The Zhongzan-Yidun arc (ZYA) terrane is a combination of the Zhongzan block and the Yidun arc, bound by the Garze-Litang suture to the east and by the Jinshajiang suture to the west (Figure 1). The Zhongzan block could have been derived from the Yangtze block because they have similar Paleozoic stratigraphic sequences and paleontological fossils (Chang, 2000; Xiao et al., 2004), and it is a result of the opening of the Garze-Litang ocean, which is a branch of the Paleo-Tethys (Deng et al., 2020) from the late Paleozoic. The Jinshajiang Ocean is another branch of the Paleo-Tethys and was closed in the Middle Triassic (Deng et al., 2020).

Magmatism in this terrane mainly occurred in the Late Triassic and Late Cretaceous periods (Xiao et al., 2004). The former is found as medium-to high-K calc-alkaline diorite, monzonite, granodiorite, granite, and volcanic rocks (Qu et al., 2002; He et al., 2013; Wu et al., 2014). The latter is mainly found as monzogranite, biotite granite, and granitic porphyry (Qu et al., 2002; Wang et al., 2014).

The eastern Qiangtang-Simao (EQST) terrane, which extends from north to south, constitutes the main and central parts of the SOB (Figure 1). The Jinshajiang-Ailaoshan suture, which is located to the east of the EQST, was the westward subduction area of the branch of the Paleo-Tethys ocean, while the Longmucuo-Shuanghu-Changning-Menglian suture located to the west was the eastward subduction area of the main Paleo-Tethyan (Deng et al., 2014a; Deng et al., 2014b). The coupling of the subduction of the Proto-Tethys oceanic plate underneath the Simao block and the opening of the Paleo-Tethys ocean suggests that the Eastern Qiangtang and Simao blocks are two Gondwana-derived microcontinents (Usuki et al., 2013; Deng et al., 2014a). The Eastern Qiangtang block was amalgamated with the Western Qiangtang subterrane as the Qiangtang Terrane in the early Jurassic (Hou et al., 2003; Yang et al., 2014) and is the south-central part, while the Lhasa terrane, which is derived from the Australian margin of the Gondwana supercontinent (Zhu et al., 2013), is the southernmost part.

Magmatism in the area developed in the late Permian to very early Triassic in continental-margin arcs, such as the Jomda-Weixi arc along the northern Jinshajiang-Ailaoshan suture (Zi et al., 2012a; Zi et al., 2012b) and the Yaxuanqiao arc in the southern part (Fan et al., 2010; Cai et al., 2014; Xu et al., 2016). The Changning-Menglian Paleo-Tethys subducted beneath the Simao Terrane, resulting in the formation of the Yunxian-Jinggu Arc. This Arc was intruded by the Lincang granitic pluton, which consists of a peraluminous S-type granite with U‒Pb zircon aging from 248 to 203 Ma (Peng et al., 2008; Dong et al., 2013; Peng et al., 2013). It is worth noting that the Eocene‒Oligocene potassic porphyry intrusions are large and widespread along the Jinshajiang-Red River deep-crustal fault zone in the Eastern Qiangtang block (Yang et al., 2014).

The Baoshan-Tengchong (BT) terrane consists of the Baoshan subterrane and Tengchong subterrane, which were once located at the northern margin of Gondwana and accreted to the Eurasian continent in the late Mesozoic (Metcalfe, 2006; Jin et al., 2014) (Figure 1). The Baoshan subterrane experienced a similar trend, but the accretionary time might have been in the late Paleozoic to early Mesozoic (Metcalfe, 2013). The Changning-Menglian suture located to the east of the BT might have preserved the subduction of the Proto-Tethys and Paleo-Tethys oceanic plates (Deng et al., 2014a; Xu et al., 2021). This suture correlates with the Longmu Tso-Shuanghu Suture in eastern Tibet.

The South China Craton (SCC) is a block with an Archean history (Gao et al., 1999). Continuous subduction and mantle plume activities following the breakup of the Columbia supercontinent resulted in large volumes of Neoproterozoic volcanic rocks (Xu et al., 2021). The Jinshajiang-Ailaoshan suture, which is accompanied by Eocene–Oligocene alkaline magmatic rocks along the western margin of the SCC (Figure 1), was a result of the India-Asia continental collision during the Cenozoic and lithospheric thickening (Wang et al., 2016; Xu et al., 2021).

