Genesis of Volcanic Rocks in the Zijinshan Ore District, SE China: Implications for Porphyry-Epithermal Mineralization

Volcanic rocks, as the extrusive counterparts of the mineralized intrusions, can provide important information on the magma source, petrogenesis, and metallogenic conditions of the coeval porphyry-epithermal system. Shanghang Basin volcanic rocks are spatially and temporally related to a series of adjacent porphyry-epithermal Cu–Au deposits, and they can be used as a window to study the related deposits. Two laser-ablation–inductively coupled plasma–mass spectrometry zircon U–Pb analyses of the volcanic rocks yield weighted mean ages of ~105 Ma, identical to the age of the coeval porphyry-epithermal mineralization. Rocks have SiO2 contents of 55.4 to 74.8 wt % and belong to the high-K to shoshonitic series, characterized by strong differentiation of light rare-earth elements (REEs) relative to heavy REEs (mean LaN/YbN = 16.88); enrichment in light REEs, Rb, Th, and U; and depletion in Nb, Ta, Zr, Hf, and Ti. The volcanic rocks display (Sr/Sr)i values of 0.709341 to 0.711610, εNd(t) values of −6.9 to −3.3 εHf(t) values of −3.95 to −0.30, and δ18O values of 6.07% –6.79% , suggesting that the parental magmas were derived from a mantle source enriched by subduction-related progress. SiO2 content shows a strong negative correlation with the contents of some major and trace elements, indicating that fractional crystallization played an important role in the generation of these rocks. A binary mixing model of Hf–O isotopes gives an estimated degree of crustal contamination of 30%. In addition, magnetite crystallized early, and the samples showed high zircon EuN/EuN* values (0.48–0.68), indicating that the parental magma had a high oxygen fugacity. The inferred suppression of plagioclase crystallization and increasing hornblende crystallization during magma evolution suggest that the magma was water rich. The high-water content and high oxygen fugacity of the magma promoted the dissolving of sulfides containing Cu and Au in the source area and contributed to the migration of ore-forming elements.


Introduction
The Zijinshan ore district is located in southwestern Fujian Province, southeastern China ( Figure 1) and is one of the most important copper-, gold-, and molybdenum-producing areas in the country. Proven reserves of gold, silver, copper, and molybdenum exceed 400 tons, 6000 tons, 4.1 million tons, and 0.11 million tons, respectively [1]. It is generally considered that the copper-gold-molybdenum mineralization was associated with large-scale magmatic and hydrothermal activities during the Late Mesozoic, with most of these deposits having formed during the Cretaceous [2][3][4][5]. The Zijinshan ore district is widely regarded as a typical porphyry-epithermal metallogenic system [6][7][8][9]. The district includes the Zijinshan high-sulfur epithermal copper-gold deposit, the Yueyang low-sulfur epithermal silver-gold-dominated polymetallic deposit, and the Luoboling porphyry copper-molybdenum deposit ( Figure 2). Previous studies have focused mainly on the mineralization process of the deposit and the structure of the ore field [7,[12][13][14][15][16], whereas little attention has been paid to the source, nature, and evolution of the parental magma from which these deposits were formed. Recent studies have suggested that the Luoboling porphyry copper-molybdenum deposit and the Zijinshan copper-gold deposit were produced by discrete magmatic-hydrothermal systems [17], but the nature of the ore-forming magma associated with the Zijinshan copper-gold deposit is still unclear. However, in the Shanghang Basin, which lies adjacent to the southwest of the Zijinshan ore district (Figure 2), the porphyry-epithermal Cu-Au deposits are spatially and temporally related to the volcanic rocks of the Shimaoshan Group [18]. The volcanic rocks are considered as the extrusive magmatic counterpart of the mineralized rocks, and their study should therefore provide important new information on the magma source, petrogenesis, and metallogenic conditions of the porphyry-epithermal system at Zijinshan.
Shimaoshan Group volcanic rocks are exposed in parts of the Zijinshan ore district but are distributed mainly in the Shanghang Basin ( Figure 2). Two periods of magmatic activity occurred in the vicinity of the Zijinshan ore district, at~165-150 Ma and~113-95 Ma [5,[19][20][21]. The ages of formation of these deposits are well established, including the Zijinshan epithermal copper-gold deposit at 102.86 ± 0.61 Ma [17] and the Luoboling porphyry copper-molybdenum deposit at 104.9 ± 1.6 Ma [2]. Clearly, there is a close temporal and spatial relationship between the Shimaoshan Group volcanic rocks and deposits of the Zijinshan ore district. In this paper, we report new zircon U-Pb ages; trace elements, Hf-O isotopes, and whole-rock major; and trace elements and Sr-Nd isotopes of Shimaoshan Group volcanic rocks to better constrain the nature of the magma (including mantle sources) and mineralization conditions (including magma water content and oxygen fugacity) associated with the formation of the Zijinshan ore district.

