The Zircon U-Pb dating results are listed in Supplementary Table S1. The zircons are mostly euhedral and reveal long to short prismatic forms, with average crystal lengths of 150–300 μm and length-to-width ratios from 2:1 to 3:1. Most zircons were transparent and colorless or pale brown. In the CL images, the zircon crystals are internally homogeneous with weak, broad zoning and without complex internal structures, as is typical of zircons formed in gabbroic magmas (Corfu et al., 2003). All zircon rare Earth element (REE) partition curves show depletion of light rare Earth elements (LREE) and enrichment of heavy rare Earth elements (HREE), with positive Ce anomalies and negative Pr and Eu anomalies. In addition, no overgrowths, mineral or fluid inclusions, or metamictization were observed in the analyzed zircons, suggesting that the zircons were not affected by post-magmatic processes. Thus, the interpretation of the zircon U-Pb isotopic results is simple (Figure 4), and the obtained ages represent the formation time of the gabbro. The Th/U value of the analyzed spots varies from 0.31 to 2.72, suggesting the zircons are of magmatic origin (Williams, 2001; Rubatto, 2002). U-Pb isotopic analyses yielded disparate zircon ²⁰⁶Pb/²³⁸U ages of 495.3 ± 3.6 Ma (mean square weighted deviation (MSWD) = 2.6) and 519.2 ± 2.9 Ma (MSWD = 0.24), respectively.

The Lu-Hf isotopes were analyzed on selected zircon grains from two different samples that were previously systematically analyzed with U-Pb. Lu-Hf analysis was performed on 41 representative zircon grains dated using the LA-ICP-MS U-Pb method. The results are shown in Supplementary Table S2. The initial Hf composition of zircon represents the ¹⁷⁶Hf/¹⁷⁷Hf value calculated at the time of zircon crystallization, namely, the U-Pb age, likely concordant with that previously obtained for the same crystal. The two-stage depleted mantle Hf model ages (T_DM Hf) were calculated using ¹⁷⁶Lu/¹⁷⁷Hf = 0.0384 and ¹⁷⁶Hf/¹⁷⁷Hf = 0.28325 for the depleted mantle (Chauvel and Blichert-Toft, 2001). The resulting ¹⁷⁶Lu/¹⁷⁷Hf values ranged from 0.282,809 to 0.282,973 with a mean of 0.282,877, indicating that the zircons were weak in radiogenic Hf. The initial zircon ¹⁷⁶Hf/¹⁷⁷Hf value varies with age. Zircons from sample 21-YSG335 had a high ε_Hf(t) between +6.54 and +17.34, with an average of 14.28. The T_DM values were restricted to the narrow range from 0.41 to 0.86 Ga. Zircons from sample 21-YSG341 were characterized by positive ε_Hf(t), ranging between +12.13 and +15.8, and T_DM values ranging between 0.44 and 0.67 Ga. Their ε_Hf(t) values were close to that of the depleted mantle evolution curve, suggesting that these zircons crystallized from magma with a juvenile signature.

Supplementary Table S3 shows that the SiO₂ contents of the 12 samples range from 45.40 wt% to 57.20 wt%, and the rocks are characterized by low TiO₂ (0.24–1.78 wt%), K₂O (0.01–0.57 wt%) and high Na₂O (0.45–5.15 wt%) contents. The high loss of ignition of rocks (1.62–6.03 wt%) indicates that the samples are slightly altered. The K₂O and Na₂O contents of the rocks may be related to alteration by K- and Na-rich fluids (such as seawater) after formation. The gabbros are enriched in MgO (4.09–25.22 wt%), Mg^# (46.9–82.0), and CaO (3.95–12.92 wt%). This reflects the combined plagioclase and clinopyroxene compositions of the initially formed basic rocks.

