ca. 243 Ma and were coeval with the diorites reported by Huang et al. [30].

Major and Trace Elements.
The results of whole-rock major and trace elements are listed in Table S2. Most hornblende gabbros are gabbro in compositions (Figure 4 Figure 5; [31,32]). In the primitive mantlenormalized multielement diagram ( Figure 6(a)), the hornblende gabbros show apparent enrichment in large ion lithophile elements (LILE, such as Rb, Ba, K, and Pb), relative to high field strength elements (HFSE, e.g., Nb, Ta, and Ti), similar to the coeval diorites in Nabang area ( Figure 6(a)). The total rare earth element (REE) contents vary from 35 to 106 ppm. In the chondrite-normalized REE pattern, the hornblende gabbros show slight REE fractionation with ðLa/YbÞ N values of 2.2 to 5.3, weakly positive to negative Eu anomalies (δEu = 1:06 − 0:81, Figure 6(b)). Besides, the hornblende gabbros have apparently high and variable Cr (618-15.8 ppm) and Ni (140-12.5 ppm) contents, and the nonlinear trends between MgO and Cr-Ni also suggest the process of fractional crystallization (Table S1).

Whole-Rock Sr-Nd and Zircon in situ Lu-Hf Isotopic
Compositions. The whole-rock Sr and Nd isotopic compositions of eight representative hornblende gabbros have been analyzed, and the results are listed in Table S3 [30]). The isotopic compositions show obvious low radiogenic Nd and high radiogenic Hf indicating the Nd-Hf isotopic decoupling (Figure 7(c)).

Mineral
Composition. The amphibole, clinopyroxene, and plagioclase were analyzed for the mineral compositions, and the results are listed in Table S5. The clinopyroxenes are enclosed in the amphibole phenocrysts as residual crystals. They belong to diopside in compositions according to the Wo-En-Fs diagram ( [33]; Figure 8 [37]. Therefore, the magmatic evolution should be evaluated when we further discuss and understand the petrogenesis of Nabang appinites. 5.1.1. Evaluation of Crustal Contamination. The mafic rocks derived from the mantle and intruding into the crust should estimate the influence of crustal contamination on their geochemical characteristics. Generally, the crustal contamination markedly modifies the elements and isotopic compositions of the primary magma [38], enriching Sr-Nd isotopes and increasing LILE. The appinites show Nb-Ta depleted, 6 Lithosphere and LILE enriched, but consistent Sr and Nd isotopes with a variation of MgO (Figure 10(a)), indicating negligible or limited crustal contamination. Therefore the compositional variations of the appinites are not mainly controlled by crustal contamination.

