Skarn classification and element mobility in the Yeshan Iron Deposit, Eastern China: Insight from lithogeochemistry

Skarns form by significant fluid-mediated exchange of mass and heat between igneous rocks and their surrounding wall rocks into which they intruded. Quantification of the mass exchange associated with skarn alteration requires knowledge of the skarn protoliths, which are often masked by metamorphic recrystallization and intense calc-silicate metasomatism. To overcome this challenge in characterizing the Yeshan skarn Fe deposit in Eastern China, a cross-section through the complete rock sequence, from the carbonate wall rock to the pluton, was systematically sampled, and analyzed for bulk-rock major and trace elements. Underpinned by the skarn zonation model, ln(SiO 2 /Al 2 O 3 ), ln(SiO 2 /TiO 2 ), and REE + Y values in the skarns were used to distinguish the various skarn protoliths. The effectiveness of the ln(SiO 2 /Al 2 O 3 ) and ln(SiO 2 /TiO 2 ) is supported by the variable mobility of Si, Al, and Ti during magma-derived fluid infiltration into the carbonate wall rocks at Yeshan. The effectiveness of REE + Y is based on their significant concentration differences in the carbonate wall rocks and igneous rocks at Yeshan. These geochemical indexes may be applicable to the characterization of protoliths and mass transfer in skarn deposits where igneous rocks intruded carbonate wall rocks.


Introduction
Skarns are the product of metasomatic processes that result in a substantial amount of heat and mass exchange between igneous rocks and wall rocks in the presence of hydrothermal fluids.Based on the characteristics of the protolith, skarns can be divided into endoskarns and exoskarns, which correspond to skarns developed in the igneous rocks and the wall rocks, respectively (Einaudi et al., 1981;Meinert et al., 2005).
With respect to skarn deposit exploitation, determination of the protoliths is essential to mineral resource estimation.It is also of economic significance for metallurgy because grinding efficiency, liberation characteristics, and metal recovery are affected by the mineral assemblages and particle size distribution, both of which are closely linked to primary lithological characteristics (e.g., Maydagán et al., 2016).Identification of the protolith is quite challenging, however, as intense thermal metamorphism and calc-silicate metasomatism can completely mask the primary textures of the protolith (e.g., Mrozek et al., 2020;Ordóñez-Calderón et al., 2017).Mrozek et al, (2020) demonstrated that immobile elements, including Al 2 O 3 , heavy rare elements, and some high field strength elements, could be used to classify exoskarns and endoskarns at the Antanima Cu-Zn skarn system.However, these elements have been shown to be mobile at other skarn deposits (Gieré, 1990;Lentz, 2005, and references therein), as well as the Yeshan skarn Fe deposit in this study.Characterization of the protolith in the Yeshan skarn Fe deposit, therefore, poses unique challenges and requires a geological-geochemical toolset that is different from that applied to the Antanima Cu-Zn skarn system.
Our study reveals that underpinned by the skarn zonation model, SiO 2 /Al 2 O 3 , SiO 2 /TiO 2 and REE + Y contents could be used to distinguish the various skarn protoliths at the Yeshan skarn Fe deposit.Additionally, the gains and losses of elements associated with the skarn development at Yeshan are quantified.
The Ningzhen ore district, situated in the easternmost part of the MLYB (Fig. 1), contains numerous deposits enriched in Pb, Zn, Ag, Cu, Fe, and Mo (Fig. 2).The Ningzhen ore district comprises mainly three tectonostratigraphic units -Mesoproterozoic metamorphic basement, Sinian-Triassic sedimentary cover, and Jurassic-Neogene superimposed cover.The basement rock is the unexposed Picheng Group, which comprises schists, amphibolites, and granulites, with a total thickness of more than 480 m (Jiangsu Bureau Geological Mineral Resource, 1989).
The overlying Sinian-Triassic sedimentary rocks comprise marine carbonate, and terrestrial shale and siltstone.The Sinian-Triassic carbonate rocks are closely related to most of the mineral deposits in the ore district, including the Qixiashan Pb-Zn-Ag, Anjishan Cu, Tongshan Cu, Weigang Fe, Jianbi Mo, and Yeshan Fe deposits (Fig. 2).The Sinian--Triassic sedimentary rocks were strongly folded in the middle Triassic by the Indosinian Orogeny.The axial lines of these folds strike northeast in the western segment of the Ningzhen ore district, but change to a more dominant east-west strike in the central portion of the district (Fig. 2).
Prior to the Jurassic, the Ningzhen ore district became an active continental margin (e.g., Zhou and Li, 2000;Li and Li, 2007).
The Early Jurassic sediments are characterized by a terrestrial succession of conglomerate, sandstone, and mudstone, and are mainly distributed in the eastern portion of Ningzhen ore district (Fig. 2).From the Late Jurassic to the Cretaceous, it has generally been accepted that this continental margin was influenced by the westward subduction of the Pacific plate and the Izanagi plate (e.g., Ling et al., 2009), with a series of NNE-, NNW-, NE-and EW-trending faults having formed and superimposed on the folds in the ore district (Fig. 2).The plate subduction event was also a trigger for the intrusion of large volumes of granitoids and the eruption of volcanic rocks.There are seven 110-100 Ma quartz diorite-granodiorite-granite intrusions exposed in the middle segment of the ore district (Zeng et al., 2013;Chen et al., 2017;Zhang et al., 2018); from west to east these are the Bancang, Qilinmen, Anjishan, Xiashu-Gaozi, Xinqiao, Shima, and Jianbi intrusions (Fig. 2) (Sun et al., 2013).The Yeshan intrusion, which is the focus of this study, Fig. 1.Geologic map showing the distribution of porphyry-skarn-stratabound Cu-Au-Mo, skarn Fe-Cu, skarn Fe, and magnetite-apatite deposits along with the Middle-Lower Yangtze River metallogenic belt (modified after Pan and Dong, 1999;Mao et al., 2011a;Mao et al., 2011b;Li et al., 2019;Hu et al., 2020;Xiao et al., 2021).Abbreviations: XGF = Xiangfan-Guangji Fault; TLF = Tancheng-Lujiang Fault; YCF = Yangxin-Changzhou Fault; CHF = Chuhe Fault.