The samples collected in this study were greyish-black with medium-to coarse-grained structures. They were mainly composed of clinopyroxene and plagioclase in nearly equal amounts. The secondary mineral is amphibole, which contains small amounts of quartz. Under a single polarized electron microscope, the entire thin section of the sample was dark green, showing a gabbro structure, and the contents of clinopyroxene and feldspar were almost equal. Field and electron microscope observations indicated that the samples collected were gabbro. The results shown in the total alkali-silica diagram are consistent with field observations and electron microscopy (Figure 5A). As the samples were likely altered by K- and Na-rich fluids, the increase in the total alkali content and loss of ignition caused a shift in the rock composition. To further determine the rock type, immobile high-field-strength elements were used for discrimination (Figure 5B), showing that all samples fell into the basalt field. Regarding the relationship between FeO^T, MgO, and Na₂O+K₂O (Figure 5C), most of the rocks were of the tholeiitic series, and the rest were of the calc-alkaline series. In the Al₂O₃-CaO-MgO diagram (Figure 5D), almost all samples were distributed within the mafic cumulative rock fields, and one of the samples fell within the ultramafic cumulative rock fields.

Yushigou gabbros have low ΣREE content ranging from 8.36 to 75.60 ppm, with low to slight enrichment in LREE in the chondrite-normalized REE distribution patterns (Figure 6A). The (La/Yb)_N values range from 0.80 to 1.66, while the (La/Sm)_N values vary from 0.54 to 1.61. The parallel REE distribution lines indicate that all samples were derived from the same magma source. In addition, the gabbro samples displayed mid-ocean ridge basalt (MORB)-like trace-element characteristics. Notably, two samples exhibited Nb and Ta depletion (Figure 6B).

The major and trace element data for clinopyroxenes in the Yushigou gabbro are presented in Supplementary Tables S4, S5. The analyzed clinopyroxene grains from the Yushigou gabbro were augmented (Figure 7A). Clinopyroxene analysis of the Yushigou gabbro showed relatively high MgO (14.28–16.09 wt%) contents, with Mg# (100 × Mg/[Mg + Fe²⁺]) values ranging from 61 to 65. The grains are characterized by relatively low Al₂O₃ (3.46–8.61 wt%) and Na₂O (0.40–1.18 wt%) contents and high Cr₂O₃ (0.27–2.37 wt%) content. These results suggest that the clinopyroxenes of the Yushigou gabbro were subalkaline and crystallized under medium- and low-pressure conditions (Figures 7B, C). The clinopyroxenes exhibit low total REE contents (9.49–38.36 ppm). The REEs in the Yushigou clinopyroxene samples had characteristics of normal (N-)MORB with a relatively flat trend (Figure 7D). Compared to HREEs, LREEs were slightly depleted.

For a basic–ultrabasic rock series resulting from separation crystallization, the projection points on the La/Sm vs. La elemental covariant map form a horizontal line (Allegre and Minster, 1978; Yang and Gu, 1990). In the La/Sm-La diagram (Figure 8A), the horizontal and linear relationship indicates that separation crystallization is the main factor controlling magma evolution rather than partial melting. The general trends of CaO and MgO in the Yushiguo gabbros indicated significant fractionation of olivine and clinopyroxene (Figure 8B). Moreover, the increase in Sr content with increasing Al₂O₃ content (Figure 8C) suggests the removal of plagioclase. However, the Yushigou gabbros do not show a significant negative Eu anomaly, indicating that plagioclase may not have been significantly fractionated during magma evolution.

Temperature is a key variable controlling magmatic phase equilibria (Neave and Putirka, 2017). In addition to composition, crystallinity, and oxygen fugacity, pressure is another primary variable that affects magmatic phase equilibria (Yoder and Tilley, 1962). Understanding the distribution of the magma storage depth within the lithosphere provides information on both oceanic and continental crustal formation mechanisms (Henstock et al., 1993; Kelemen et al., 1997; Annen et al., 2006). Therefore, it is essential to determine magma storage pressures and depths.