Geological Background and Sampling
Southeastern China comprises the Cathaysia Block in the southeast and the Yangtze Block in the northwest [22,23]. According to the different crustal evolutions on opposing sides of the Zhenghe-Dapu Fault, the Cathaysia Block can be further divided into the Western Cathaysia and Eastern Cathaysia regions [24,25]. Magmatic activity in southeastern China during the Late Mesozoic was concentrated mainly in the Cathaysia Block, where it migrated seaward over time and increased in intensity [10,26,27]. Volcanic-intrusive magmatism during the Yanshanian (65-220 Ma) produced mainly felsic rocks with some basalts and rare intermediate rocks [28][29][30]. Compared with the Jurassic, there are fewer Cretaceous granitoid rocks but more abundant volcanic rocks [10,30,31]. Jurassic volcanic rocks and granitoids are distributed mostly in Western Cathaysia, whereas Cretaceous basaltic rocks and granitoids appear mainly in Eastern Cathaysia [4,32] (Figure 1a). Fujian Province, located in the eastern part of southeastern China, is divided into three tectonic belts by the Changle-Nanao and Zhenghe-Dapu faults. From east to west, these belts are the Pingtan-Dongshan Metamorphic Belt, the Yanshanian Magmatic Belt, and the Early Paleozoic Fold Belt [20,33,34] (Figure 1b). The Cretaceous volcanic rocks in Fujian Province can be divided into upper series named the Shimaoshan Group and lower series named the Moshishan Group, with most of the volcanic rocks in the Shanghang Basin belonging to the Shimaoshan Group. The Shimaoshan Group comprises the Huangkeng Formation and the overlying Zhaixia Formation, with each formation being further divided into upper and lower members.
The Huangkeng and Zhaixia formations of the Shimaoshan Group show substantial variation in thickness between areas of tectonic subsidence and areas of tectonic uplift [21]. The thickness of the Huangkeng Formation in areas of tectonic subsidence reaches 700 m, whereas in areas of tectonic uplift, the rocks of this formation are scattered across the Ermiaogou, and Zijinshan ore districts with a thickness of only tens to a few hundreds of meters. Rocks of the lower member of the Huangkeng Formation comprise mostly conglomerate, glutenite, and siltstone, mixed with purple-gray dacite in the middle, and have a total thickness of about 230 m. The lower member unconformably overlies the Zijinshan Complex in the Yueyang mining area and the Louziba Group in the Ermiaogou mining area. The upper member of the Huangkeng Formation is composed of purple-gray andesite, andesitic breccia, light-gray dacitic crystal tuff, and dark-gray andesite basalt, with a total thickness of >350 m [35]. The Zhaixia Formation is distributed mainly in the southwest part of the Zijinshan ore district. Its lower member is composed of thin layers of purple-red silty mudstone intercalated with purple-red rhyolitic sedimentary tuff and rhyolitic crystal tuff, with a total thickness of up to 372 m. The upper member consists mostly of gray-white rhyolite, purple-red volcanic breccia, and rhyolitic breccia tuff, with a total thickness of up to 1037 m [35].
The ages of volcanic rocks of the Shimaoshan Group are well established [21,36,37]. By combining multiple dating methods (such as zircon LA-ICP-MS, whole-rock Rb-Sr isochron, zircon SHRIMP), estimates of the formation age of the Huangkeng Formation volcanic rocks are clustered within 93-125 Ma and those of the Zhaixia Formation within 93-121 Ma [37]. The formation age of Shimaoshan Group volcanic rocks was constrained to 104-94 Ma by using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) zircon U-Pb dating [36]. A whole-rock Rb-Sr isochron age of rhyolite in the Zhaixia Formation of 94 ± 7.7 Ma and an age of dacite in the Huangkeng Formation of 125 ± 9.8 Ma were reported [38]. Zircon U-Pb dating of purplish-red rhyolite and breccia from the Huangkeng Formation has yielded ages of 110.1 ± 0.7 Ma and 113.0 ± 1.9 Ma, respectively [21]. Zircon U-Pb dating of dacite and tuff in the Zijinshan ore district has given ages of 110 ± 1 Ma and 111 ± 1 Ma, respectively [19]. Zircon U-Pb ages were obtained for limestone and rhyolite in the Yueyang mining area of 105 ± 1 Ma and 102 ± 1 Ma, respectively [39]. The above data together indicate that Cretaceous magmatism in the Zijinshan ore district lasted for about 20 Myr.
Samples for this study were obtained from the Zijinshan ore district and the Shanghang Basin. The Shanghang Basin is controlled by a NW-trending fault with an area of 100 km 2 . The southwestern and northeastern margins of this basin are controlled by deep NW-trending faults ( Figure 2) [18]. The Shimaoshan Group, mainly the upper member of the Huangkeng Formation, is exposed in the Shanghang Basin with a thickness of >350 m. Most of the samples for this study were collected from drill cores (ZK5710 and ZK5903), with a smaller number of samples being obtained from surface exposures. A total of 62 rock samples were collected, combining the characteristics of hand specimens and the characteristics of major elements; we divided them into three categories, namely, trachyte, trachyandesite, and rhyolite-dacite, respectively. Most of the volcanic rock samples have undergone variable degrees of alteration. We have selected a more representative sample for a brief introduction from each category. Rhyolite is pink with a patchy structure (Figure 3a), in which the phenocrysts are mainly alkaline feldspar and the matrix is cryptocrystalline (Figure 3b). Trachyte is dark gray with a patchy structure and contains phenocrysts of mainly feldspar and subordinate amphibole (Figure 3c,d). Trachyandesite is dark gray, with phenocrysts of mainly plagioclase and subordinate magnetite in a matrix of grains with a preferred orientation (Figure 3e,f).