Fractional Crystallization and Evolution
Process. The nonlinear trends between MgO and other oxides indicate a multistage process of fractional crystallization [31,32].
According to the compositional variation of Nabang appinites, the processes of fractional crystallization are subdivided into three stages of paths at least ( Figure 5). The sharp decrease of MgO companied with Ca/Al ratios in the appinites indicates that clinopyroxene is the dominant phase in the earliest stage of crystallization and fractionation. Owing to high MgO and CaO content in the clinopyroxene, it raises the Al 2 O 3 , K 2 O and Na 2 O but decreases MgO contents and Ca/Al ratios in the residual melt during clinopyroxene fractionation ( Figure 5). If the clinopyroxene differentiation predominates, the SiO 2 content of evolving melt could be depleted or constant [39]. This is observed in the variable trend between MgO and SiO 2 , slightly decreasing SiO 2 with decreasing MgO within the first stage ( Figure 5(a)). Dy/Dy * is an important parameter to limit the curvature of the REE pattern [40]. During fractional crystallization, the amphibole and clinopyroxene are the major minerals capable of significantly decreasing Dy/Dy * [40]. Clinopyroxene may reduce or increase Dy/Yb with fractionation due to variable partition coefficient with D Dy /D Yb ðcpxÞ ranging from 0.72 to 1.38 in peridotite to rhyolite, but amphibole fractionation generally reduces Dy/Yb because of D Dy /D Yb ðampÞ ranging from 1.32 to 2.35 in peridotite to rhyolite ( [40]; GERM 7 Lithosphere database). The clinopyroxene differentiation in the first stage increases Dy/Yb but decreases Dy/Dy * for the evolved melt (Figure 11(a)). Besides, the alkalis and Al 2 O 3 contents increasing is against with the plagioclase crystallization and fractionation in the first stage. The suppression plagioclase crystallization occurs at a condition of high H 2 O content or/and high pressure [41][42][43][44][45]. It is consistent with the consensus on the high H 2 O content for the appinite magma [10] and the high pressure at 7.1 kbar, a calculation based on the clinopyroxene composition. The unchanged and consistent δEu values and gradually increasing Sr contents with decreasing MgO and SiO 2 also suggest that the plagioclase is not involved in the fractional process in the first stage (Figures 11(b) and 11(c)). Furthermore, the slight increase of Fe 2 O 3 T and minor change of TiO 2 with decreasing MgO in this stage argue against with the fractional crystallization of magnetite, which could be the late crystallization phase ( Figure 5). Therefore, the clinopyroxene fractionation and suppression plagioclase crystallization are the mainly evolved processes in the first stage.
After the first stage of clinopyroxene crystallization and fractionation, an important feature is to strengthen H 2 O content in the residual melt, which allows the reaction of early formed clinopyroxene with hydrous melt and forms the hydrous mineral, such as amphibole. This is marked in the second stage of evolution for the Nabang appinites. It is evidenced by residual clinopyroxene enclosed in amphibole ( Figure 9) and further confirmed by the fact that the quartz crystals were enclosed in almost all of the amphibole phenocrysts or occurred as interstitial phases around amphibole CaO (wt.%) (i) Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Figure 5: Diagrams of MgO versus major elements for the Nabang appinites. 8 Lithosphere phenocrysts ( Figure 2). Compared with clinopyroxene, the amphiboles have lower SiO 2 and CaO and higher FeO, Al 2 O 3 , and alkalis. Therefore, it would release exceeded silica and CaO and absorb FeO, Al 2 O 3 , and alkalis from the melt during the clinopyroxene reaction to amphibole, resulting in slightly increasing SiO 2 and CaO, decreasing or/and constant alkalis, and a relatively small slope of Ca/Al trend with decreasing MgO in the further evolved melt ( Figure 5). Compositional variations of zoning amphiboles reflect these elemental exchanges, reaction-replacement amphibole core with more primitive elements than amphibole rim, such as higher SiO 2 and MgO core (Figures 9(a)  9 Lithosphere 10). The plagioclases in the second stage are likely to onset of crystallization, but a small amount of plagioclases is involved in the fractionation, because of the trend of decreasing δEu with decreasing MgO and relatively slow increasing Sr with increasing SiO 2 comparing to the first stage of evolution ( Figure 11). The accessory mineral, such as Ti-Fe oxide (magnetite) and apatite could be fractionated in the second stage because of the decreasing Fe 2 O 3 T, TiO 2 and P 2 O 5 with decreasing MgO (Figure 5). Therefore, the amphiboles with accessory minerals fractionation, especially magnetite, are predominant in the second stage of the Nabang appinite evolution.
In the third stage of the fractional process, the major elements of Al 2 O 3 and Na 2 O and trace elements of Sr with δEu are subsequent decrease in the evolved appinites, indicating the dominant plagioclase fractionation (Figures 5 and 11). The plagioclase fractionation results in the potassic enrichment in the intermediate composition ( Figure 5). In addition, the successive decreasing MgO and Fe 2 O 3 T and Dy/Dy * and Dy/Yb indicate the amphibole is also involved in the fractional phase in the last stage ( Figure 5).
The compositional variations of Nabang appinites were controlled by the multistage fractional process, which is sensitive to pressure and H 2 O content in magma and changes the crystallization sequences. It is consistent with the experimental results [41,44,45]. source. The immobility of HFSE and HREE and their low concentrations in the slab or sediment-derived hydrous fluid/melt could be almost unaffected by subduction-related processes [47][48][49]. Thus, these trace elements were used to reveal the characteristics of mantle sources. On the diagrams of Zr/Yb versus Nb/Yb and Ta/Yb versus Nb/Yb, the lessevolved appinites fall close to the N-MORB, indicating man-tle sources with an N-MORB affinity (Figures 10(b) and 10(c)). It is further confirmed by the depleted zircon Lu-Hf isotopic compositions with ε Hf ðtÞ values of 6.0 to 16.4. However, their slightly enriched Sr and Nd isotopic compositions indicate that the mantle sources should be metasomatized by fluid/melt derived from subducted sediments or/and altered oceanic crust (AOC). The metasomatized mantle agent is generally estimated based on the variable mobility of incompatible trace elements in different subducted-related fluids/melts [50]. The LILEs are easily mobile in the hydrous fluid, and Th is mainly enriched in subducted sediments and participated into the melt [49]. The less-evolved appinites have markedly higher Sr/Nd  and Ba/Th (444-883) and slightly higher Th/La (0.05-0.10) and Th/Nd  [49,52]). Consequently, AOC has less effect on the Sr and Nd isotopic composition to the fluid/melts than subducted sediments. The Sr-Nd isotopic compositions and trace elemental ratios are widely utilized to quantify the subducted-related fluid/melt contribution. The parameters in the calculation are listed in Table S6. The results show that the sources for Nabang appinites were N-MORB-type mantle metasomatized by~2% subducted sediment-derived fluid (Figure 12). In the process of metasomatism, the Nd-Hf isotope decoupling could achieve, due to the higher mobility of Nd with respect to Hf [47,53], which is a common feature observed in Phanerozoic arc systems [54].
The REE concentrations and ratios are generally used to constrain the natures of source mineralogy and the degree of partial melting [55]. The HREEs more readily participate into the garnet, and MREE/HREE and LREE/MREE ratios are sensitive to the degrees of partial melting of garnet-or spinel-lherzolite sources. The melts produced by medium or small degree melting of garnet-lherzolite sources have high Sm/Yb and La/Sm, while the melts produced by spinel-lherzolite sources will have similar Sm/Yb but decreasing La/Sm with increasing melting degree [55]. We attempt to formulate the nature of sources and partial melting process for the Nabang appinites. The nonmodal batch melting model is used after Shaw [56]. The starting materials are based on the mentioned above, 98%DDM + 2% subducted sediments-derived fluids. On the La/Sm against Sm/Yb diagram (Figure 13), the Nabang appinites are close to the melting trajectories drawn for spinel-lherzolite source with 5-10% melting degree.