In this study, the clinopyroxene thermometer proposed by Putirka (2008) was used to calculate the temperature and pressure of the gabbro in the Yushigou ophiolite. The formation temperature of clinopyroxene is 1,221.3°C–1,376.6°C, and the pressure at this temperature is 4.5–15.0 kbar. The temperature range of the Yushigou ophiolite was 1,174°C–1,402°C, and the mineral phase equilibrium temperature was approximately 1,230°C, which is consistent with the temperature measured in this study. The consolidation equilibrium temperature of mantle magmatic rocks in the Yushigou ophiolite is speculated to be approximately 1,200°C (Yan, 2014). In this study, temperature, pressure, and depth models (Lee et al., 2009) were used to estimate the mantle magmatic depth of the Yushigou ophiolite. The simulation estimated the depth of the clinopyroxene crystals, indicating that the mantle magmatic rocks in the Yushigou ophiolite began to consolidate at approximately 60 km (Figure 8D).

The Yushigou gabbro is a magmatic rock of the tholeiitic series, featuring mild large-ion lithophile elements and high-field strength elements, which are typical geochemical characteristics of MORB. The geochemical characteristics of the trace elements show that the gabbro formed in the depleted mantle. The rare Earth distribution curve and trace element multi-element diagram of the Yushigou gabbro are nearly parallel to those of N-MORB (Figure 6), and the Th/Ta, Th/Yb, and Zr/Nb values of trace elements in the Yushigou gabbros are similar to those of the depleted mantle (Th/Ta=2.2, Th/Yb=0.25, Zr/Nb=18, from Condie (1989)). As shown in the Nb vs. Zr diagrams (Figure 9A), the Yushigou gabbros fall near the primitive mantle line, which is close to the depletion type. In addition, the Hf isotopes of the zircons reflect the characteristics of the source area. No inherited or captured zircon was found in the gabbro-zircon CL image or in situ micro-survey of the Yushigou ophiolite, indicating that the zircon crystallized in a homogeneous, unmixed magmatic source. The relatively high ε_Hf(t) values of the Yushigou gabbros indicate that they primarily originated from depleted mantle (Figure 9B).

Ba is a fluid-active element (Kessel et al., 2005), which easily enters the mantle via fluid migration during subduction (Morris and Ryan, 2003). Because of the significant difference in the Ba content between the crust and mantle, Ba can be used to trace the recirculation-related processes of subducting materials (Elliott et al., 1997; Murphy et al., 2002; Pearce and Stern, 2006; Kuritani et al., 2011). A high Ba/Th value indicates aqueous fluid from the dehydrated ocean crust or sediment, and a high Th/Nb value indicates the addition of partially melted material from subducted sediments. The studied samples showed high Ba/Th and Th/Nb values, indicating that they were affected by two subduction components (Figure 9C), and the additional subduction components comprised 1%–3% of the total composition (Figure 9D).

Recent research has shown that, in contrast to N-MORB, garnet source basalts (G-MORB) exhibit a significant garnet signature (e.g., Montanini et al., 2008; Saccani et al., 2008; 2013). This garnet signature can be highlighted using LREE/HREE and middle rare Earth elements (MREE) to HREE ratios, such as Ce/Yb and Dy/Yb. In the chondrite-standardized (Ce/Yb)_N vs. (Dy/Yb)_N diagram, all sample sites are located on the N-MORB side (Figure 10A), which is significantly different from the G-MORB and depleted of HREE. This indicates that the Yushigou gabbro was derived from a mantle source of non-garnet peridotite (Saccani, 2015). The Ce/Yb values of gabbros ranged from 2.18 to 4.63, which indicates that the gabbros were sourced from the stable zone of spinel less than 70 km deep (Xiao et al., 2003). This is consistent with the formation depth of clinopyroxene in the simulation, where spinel lherzolite partial melts, mantle residues, and melts have similar Sm/Yb values, and Sm values decrease with an increase in the partial melting degree (Aldanmaz et al., 2000). Therefore, the Yushigou gabbro falls within the spinel lherzolite region, and the Sm values vary widely, whereas the Sm/Yb values are relatively constant (Figure 10B).