Previously published Nd isotopic data from 220 samples were used to evaluate the crustal evolution in this region over time based on existing zircon U‒Pb ages or other isotopic ages, such as Ar–Ar or K–Ar ages. To produce a data set for the SOB, a consistent method was used to recalculate the data (Maboko and Nakamura, 1996). The specific calculation methods and parameters are as follows (Jacobsen and Wasserburg, 1980; Wu et al., 2002). ϵ N d t = ( 143 N d / 144 N d i / ( 143 N d / 144 N d ) C H U R − 1 ] × 10 4 ( 143 N d / 144 N d ) i = ( 143 N d / 144 N d ) − ( 147 S m / 144 N d ) e λ t − 1 f S m / N d = ( 147 S m / 144 N d ) / ( 147 S m / 144 N d ) C H U R − 1 T D M 1 = 1 λ ln ⁡ ⁡ 1 + ( 143 N d / 144 N d ) − 0.51315 ( 147 S m / 144 N d ) − 0.2137 T D M 2 = T D M 1 − T D M 1 − t f C C − f S / f C C − f D M where λ=6.54×10⁻¹², (¹⁴³Nd/¹⁴⁴Nd)_CHUR=0.512638, (¹⁴⁷Sm/¹⁴⁴Nd)_CHUR=0.1967, f_CC=−0.4, and f_DM=0.08592.

The data that were collected and recalculated are listed in Supplementary Table S1. The age-corrected initial Sr isotopic ratios showed large variations from 0.700899 to 0.760444 (Figure 3A), but most were between 0.703 and 0.715 (Figure 3B). The ε_Nd(t) values varied from −14.56 to 5.94, and most were <0 and >−10 (Figure 3). Positive ε_Nd(t) values from the EQST and SCC terranes were observed to a large extent in mafic rocks, followed by felsic dikes.

Additionally, most samples had one-stage model ages (T_DM) between 0.5 and 3.2 Ga, with f_Sm/Nd values ranging from −0.68 to 0.75, and most rocks had f_Sm/Nd values between −0.2 and −0.6 (except 55 samples) (Figure 4). The model ages generally become older with increasing f_Sm/Nd, which shows a linear correlation. Notably, the T_DM of the rocks from the syn-collisional stage was usually younger than that from the pre-collisional stage. As shown in Figure 5, T_DM2 was applied only to samples with values of f_Sm/Nd > −0.2 and < −0.6 (Wang et al., 2009). The model age increased with decreasing ε_Nd(t) values, which suggests a linear correlation.

Since the dataset was small, it was reasonable to apply the inverse distance weighted interpolation method, which uses the 12 nearest neighbors at a “power” (Mole et al., 2014; Hou et al., 2015), in ArcGIS to contour the ε_Nd(t) and T_DM maps (Wang et al., 2009; Mole et al., 2014; Wang et al., 2016). During the data processing, we attempted to minimize the subjective influence caused by manual screening to show the original distribution of the data as much as possible. All Nd isotope data were classified by the geometric interval method in ArcMap to ensure that the variation between intervals was fairly consistent and that the amount of data in each class range was approximately the same. Particularly, regarding the samples with different Nd isotope values in the same geographical coordinates, the arithmetic mean values were used as the points for mapping to avoid the bias caused by random selection in ArcGIS. As a result, 199 Nd isotopic values were used for mapping. The ε_Nd(t) values of rocks in the SOB were mostly negative, while positive values were observed in the southern, northern, and central parts of the EQST. The most negative values of ε_Nd(t) occurred in the BT and SCC, and the other areas contained ε_Nd(t) values between −9.3 and −1.9 (Figure 6). Similarly, the youngest Nd model ages (0.58–0.87 Ga) occurred in the same position as the ε_Nd(t) values in the EQST, and the oldest (2.29–3.03 Ga) was found in the SCC and BT, while the other Nd model ages were mostly between 2.29 and 1.17 Ga (Figure 7).