Zircon U-Pb Dating and Trace-Element Analyses
Conventional density and magnetic methods were used to separate zircon grains from two samples (SH1604, trachyte and ZK5710-30, trachyte), following which grains were hand-picked under a high-resolution optical microscope. Representative zircon grains were set in epoxy resin and then polished to expose their cores. Cathodoluminescence (CL) imaging was used to examine zircon morphology and internal textures and to select analysis locations. Zircon U-Pb dating and trace-element analyses were carried out concurrently by LA-ICP-MS at the State Key Laboratory of Ore Deposit Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences (SKLODG, IGCAS), Guiyang, China. A GeolasPro 193 nm ArF excimer laser (Coherent, Gilching, Germany) was used to ablate zircons, and an Agilent 7900 ICP-MS instrument (Agilent, Santa Clara, CA, USA) provided ion signal intensities. A mixture of helium and argon was used as the carrier gas to transport the aerosol to the ICP-MS instrument. Zircon 91500 was used as an external standard, and reference materials NIST 610, BIR-1G, BCR-2G, BHVO-2G, PL-1, and QINGHU were used to calibrate the trace-element data. Adopted values of these reference materials are from the GeoReM database [40]. Each sample analysis involved 18 s background acquisition and 50 s data collection. Operating conditions included a laser beam diameter of 32 µm, a repetition rate of 5 Hz, and an energy flux of 8 J/cm 2 . ICPMSDataCal software (Version 11.8, Sample Solution, Wuhan, China) was used to process the U-Pb isotopic and trace-element data [41]. Si was used as an internal standard, and multiple elements were used as external standards to calibrate analyzed element contents. Isoplot 4.15 software (Berkeley Geochronology Center, Berkeley, CA, USA) was used to construct concordia plots and calculate weighted mean ages [42].

Whole-Rock Major-and Trace-Element and Sr-Nd Isotope Geochemical Features
Rock samples were cut into small pieces and fresh parts were selected and ground into powder for whole-rock geochemical analyses. An Axios (PW4400) X-ray fluorescence spectrometer (Thermo Fisher, Waltham, MA, USA) was used to determine contents of major elements at the SKLODG, IGCAS. Loss-on-ignition (LOI) values were measured by heating 1 g sample powder to 1100 • C for 1 h.
Trace-element analyses were also conducted at the SKLODG, IGCAS, by using an ICP-MS instrument (PE DRC-e) produced by PerkinEmer company (Waltham, MA, USA). For each analysis, a 50 mg subsample of rock powder was dissolved at 190 • C in a mixture of HF and HNO 3 , and placed in a PTFE-lined stainless-steel bomb for 48 h. To monitor signal drift during counting, Rh was applied as an internal standard. International standards GBPG-1, AGV-2, and AMH-1 were used to monitor analytical quality, with analytical accuracy generally better than ±10%. Details of the experimental procedures followed are given by [43].
A TRITON thermal-ionization mass spectrometer (Thermo Fisher, Waltham, MA, USA) was used to determine Sr-Nd isotope compositions of whole-rock samples at the SKLODG, IGCAS. Samples were dissolved by using the same method as that used for trace elements. Cationic ion-exchange procedures were applied to effectively separate Sr and Nd in solution [44]. Sr and Nd isotopic ratios were normalized with respect to 86 Sr/ 88 Sr = 0.1194 and 146 Nd/ 144 Nd = 0.72419. Details of the experimental method and procedure are given by [45].

