In this study, we present new mineral chemical data on the garnet from the Hongshan Cu-Mo deposit, one of the largest skarn deposits in the Sanjiang Tethyan tectonic domain, in SW China. The aims are to unravel the formation mechanism of the oscillatory-zoned garnet at Hongshan, and to explore possible applications using garnet trace element to distinguish different types of skarn deposits.

Two generations of garnets, i.e., poorly-zoned garnet I (Grt I) and the well-zoned garnet II (Grt II), are identified from the endoskarn and exoskarn, at Hongshan based on field and petrographic observations. EPMA and LA-ICP-MS analyses show that Grt I contains a narrow compositional range (Adr_40.07-61.25 Grs_36.65-57.08), and chondrite-normalized HREE-enriched and LREE-depleted patterns with minor positive Eu anomalies. In contrast, Grt II shows a wide compositional range (Adr_27.78-67.53 Grs_30.67-70.92 to almost pure andradite Adr_94.41-99.99 Grs_0.00-4.21) with interlayered Al-rich and Fe-rich zones. In Grt II, its Al-rich zone is chemically similar to Grt I, whereas its Fe-rich zone exhibits chondrite-normalized LREE-enriched and HREE-depleted patterns with distinct positive Eu anomalies. The positive REE³⁺ vs. Mg correlations, and the poor REE³⁺ vs. Y and Ca correlations, altogether indicate that the REE³⁺ incorporated into garnet was co-influenced by the menzerite-type substitution mechanism (Ca2+-1VIIIREE3++1VIIIAl3+-1VIMg2++1VI), fluid chemistry, and the physicochemical conditions of precipitation. The higher U contents and HREE-enriched patterns of Grt I suggest that it was formed under a relatively reducing and near-neutral pH condition. Varying U contents and REE patterns across different oscillatory zones indicate that Grt II was probably formed under alternating physicochemical fluctuations led by repeating crack-seal cycles in the hydrothermal system. Besides, chemical data compilation suggests that the garnet from different types of skarn deposits has different compositions, and can be used as mineral prospecting tools. We suggested that discrimination plots, such as δEu vs. V and δCe vs. U, can distinguish Cu skarn deposit from other skarn deposit types, whilst the U vs. W plot can discriminate between Cu, W, W-Sn and W-Mo skarn deposits.