Hf–Nd isotopic data were used to distinguish between juvenile and old crustal components (Wang et al., 2009; Mole et al., 2013; Wang et al., 2016; Granseth et al., 2021; Xu et al., 2021). Juvenile crust, with isotopic values that plot on or close to the depleted mantle evolution line, is regarded as crustal material that is generated directly from the depleted mantle or remelted from material recently extracted from the depleted mantle. In contrast, the old or reworked crust refers to the preexisting crust that was remobilized by partial melting and/or erosion with sedimentation (Belousova et al., 2010; Hawkesworth et al., 2010; Hou et al., 2015). The parent and daughter elements of the Sm-Nd isotope system are both rare earth elements (REEs) and generally behave similarly, which can be used to effectively suggest crustal processes and the source of the rocks in question (DePaolo, 1988; Champion, 2013). During the formation of juvenile crust, the Earth’s crust is more enriched in Nd and Sm and has lower Sm/Nd ratios than the complementary depleted mantle reservoir as a result of lanthanide contraction. In other words, ε_Nd(t) values vary over time because of the different Sm/Nd ratios of mantle and crustal reservoirs (Champion, 2013). The values of ε_Nd(t) and Nd model ages are tools for distinguishing the possible sources of magma and the formation age of crustal source rocks separately (Hou and Wang, 2018) (Figure 8). A comparison of Hf isotopic mapping from previous studies (Figure 9) and Nd isotopes are discussed below.

Mo et al. (1993) stressed that the Zhongzan block, sandwiched between the Ganzi-Litang Suture and Jinshajiang Suture, was a breakaway part of the SCC. The old Nd model ages of granites occurred in the ZYC, which is connected to the upper middle part of the EQST through the Jinshajiang Suture (Figure 7); these granites are mostly near the Yidun Arc, with intrusion ages of 103.7 Ma to 75.2 Ma from the north to the south (Qu et al., 2002), and older granites, such as the Daocheng granite, have an age of 216 Ma (He et al., 2013). The samples from the EQST also have an age of ca. 245 Ma (Wang et al., 2014). The old Nd model age and low ε_Nd(t) values may suggest that these felsic rocks originated from the same ancient crust of the SCC basement.

Many metal deposits occur in the SOB to form several metallogenic belts. Figures 6–8 show a strong empirical correlation between the different lithospheric architectures and the locations of distinct ore deposits. All porphyry Cu deposits are confined to juvenile crustal blocks, whereas skarn and hydrothermal vein-type W–Sn deposits are located in the low-ε_Nd(t) area, while skarn and hydrothermal vein-type Pb‒Zn–Cu–Ag deposits occur between the two types of deposits (Figure 6).

The Cenozoic porphyry-skarn Cu–Mo–(Au) deposits (Wang et al., 2005; Hou et al., 2006; Lu et al., 2013) occur in the juvenile crust within the EQST or along the suture with high ε_Nd(t) (−3.3–0) and young T_DM ages (0.58–1.17 Ga). Representative porphyry Cu–Mo–(Au) deposits in the SOB are the Machangqing deposit [39 Mt ore, with 0.64% Cu and 56 Mt with 0.08% Mo (Lu et al., 2013)], the Beiya Au deposit [26 Mt ore, with 2.26 g/t Au, (Lu et al., 2013)], and many smaller deposits, such as the Changan, Tongchang, and Yao’an deposits. These deposits are generally associated with potassic-enriched intrusions that were formed in the post-collisional stage (Lu et al., 2013; Lu et al., 2013; Lu et al., 2015). However, no significant porphyry Cu deposits were found in regions with low ε_Nd(t) values and old T_DM ages. This distribution pattern also occurs in Lhasa Terrane, as revealed by Hf isotopic mapping (Hou et al., 2015). This comparison indicates that the juvenile lower crust might have exerted first-order control on the formation of the porphyry Cu deposits. In the SOB, the juvenile component is mainly derived from the post-collisional stage and characterized by high-K calc-alkaline signatures with high Sr/Y values and ε_Nd(t) values (e.g., Lu et al., 2013). Arc magma interacted with the crust in an open environment; then, the fO₂ of magma decreased, which led to the accumulation of sulfides in the juvenile lower crust. At the post-collision stage, an upwelling of the asthenosphere triggered the melting of metal sulfide-rich crust, which resulted in the release of Cu into the magmatic system as the material source of Cu deposits (Hou et al., 2015). Furthermore, as representatives of the cumulates or residuals of Neoproterozoic arc magmas, the amphibolite xenoliths hosted by the Cenozoic stocks are enriched in Cu and Au (Hou et al., 2017), which reinforces the conclusion that the Cu–Au-enriched low-crustal cumulates might be the metal source of the deposits, so the Cu–Mo–(Au) deposits typically occur at the cratonic edges or sutures.