Zircon O Isotope Analyses
Determinations of zircon O isotopic composition were conducted by a Cameca IMS-1280HR secondary-ion mass spectrometer (Camera, Paris, France) at the SIMS laboratory in the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou, China. Analytical procedures followed those described by [46]. Under conditions of 10 kV and an intensity of~2 nA, the Cs + primary ion beam was accelerated and then rasterized over a 10 µm area. Analysis spots had a diameter of 20 µm. The width of the inlet gap of the multicollector Faraday cup (FC) of 16 O and 18 O was 120 µm, and the width of the outlet gap was 500 µm. The intensity of 16 O was~1 × 10 9 counts per second. In multicollector mode, two off-axis Faraday cups were used to measure O isotopes. Each analysis duration included 120 s for pre-sputtering, 60 s for automatic beam formation, and 40 s for O isotope integration. Vienna Standard Mean Ocean Water (VSMOW) composition was used as the standard to normalize δ 18 O values. Standard zircon 91500 was used to correct the instrumental mass fractionation factor, with δ 18 O VSMOW = 9.9% [47].

Zircon Lu-Hf Isotope Analyses
In situ analyses of Hf isotopic compositions of zircon were performed using LA-MC-ICP-MS at the SKLODG, IGCAS. Analyses of Hf were conducted on the same spots as used for O isotope analyses. A RESOlution S-155 LA system (Australian Scientific Instruments, Canberra, Australia) and a Nu Plasma III MC-ICP-MS instrument (Nu Instruments, Wrexham, Wales, UK) were used together for the analyses. After homogenization of the beam-delivery system, the 193 nm ArF excimer laser was focused on the surface of the zircon with an energy density of 6.0 J/cm 2 . Laser ablation involved a spot diameter of 40 µm, a repetition frequency of 6 Hz, and an ablation duration of 40 s. Helium was used as a carrier gas to transport the aerosol to the Nu Plasma III MC-ICP-MS instrument (Nu Instruments, Wrexham, Wales, UK). For quality control, five standard zircons (GJ-1, 91500, Plešovice, Mud Tank, and Penglai) were processed for every 30 unknown analyses, and one standard zircon (Penglai) was processed for every five unknown analyses. The influence of mass deviation on Hf was corrected according to an exponential law by using the value of 179 Hf/ 177 Hf = 0.7325 [48]. The isobaric interference of 176 Lu on 176 Hf was corrected by using the recommended 176 Lu/ 175 Lu ratio of 0.02655 [49] to calculate values of 176 Lu/ 177 Hf, and the isobaric interference of 176 Yb on 176 Hf was corrected by using the recommended 176 Yb/ 172 Yb ratio of 0.5887 [50] to calculate values of 176 Yb/ 177 Hf.

Zircon U-Pb Dating and Trace Elements
The U-Pb isotope data are listed in Supplementary Table S1, and representative zircon CL images and U-Pb concordia diagrams are shown in Figures 4

Whole-Rock Major and Trace Elements
Sixty-two volcanic rock samples from the Shanghang Basin were analyzed for major-and trace-element compositions, with results given in Supplementary Table S2. Sample contents were standardized to 100% after accounting for LOI. These samples have SiO 2 contents ranging from 55.   represents primitive arc melt [52]. The dotted line represents the field of experimental crust melts, data from [36]. the solid line represents the field of metabasaltic and eclogite experimental melts at 1-4 GPa, data from [53].

Whole-Rock Sr-Nd Isotopes
Sr-Nd isotopic compositions of the studied volcanic rocks are presented in Table 1 and Figure 9a. Initial 87 Sr/ 86 Sr ratios and ε Nd (t) values were calculated according to the formation age determined by zircon U-Pb dating. The volcanic rocks have ( 87 Sr/ 86 Sr) i ratios between 0.708901 and 0.711610 and ε Nd (t) values between −6.9 and −3.3. The higher ( 87 Sr/ 86 Sr) i values of the volcanic rocks were attributed to post-magmatic alteration, since the Rb and Sr contents of the rocks has been modified by alteration. As a result, the compositions of Sr isotope are excluded in the following discussion.
According to these data, the calculated two-stage depleted-mantle Nd model ages (T DM2 ) range from 1.47 to 1.19 Ga (Table 1).