Active elements (Cs, Rb, Sr, and Na) migrate under the influence of alteration and are not effective indicators of tectonic environments; however, most high-field elements (Ta, Nb, Zr, Hf, Ti, Th, and REE) are unaffected (Mullen, 1983). In this study, high-field elements were used to identify the tectonic environment of the Yushigou ophiolite suite gabbro. Studies have demonstrated that both Th and Ta are closely related to subduction. Both are highly incompatible elements, and their elemental ratios remain relatively stable during mantle melting or crystallization differentiation. However, during subduction magmatism, the sediments are rich in Th and depleted in Ta. The addition of a small amount of sediment melt can lead to an increase in the Th content of the magma, and an increase in the Th/Yb value can reflect the contribution of sediment melt in the source region (Elliott et al., 1997; Class et al., 2000; Singer et al., 2007). In contrast, the Ta content is sensitive to subduction fluids. The covariant relationship between the Th/Yb and Ta/Yb values suggests that the components (including sediments and fluids) derived from subduction contributed to the generation of the Yushigou gabbros (Figure 10C).

Tectonic discrimination diagrams of the elemental distributions of Si, Al, and Ti in the clinopyroxene compositions provide a distinct classification of magma types (Beccaluva et al., 1989). In the clinopyroxene Si versus Al diagram (Figure 10D), the sample fell into and near the MORB field. However, in the clinopyroxene Ti versus Al^Ⅳ diagram (Figure 10E), the sample falls within the island arc tholeiite (IAT) field. Grimes et al. (2007) used U, Th, Hf, Y, and Yb (as a monitor for HREEs) to discriminate crystallized zircons from the MORB mantle from those formed in continental magmatic settings. During subduction, easily migrated incompatible elements (such as large ion lithophile elements, LREEs, and U) are separated from non-migrated high-field strength elements (such as Nb, Y, and HREEs). These easily migrated incompatible elements are mobilized by plate-derived fluids to increase the U/Yb value in the magma (Grimes et al., 2015). Zircons from the Yushigou gabbros were mostly within the 90% confidence interval of the MORB zircons, as presented in Figure 10F, although several overlapped with primitive magmatic arc zircons. This suggests that zircons from the Yushigou gabbro formed in a subduction-related environment.

Three explanations have been proposed for the tectonic evolution of the North Qilian orogenic belt. The first suggests that the ancient ocean was a part of the Proto-Tethys and that the tectonic evolution of the North Qilian orogenic belt can be regarded as a portion of the Tethyan tectonic domain (Wang and Liu, 1976). The second hypothesis proposes that the ancient ocean represented a limited extensional oceanic basin developed on the southern margin of the North China Craton. The continental margin rifted and expanded to form this basin, which then closed to form an orogenic belt (Feng and He, 1996; Xia et al., 1996; 1998). The final explanation emphasizes that the ancient ocean in the North Qilian orogenic belt was part of the Paleo-Asian Ocean (Zhang et al., 1997). After several years of discussion, most scholars agree with the second hypothesis (Feng and He, 1994; Qian et al., 2001; Du et al., 2006a; 2006b; 2007; Zhu and Du, 2007).

After the division of the Rodinia supercontinent, the Qilian Ocean opened and expanded as part of the Iapetus Ocean from approximately 710–520 Ma (Song et al., 2013b). Basic rocks with mid-ocean ridge or back-arc basin characteristics have been identified in the Yushigou area with ages ranging from 550 to 521 Ma (Xia et al., 1998; Shi et al., 2004; Hou et al., 2006). Initial ocean subduction and infant arc magmatism occurred from 520 to 490 Ma, which caused partial melting of the mantle wedge and formed infant arc basalts from 517 to 505 Ma (Song et al., 2013b). Wu et al. (2010) identified arc volcanic granites with an age of 512 Ma. Song et al. (2013a) reported that the magmatic zircon age of the Chaidanuo intrusion was 516 ± 4 Ma, recording the oldest arc magmatic activity in the North Qilian Orogenic Belt. A report on the Dachadaban boninite (505 Ma) also suggested a pre-arc environment related to the inner oceanic island arc in the North Qilian orogenic belt (Chen et al., 1995; Meng et al., 2010). This implies the formation of back-arc basins (Song et al., 2013b). The study from Hou et al. (2006) of isotope means indicated that the Yushigou ophiolite most likely formed in a mid-ocean ridge or mature back-arc basin. Combined with previous data and this study, the back-arc basin of the North Qilian orogenic belt may have evolved to a relatively mature stage from 519 to 495 Ma.