The W–Sn deposits are mainly located in the BT areas, with negative ε_Nd(t) (−13∼−6) and ε_Hf(t) values, and form the Tengchong-Lianghe Sn metallogenic zone. Previous studies have shown that the occurrence of these deposits tends to occur in the Late Cretaceous to early Eocene (Jiang et al., 2012; Chen et al., 2014; Cao et al., 2017; Sun et al., 2017), which is in concordance with the collisional timeline. A representative deposit, the Lailishan deposit, with an Ar-Ar age of hydrothermal muscovite of 50.4 Ma (Cao et al., 2017), is a large tin deposit located in the midwestern region of Tengchong. The tin orebodies occur in the lower contact zones and the surrounding fractured zones of the granitic intrusion, with a weighted mean U–Pb age of 50.6 Ma (Hou et al., 2007; Cao et al., 2017). The Lailishan granite is characterized by high K, F, and S contents and a high initial ⁸⁷Sr/⁸⁷Sr ratio (Lu and Wang, 1993), and it is rich in Al (Al₂O₃= 14.70%–15.27%), alkalis, Ca, REEs, and Ba (Sun et al., 2017), which suggests that the granites are S-type and related to the crustal anatexis. The ε_Nd(t) values are most negative when the T_DM ages are the oldest. Both the geochemical characteristics and the results of Nd isotopic mapping suggest that the related igneous rocks are likely derived from the partial melting of predominantly ancient crustal material, which is consistent with the conclusion that S-type magmas are geographically related to W–Sn mineralization (Wang et al., 2014). Correspondingly, the enrichment and mineralization of W–Sn are not only related to the high differentiation of granitic magma but also restricted by the composition of the source area. The origin of W–Sn deposits is the partial melting of the sedimentary rocks, which enriched the metal composition in the muscovite and biotite (Romer and Kroner, 2015). The age of the intrusion is within the syn-collision stage, which means the continental collision and subsequent crustal thickening in the BT resulted in the formation of S-type granites (Zhu et al., 2018) and related W-Sn deposits in the region (Sun et al., 2017).

Skarn and hydrothermal vein-type Pb‒Zn‒Cu‒Ag deposits, such as the enormous Jinding Pb‒Zn deposit (200 Mt ore with 6.1% Zn and 1.3% Pb) (Xue et al., 2003; Leach et al., 2017) and the Baiyangping Pb‒Zn‒Cu‒Ag deposit (Xue et al., 2000; Zou et al., 2015), occur within the EQST with relatively low ε_Nd(t) and old T_DM ages. The ⁴⁰Ar–³⁹Ar plateau age of silicification quartz in the Baiyangping deposit is 62.7 Ma (Xue et al., 2003), and different age dating methods and samples provide varied ages for the Jinding deposit between 129 and 21 Ma (e.g., Yalikun et al., 2018; Wang et al., 2022). However, almost all the Pb‒Zn–Cu–Ag deposits that occur in the area have the Nd isotopic characteristics of old crust, which indicates that the metal source was likely to be crustal materials. Similarly, geochemical and other isotopic characteristics, such as Pb isotope results (Hao et al., 2017), also support this conclusion. It is worth noting that the formation of the Pb‒Zn‒Cu‒Ag deposits is controlled by trust-nappe structures (Hou et al., 2008).

In summary, the continent-continent collision of India and Asia resulted in the formation of distinct metal deposits based on the different locations and sources of partial melting. The juvenile crust was the base of porphyry-skarn Cu–Mo–(Au) deposits, and the melting of old crust or upper crust contributed to the occurrence of W–Sn deposits and skarn and hydrothermal vein-type Pb‒Zn‒Cu‒Ag deposits controlled by trust-nappe structures (Figure 10).