In Situ Zircon Hf-O Isotopes
One sample (ZK5710-30) was selected for analysis of zircon Hf-O isotopes, due to their high magmatic zircon content. Results are presented in Table 2 and Figure 9b. The ε Hf (t) value and the two-stage depleted mantle Hf model age were calculated according to the age obtained by zircon U-Pb dating. Twenty-one analyses for sample ZK5710-30 yielded 176 Hf/ 177 Hf ratios between 0.282597 and 0.282701, and the calculated ε Hf (t) values range from −3.95 to −0.30, corresponding to two-stage Hf model ages of 1.47 to 1.19 Ga. Fifteen O isotope analyses were obtained with δ 18 O values ranging from 6.07% to 6.79% , which are higher than the mantle value of 5.3% ± 0.3% (Figure 10).

Timing of Volcanism and Relationship to Porphyry-Epithermal Mineralization
In the Zijinshan porphyry-epithermal metallogenic system, the direct ore-forming magma of the epithermal deposits has not been discovered. However, Shimaoshan Group volcanic rocks are distributed extensively in and around the ore district ( Figure 2). Understanding the relationship of the volcanism in the Shanghang Basin to porphyry-epithermal mineralization at Zijinshan can help to understand the nature of the metallogenic magma source area and the controls on metallogenesis. Our zircon U-Pb dating results show that volcanism in the Shanghang Basin occurred at~105 Ma, synchronous with the metallogenic age of the Zijinshan Cu-Au deposit (103 ± 1 Ma, [67]) and the Luoboling porphyry Cu-Mo deposit (104.9 ± 1.6 Ma, [2]; 104.6 ± 1.0, [5]). The Luoboling granodiorite porphyry and the studied volcanic rock samples partially overlap in the K 2 O vs. SiO 2 diagram, mainly in the trachyte-dacite-rhyolite field (Figure 6a). However, the volcanic rocks have a wider range of SiO 2 (55-75 wt %), and the Luoboling porphyry are all acidic (63-73 wt %) (Figure 6a), indicating that the magma that formed the porphyry underwent a higher degree of evolution relative to that of the volcanic rocks. Both the studied volcanic rocks and the Luoboling porphyry belong to the high-K calc-alkaline and shoshonitic series in a Na 2 O + K 2 O vs. SiO 2 diagram (Figure 6b). Moreover, the volcanic rocks have similar REE and trace-element patterns to those of the Luoboling porphyry ( Figure 8). These features, together with the similar ages and the close temporal and spatial relationships between the volcanic rocks and porphyries, indicate that the volcanic rocks and the magma were derived from a common magma chamber beneath the volcanic basin and underwent similar fractionation pathways.