S. Zhao et al.
is located in the northernmost portion of the Ningzhen ore district and has a zircon U-Pb age of 128 Ma (Chen et al., 2018), suggesting that it crystallized from a different magmatic pulse than those from which the other intrusions in the Ningzhen ore district formed (Fig. 2).

Deposit geology
The Yeshan skarn Fe deposit is located about 20 km northeast of Liuhe City, Jiangsu Province (Fig. 2).The ore field is located proximal to the Jiangpu-Liuhe fault (JLF) to the southeast and the Tianchang-Yizheng fault (TYF) to the southwest (Fig. 3).The Yeshan deposit was initially produced as an open-pit mine, but began underground production in 1971.With an iron ore reserve of 23.7 million tons (Mt) as of 2015 (Yeshan Mining Co. Ltd., unpublished report), the deposit represents an important source of iron in the Ningzhen ore district.Although the Yeshan deposit is, by definition, a classic example of Fe skarn mineralization, no systematic geological or geochemical studies of the deposit have been published.Literature pertinent to the deposit are in Chinese and mostly pertain to the genesis of the pluton (Zi et al., 2011;Chen et al., 2018).Discussions on the mineralization mechanisms are elusive (Sun et al., 2017).
The deposit comprises three ore zonesthe North, Eastern, and Tieshigang ore zones (Fig. 3) -and contains nine major tabular-and lens-shaped economic ore bodies that occur within the pluton (in the case of the Tieshigang ore zone) or at the contact zone between the pluton and the wall rock (in the case of the North and Eastern ore zones).The individual ore bodies have thicknesses of 0.55-54.7 m and lengths of 10-320 m.In the North ore zone, the ore bodies strike northwest, dip at 25-65 • to the southwest, and contain average Fe and Cu grades of approximately 44.24 wt% and 0.40 wt%, respectively.In the Eastern ore zone, Fe ore bodies strike east-west and dip at 40-75 • to the south, with an average Fe grade of 37.58 wt%.The Eastern ore zone also contains B mineralization hosted by camsellite and ludwigite, with a reserve of 280,000 tons at an average ore grade of 11.32 wt%.No Cu mineralization exists in the Eastern zone.In the Tieshigang ore zone, Fe ore bodies strike northeast and dip at 75-85 • to the southeast, and contain average Fe and Cu grades of approximately 42.09 wt% and 0.15 wt%, respectively.The ore-bearing pluton at Yeshan can be subdivided into three facies based on petrographic and geochemical characteristics (Fig. 3) -granodiorite in the eastern portion, diorite in the western and southeastern portions, and quartz monzonite in the central portion.Zircon U-Pb dating yields ages of 128 ± 2 Ma and 129 ± 2 Ma for the granodiorite and southeastern diorite, respectively (Chen et al., 2018).
The wall rocks of the Yeshan deposit are controlled by NNW-and NEtrending faults (Fig. 3).The ore-related wall rocks comprise Sinian and lower Cambrian strata that are separated by four nearly parallel thrust faults (Fig. 3).The Sinian Huangxu Formation is mainly composed of limestone and dolomitic limestone.It is uncoformably overlain by the Dengying Formation, which consists of a dark grey dolostone intercalated with siliceous and muddy layers.The lower Cambrian strata, including the Hetang and Mufushan formations, unconformably overly the Sinian Dengying Formation.The Hetang Formation is characterized by black carbonaceous and/or siliceous shale, whereas the Mufushan Formation consists predominantly of dolomite intercalated with silicified hornfels.
Other wall rocks are unrelated to skarn mineralization in the Yeshan deposit, and include Jurassic and Cretaceous strata (Fig. 3; Chen et al., 2018).The Jurassic Xihengshan Formation is composed of quartz sandstone intercalated with thin layers of siltstone.The Jurassic Longwangshan Formation is a suite of pyroclastic rocks containing andesite intercalated with mudstone and tuff.The Cretaceous Pukou Formation consists of packsand, siltstone, and silty mudstone.The Cretaceous Chishan Formation comprises quartz sandstone, siltstone, and mudstone (Chen et al., 2018).