Implications for High-Alumina Basalt and Continental
Arc Magmatic System. In the Nabang appinites, it is noteworthy that Al 2 O 3 contents (>16%) are high enough in the evolved members (MgO < 6%), which are similar to the compositions of low MgO high-alumina basalt and basaltic andesite (HAB) in the arc or midocean ridges [41,44,57]. 12 Lithosphere Several models have been proposed for the origin of HAB magma: (1) directly derived from partial melting of subducted slab [58]; (2) evolved liquid that has experienced selected accumulation of plagioclase inducing the rise of alumina [57]; and (3) the delayed plagioclase of hydrous mantlederived basaltic magma at high-pressure condition [41,44]. As mentioned above, the variable compositions of Nabang appinites are mainly controlled by the multistage fractional crystallization. According to the experimental results and natural sample researches, the water content and crystallization pressure of basaltic magma play important roles in the crystallization sequences and phases and influence on the chemical differentiation (e.g., [42][43][44][45]59] Figure 5). The pressure is also an important factor in controlling the liquidus phase. At relative low pressure, the liquidus minerals of mantle-derived basaltic magma are olivine with magnetite, while the stability field (liquidus volumes) of clinopyroxene expends and crystallizes very early at the high-pressure condition, even with low magmatic water content [41,44]. The increasing Fe 2 O 3 T and Al 2 O 3 and decreasing CaO with decreasing MgO in the early evolution of the Nabang appinites indicate the Ca-and Mg-rich mineral differentiation and Al-rich mineral instability, such as fractionation of clinopyroxene and suppression of plagioclase, which requires a relatively high-pressure condition [41,45]. However, the garnet could be a common phase during the crystallization of primary, hydrous arc magma at the high-pressure condition (generally more than 1-1.2 GPa; [5,42]). In the Nabang appinites, the constant Gd/Yb and Dy/Yb ratios and slightly increasing Fe 2 O 3 T with decreasing MgO contents for the less-evolved appinites indicate the garnet is not involved in the fractionation phase [60]. However, it is noteworthy that the Gd/Yb and Dy/Yb show unobvious rise but display flatter pattern model of HREE for the high MgO garnet-poor pyroxene, because of small amounts of garnet, which is considered high-pressure cumulate and named arclogite in Mesozoic Sierra Nevada from [5,61]. Therefore, the fractionation of garnet for the less-evolved appinites cannot be completely ruled out. The early fractionation phase is dominated by clinopyroxene, and crystallization pressure is more than 7.1 kbar, corresponding to the lower crust, for the less-evolved appinites. This is further confirmed by a thick crust at 52-59 km for Tengchong Block during Middle Triassic, estimated according to the Sr/Y ratios after Chapman et al. [62]. The early fractionation products of could be similar to high-MgO garnet-poor pyroxene in the Sierra Nevada batholith [5,6]. The suppression of plagioclase and magnetite along with fractionation of clinopyroxene for the slightly hydrous primary magma drives the depletion of Si, Mg, and Ca, as well as the enrichment of Al and slightly increasing Fe in the evolved melt at high-pressure condition. This fractional process also results in the rise of Fe/Mg values and drives the magma transition from calc-alkaline to tholeiitic. Subsequently, the water should increase during the evolution of slightly hydrous magma [63,64] and induce the reaction between extant anhydrous minerals (clinopyroxene) and hydrous melt in the lower crustal mush zones, which ultimately forms hydrous minerals, such as amphibole [65]. This reaction is evidenced that the residual clinopyroxenes are enclosed into the amphiboles, and a partial of amphibole phenocrysts have more primitive compositions than the matrix amphiboles. Therefore, the partial of amphibole phenocrysts are the reaction-replacement origin, and the amphiboles in the matrix are the magmatic origin.
Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2+3 Figure 11: Dy/Yb versus Dy/Dy * after Davidson et al. [40] (a), MgO versus δEu (b), and SiO 2 versus Sr (c) for the Nabang appinites. 13 Lithosphere the amphibole is considered an important and cryptic fractional mineral within the arc magmatic system [39,40,65]. The hornblendite cumulates are also found in the andesitic rocks in Central Andes [7], and the amphibole compositions generally record the fractionation process occurring at the mid-low crust in depth [7,8]. The amphiboles in the matrix of Nabang appinites record their crystallization and fractionation pressure at 1.1-5.9 kbar, corresponding to the middle crust in depth. Therefore, the amphibole fractionation is identified posterior to the fractionation of clinopyroxene in the Nabang appinites. The fractionation products could be similar to the hornblendite cumulates in Central Andes [7]. The magnetite and a small amount of plagioclase crystallize along with or posterior to the amphibole. The fractionation of amphibole and magnetite with a small amount of plagioclase further drive the residual melt to intermediate to acid composition with the depletion of Fe and enrichment of Si and Al. Furthermore, the Fe-rich mineral fractionations drive the melt from tholeiitic to calc-alkaline trend. Therefore, the high Al 2 O 3 contents in the evolved member of Nabang appinites result from the fractionation of clinopyroxene and amphibole but the suppression of plagioclase within the lower to the middle crust. The extensive plagioclase fractionation mainly occurs at the last stage of magmatic evolution and finally leads to the Al 2 O 3 , Na 2 O, and Sr content decrease.
In summary, the broad spectrum of compositions in the Nabang appinites are the most accessible witness to the differentiation processes during the upward-moving of the   [52]. Global subducted sediment composition from Plank and Langmuir [49]. The AOC and GLOSS fluid was calculated assuming the 2% Rayleigh distillation. The GLOSS melt was calculated assuming 20% modal batch melting. The subducted AOC compositions refer from Brenan et al. [76] with a modal compositions of 60% garnet ðGrtÞ + 40% clinopyroxene ðCpxÞ and the partition coefficients for Grt and Cpx from Brenan et al. [76,77]. The partition coefficients for calculation of GLOSS fluid and melts refer from Johnson and Plank [78]. The parameters are listed in Table S6. 14 Lithosphere slightly hydrous magmatic system in a thick crust. They have experienced the multistage fractional crystallization at a different depth, along with a pathway of clinopyroxene fractionation at the lower crustal pressure, amphibole+magnetite fractionation at the middle crust pressure, and plagioclase fractionation in the magma emplacement level ( Figure 14). The Nabang appinites show a polybaric differentiation history in the transcrustal magmatic system, proposed by Cashman et al. [9], which is different from the traditional consensus of a melt-dominated magma chamber for the igneous processes. The multiple vertical fractional crystallization could form the corresponding igneous cumulates, such as pyroxenite and hornblendite [5,7]. After the multistage fractionation, the mantle-derived hydrous melt are driven to variable compositions and finally generate the intermediate to acid magma. If the mafic-ultramafic pyroxenite cumulate delaminates into the mantle because of its high density [4,6], compositions of the latter remaining could be consistent with the bulk compositions of continental crust.  [79], respectively. Composition of Nabang (NB) was using 2%GLOSS fluid + 98%DMM . The detail calculated process and REE partition coefficients refer from Aldanmaz et al. [55].

Paleo-Tethys
O c e a n ic c r u s t a n d li t h o s p h e r ic m a n lt e

Conclusion
Based on the geochronological, geochemical, isotopic, and mineralogical analyses of the Nabang appinites, it is concluded as the following: (1) The appinites in Tengchong Block are generated during the Middle Triassic with zircon U-Pb age of ca. 243 Ma. (2) The sources of appinites have an affinity with N-MORB but are metasomatized by 2% subducted sediments-derived fluid. The primary magma for the Nabang appinites was derived from about 5-10% melting of such mantle sources. (3) The compositional ranges of the appinites indicate they underwent the multistage fractional crystallization within a thick crust, clinopyroxene fractionation at the lower crustal pressure, amphibole+magnetite fractionation at the middle crustal pressure, and plagioclase fractionation in the magma emplacement level. The high Al 2 O 3 contents in the evolved appinites can be resulted from fractionation of clinopyroxene and amphibole but suppression of plagioclase at the lower-middle crustal depth.

Data Availability
The data are listed in the Supplementary Materials.

Conflicts of Interest
The authors declare that they have no conflicts of interest. Lithosphere