Magma Source of the Volcanic Rocks
The Shimaoshan Group volcanic rocks in the Shanghang Basin show high MgO contents and high Mg # values relative to those of experimental crustal melt (Figure 7b), suggesting that the volcanic magmas cannot have originated from partial melting of crustal materials. The two-stage Hf model ages of 1.41 to 1.18 Ga (Table 2) are substantially younger than those of basement metamorphic rocks of the Cathaysia Block (>1.85 Ga, [25]) (Figure 9b), indicating that the crust was not the only source of the volcanic rocks. In addition, the Sr-Nd isotope data of the studied samples differ substantially from those of Neoproterozoic and Meso-Paleoproterozoic basement rocks of the Cathaysia Block (Figure 9a), with two-stage Nd model ages of 1.47 to 1.19 Ga ( Table 1) that are much younger than those of basement metamorphic rocks of the Cathaysia Block (>1.74 Ga, [68]). Experimental studies on melting have shown that continental crust melts are usually sodium rich [69,70], but our samples are high-K calc-alkaline and shoshonitic. Therefore, Shimaoshan Group volcanic rocks were not derived from a predominant crustal source.
The enriched Sr-Nd-Hf isotopic features of the samples, together with the older Hf and Nd model ages (1.47-1.19 Ga), indicate that the parental melts of the studied volcanic rocks were derived from enriched mantle or a mixture of mantle-derived magma and crust-derived magma. Nb/Ta values can serve as an important indicator of crust-mantle interaction [71]. Continental crust has an average Nb/Ta ratio of 11.43 [62] and primitive mantle of 17.8 [72]. The Shanghang Basin volcanic rocks show Nb/Ta ratios that range from 7.03 to 18.18, with a mean value of 10.11, which is near the value of average continental crust. The low Nb/Ta ratios cannot fully interpret as the result of crustal contamination, since an unreasonable proportion of crustal materials is required to cause such a low ratio. All the Cretaceous mafic plutons in the Fujian coastal area of southeastern China have a narrow range of negative ε Nd (t) values (−1.3 to −3.3; [73]). This, together with the mantle-like δ 18 O value of these mafic plutons [55] indicates that these rocks reflect the characteristics of a mantle source and that crustal assimilation was negligible during magma ascent. However, the volcanic rocks are more enriched in Sr-Nd compared with the mafic plutons, consistent with minor crustal contamination of the parental magma.
Zircon O isotopes can effectively distinguish crustal contamination [74] because mantle-derived magma and upper-crust-derived magma have distinct O isotope characteristics, with δ 18 O values of 5.3% and 10% -30% , respectively [75]. δ 18 O values of our samples vary from 6.07% to 6.79% (Table 2) and are much higher than the mantle δ 18 O range (Figure 10a). This, combined with the negative correlation between zircon δ 18 O and ε Hf (t), indicates that the volcanic rocks in the Shanghang Basin were derived from enriched mantle but also affected by some crustal contamination. We performed quantitative modeling on the basis of the zircon δ 18 O and ε Hf (t) isotopic compositions by magma mixing of two end members to estimate the degree of crustal contamination. The Cretaceous gabbro near the study area and South China S-type granite were selected as the two end members, representing the mantle and crust components, respectively. The final composition of the mixture depends on the ratio of the two end members, each curve represents a mixed case, and the curvature of the curve is controlled by Hf m /Hf c. The results indicate that the degree of crustal contamination is <30% (Figure 10a).
The samples show strong fractionation between LREEs and HREEs, enrichment of LILEs, and depletion in HFSEs, with negative Nb-Ta and Ti anomalies (Figure 8d-f). These characteristics are consistent with those of non-crust-contaminated Cretaceous mafic suites in the Cathaysia Block [55,76]. Slab-derived fluids or subducted sediments can enrich mantle sources, giving rise to enrichment in LILEs and depletion in Nb, Ta, and Ti [77], consistent with our results (Figure 10b). The studied volcanic samples have very similar Nd isotope characteristics to those of the Cretaceous gabbro in the Fujian coastal area (Figure 9a) [55], further suggesting a similar mantle derivation with the gabbro. Moreover, their high La/Nb and Ba/Nb ratios are consistent with the characteristics of arc volcanic rocks (Figure 10c), which are generally considered to be related to subduction [78]. In addition to the above indicators, the ratios of incompatible trace elements with similar distribution coefficients, such as Nb/U, Nb/Ta, and Ce/Pb, can also be used to distinguish the source region, as they show negligible change during the melting process [79,80]. The Nb/U values (1.49-7.58) of the studied samples are much lower than those of MORB and OIB (Figure 10d), indicating that these rocks were metasomatized by subduction-related hydrous fluids. Thus, we suggest that these volcanic rocks were derived from mantle material enriched by input of subduction-related materials.
The extensive magmatic activity during the Cretaceous in the Cathaysian Block is generally considered to have been the result of westward subduction of the Paleo-Pacific Plate [10,81], which could account for the inferred enrichment of mantle material. The Zijinshan ore district volcanic rocks share similar REE and trace-element patterns and Sr-Nd-Hf isotope characteristics with typical arc volcanic rocks, suggesting a subduction affinity. In addition, the volcanic rocks represent high-K calc-alkaline to shoshonitic lavas, which are different from the intermediate, low-K to calc-alkaline rocks in modern arc environments [82]. High-K, calc-alkaline to shoshonitic rocks are also found in lithospheric extensional settings [83]. Accordingly, the characteristics and age of volcanic rocks of the Shimaoshan Group suggest that the coastal region of southeastern China was situated in a lithospheric extensional setting during the Cretaceous. It has been proposed that the regional tectono-magmatic evolution in southeastern China during the late Yanshanian was controlled by break-off and rollback of the subducting paleo-Pacific Plate [73]. Thus, we suggest that this geodynamic process caused regional extension and partial melting of the mantle lithosphere in the Zijinshan district and the Shanghang Basin to produce magmas that ultimately formed the volcanic rocks.

Petrogenesis
The SiO 2 contents of melts directly produced by partial melting of mantle material are less than 57 wt % [84,85]. However, the SiO 2 contents of the studied Shimaoshan Group volcanic rocks range from 56 to 71 wt %, suggesting that these rocks do not represent the primary magma derived from partial melting of the mantle. Similarly, the Mg # values of primitive arc melts are generally >70 [86], compared with <67 (most <50) of the studied rocks, meaning that these volcanic rocks were formed from mantle-derived magma after a high degree of fractional crystallization in a deep magma chamber [87]. This interpretation is supported by the lower contents of some compatible elements in the volcanic rocks. For example, the Cr (1.2-48.3 ppm) and Ni (1.8-35.6 ppm) contents of the samples are considerably lower than those of primitive arc magma (Cr = 364 ppm, Ni = 168 ppm) [88] (Figure 7). Given that Cr tends to enter Cr-spinel and clinopyroxene, and Ni generally enters olivine, fractional crystallization of olivine, clinopyroxene, and Cr-spinel is inferred to have occurred during the early stage of magma evolution. Moreover, the volcanic rocks are not depleted in Al 2 O 3 content relative to primitive arc magma (Figure 7c), suggesting that plagioclase crystallization was negligible in the early stage of magma differentiation. In addition, there is no negative Eu anomaly apparent in an REE variation diagram (Figure 8a,c,e), also indicating that early fractional crystallization of plagioclase was negligible.
The SiO 2 content of the studied samples is negatively correlated with various elements, including Fe 2 O 3 T , MgO, Al 2 O 3 , and TiO 2 (Figure 7). These negative correlations can be interpreted in terms of fractionation of mafic minerals, such as magnetite, hornblende, and biotite, which tend to remove most of the Fe, Mg, and Ca from melts and cause them to evolve toward more silicic compositions [4]. Owing to different partition coefficients of the various elements, a correlation diagram of the element ratios can qualitatively determine separation of mineral phases. For example, fractionation of biotite can lead to an increase in SiO 2 /Al 2 O 3 and a decrease in Sc/Th (Figure 11a), and fractionation of hornblende can result in a positive correlation between Dy/Yb and Nb/Ta (Figure 11b). These patterns suggest that the magma underwent fractional crystallization of amphibole and biotite when it evolved from an andesitic to a dacitic composition. The observed decreases in V and Ti with SiO 2 may be related to the separation of Fe-Ti oxides, such as magnetite, as supported by microscope observations (Figure 3f). Moreover, fractional crystallization and partial melting trends can be distinguished in diagrams of Ni versus Rb and Ni versus Ba (Figure 11c,d). In summary, the parental magma underwent fractional crystallization of Cr-spinel and clinopyroxene in the deep magma chamber with suppression of plagioclase crystallization. During magma ascent, biotite and hornblende were the main mineral phases. After mineral separation, the magma was emplaced at shallow or surficial crustal levels, forming the Shimaoshan Group volcanic rocks in the Zijinshan district.

Implications for Porphyry-Epithermal Mineralization
The water content and oxygen fugacity of the magma are the most important controls on the metallogenic potential of porphyry magmas [89][90][91]. In a comagmatic system, important physicalchemical information of the parental magmas can be obtained from volcanic rocks [90,92]. In turn, the f O 2 -H 2 O conditions in the parental magma of volcanic rocks provide insights into the metallogenic potential of the porphyries in the Zijinshan ore district.
It is difficult to evaluate the original magmatic oxidation state in altered plutonic rocks, but normalized zircon Ce and Eu anomalies of volcanic rocks can be used to provide qualitative estimations [89,93,94]. The element Eu has two valence states: Eu 3+ and Eu 2+ . Eu 3+ is a more compatible ion than Eu 2+ and can therefore more readily replace zirconium ions in the zircon crystal lattice, but it requires more oxidizing conditions. Thus, the ratio Eu N /Eu N * can be used as a suggestion of magma oxygen fugacity [89,95]. To avoid the influence of mineral inclusions (such as apatite and titanite), we used Ca > 200 ppm or La > 0.3 ppm as an indicator of apatite contamination and Ti > 20 ppm to reflect titanite contamination [96]. At the same time, plagioclase crystallization would also influence on Eu N /Eu N *. As a result, samples with similar SiO 2 content (63-66 wt %) were selected for comparison, which represented that they have similar degrees of differentiation. Zircon Eu N /Eu N * values of the volcanic rocks are similar to those the Luoboling porphyry deposit and to some other porphyry deposits in the world [89,96,97] (Figure 12a), suggesting that the parental magma was of high oxygen fugacity. In a diagram of Fe 2 O 3 T versus MgO (Figure 12b), Fe 2 O 3 T content is positively correlated with MgO content for the studied rocks, with the change in slope marking magnetite saturation, indicating that magnetite had crystallized and separated from the magma. The onset of magnetite saturation from basaltic magma is controlled largely by oxygen fugacity [98]. As shown in Figure 12b, the onset of magnetite saturation in the parental magma of the volcanic rocks was like that in other typical arc magmas worldwide. This early saturation of magnetite suggests that the oxygen fugacity of magma that formed the studied Shimaoshan Group volcanic rocks was relatively high. Previous studies have shown that water-rich magma is liable to dissolve volatiles during its evolution, which is conducive to enrichment and migration of ore-forming elements [103]. A high-water content of the magma inhibits early crystallization of plagioclase and promotes crystallization of amphibole, which can generate some trace element anomalies. For example, the magma has a higher Sr content when crystallization of plagioclase is inhibited, because there is very little Sr to replace Ca into plagioclase lattice. At the same time, medium REEs and Y can readily enter amphibole in intermediate to acidic magma, and the crystallization of amphibole in water-rich magma reduces its Y content. Therefore, water-rich magma commonly shows a high whole-rock Sr/Y ratio [104,105], consistent with our results (Sr/Y = 4.4-54.6, mostly >20), which suggest separation of amphibole. REE patterns also suggest that these elements entered amphibole during the crystallization process (Figure 8a,c,e) [96]. In addition, our samples have a whole-rock mean [La/Sm] N value of 4.53, much higher than that of Zijinshan barren rocks (3.8, [35]), and have low mean Dy/Yb values of 0.2 in the zircon and high whole-rock V/Sc ratios (mean = 9.12), indicating a high water content of the magma [91,94]. Furthermore, Eu 2+ can easily replace Ca 2+ in plagioclase, meaning that the Eu anomaly can be used to indicate the crystallization of plagioclase: a strong negative anomaly indicates early crystallization of plagioclase; and a weak negative anomaly suggests either that the magma had a high water content that inhibited crystallization of plagioclase or that the magma had a high oxygen fugacity with Eu existing mainly as Eu 3+ . Chondrite-normalized REE patterns (Figure 8) show either a weak negative or negligible Eu anomaly (Eu N /Eu N * = 0.89-1.06), suggesting that fractional crystallization of plagioclase can be ignored.
High oxygen fugacity can facilitate porphyry mineralization in multiple stages of magmatic evolution [106]. In addition, high water contents of ore-forming magma are a key control on the formation of porphyry deposits [107,108]. Copper, gold, and molybdenum are all chalcophile elements with high partition coefficients between sulfides and melts, with the behaviors of Cu and Au being strongly controlled by sulfides [109]. Oxygen fugacity controls sulfur speciation, which in turn is the dominant influence on S solubility of natural silicate melts [110]. The solubility of sulfate is one order of magnitude higher than that of sulfide [110]. Under conditions of high oxygen fugacity, most of the sulfur is removed in the form of sulfate during partial melting of a mantle source [110], leading to the release of abundant chalcophile-forming metals from the magma [111]. In contrast, sulfide saturation in the magma is suppressed by the high solubility of sulfur, resulting in chalcophile-forming metals (e.g., Cu and Au) being retained in the magma [112]. Moreover, high oxygen fugacity favors the presence of Mo as MoO 4 2− , which promotes enrichment of Mo in residual melt [113]. Water-rich magma is more likely to reach water saturation, which is conducive to the extraction and dissolution of volatiles. These volatiles are readily combined with gold, copper, and other ore-forming elements to form soluble complexes, which facilitates the enrichment and migration of ore-forming elements. Combining all of the above, we conclude that the high water content and high oxygen fugacity of the source magma promoted mineralization of the Zijinshan epithermal-porphyry deposit system.

Conclusions
(1) Zircon U-Pb isotope dating yielded an age of 105 Ma for the Shimaoshan Group volcanic rocks, which is identical to the mineralization age of the nearby porphyry-epithermal ore deposits in the Zijinshan district. This, together with the similar geochemical features, suggests that the volcanic rocks and the mineralization were derived from magmas residing in a common magma chamber beneath the volcanic basin and that these magmas underwent similar fractionation pathways.
(2) Major-and trace-element geochemistry reveals that the volcanic rocks from the Shanghang Basin are predominantly high-K and shoshonitic rocks. Sr-Nd-Hf and O isotopes of the rocks indicate that they were derived predominantly from partial melting of the subduction-modified mantle in a back-arc extension setting as a response to rollback of the subducting paleo-Pacific Plate. Extensive fractional crystallization and crustal contamination account for the compositional variation in the volcanic rocks.
(3) High Eu N /Eu N * values of the studied volcanic rocks and the inferred early crystallization of magnetite imply a high oxygen fugacity of the parental magma. Inferred hornblende crystallization and inhibition of plagioclase crystallization indicate a high-water content of the magma. High oxygen fugacity and high-water content promoted the dissolution of Cu-and Au-bearing sulfides in the source area and the concentrating of Cu and Au in the magma to form metal-and water-rich magma, followed by extraction of these elements from the magma by volatiles. We conclude that the high oxygen fugacity and high water content were important controlling factors in the formation of the Zijinshan ore deposit.

Conflicts of Interest:
The authors declare no conflict of interest.