Samples and analytical techniques
Samples selected for this study were taken from drill core ZK302 at the Xiaomiaochen area of the Yeshan deposit (Fig. 4a), which is to the northern of the North ore zone (Fig. 3).The drill core extends to a depth of ~900 m and intersects a complete sequence of rocks that make up the deposit, from the carbonate wall rock, skarn, to the igneous pluton.More than 60 samples were collected along the length of the drill core at intervals of approximately 20 m (Fig. 4b).

Whole-rock geochemistry
The major-element composition of the skarns (n = 21) and quartz monzonite (n = 15) was determined using a Thermo Scientific ARL-9900 X-ray fluorescence spectrometer (XRF) at the State Key Laboratory for Mineral Deposits Research, Nanjing University.Prior to XRF analyses, rock powders were heated for 4 h at 105 • C to remove pore and surface water.The whole-rock powders (1.0 g) were then mixed with Li 2 B 4 O 7 + LiBO 2 + LiBr (11 g) and fused into glass disks.Loss on ignition (LOI) was determined by changes in the weight of the samples after being heated at 1050 • C for 3 h.Reference materials GBW-07103, GBW-07135, BHVO-2, and BCR-2 were used as external standards to monitor analytical accuracy.Analytical precision for major elements is better than 2%.The major-element composition of marbles (n = 14) was measured by a PANalytical Magix Fast XRF using the ME-XRF26 package at ALS Chemex Co., Ltd (Guangzhou, China).Analytical uncertainty is better than 5%.The trace-element composition of the skarns (n = 21), quartz monzonite (n = 15), and marbles (n = 14) was determined by a Perkin Elmer Elan-9000 inductively coupled plasma mass spectrometry (ICP-MS) using the ME-MS81 package at ALS Chemex (Guangzhou).A prepared whole-rock power was added to lithium borate flux (LiBO 2 / Li 2 B 4 O 7 ), mixed well, and fused in a furnace at 1025 • C.After cooling, the melt was dissolved in the HCl-HNO 3 -HF mixture for analysis.Analytical precision and accuracy monitored by USGS rock standards (BCR-2, BHVO-1 and AGV-1) is better than 10%.

Isocon analysis
Mass transfer caused by hydrothermal alteration can be quantified by comparing the bulk chemical composition of original and altered rocks.The graphical isocon method of Grant (1986) is used here to investigate element mobility during skarn formation.The basic equation for the isocon method is.
where C A i and C O i represent the concentration of element i in altered and original rocks, respectively, and M A and M O denote the total mass of the altered and original rocks, respectively.The slope of the isocon line is defined by M O /M A .The gain or loss of element i is given by.Given that the magnitude of numeric values varies among elements, the geochemical data of original and altered rocks are artificially scaled for portrayal in the isocon diagram.Elements are divided into two groups by the isocon line.Elements plotting above the isocon line were gained during the hydrothermal alteration, while those plotting below the isocon line were lost.
Magnetite, garnet, diopside, and epidote are the main constituents of the ore body at Yeshan (Fig. 5i), with magnetite being the main ore mineral and chalcopyrite being a minor component.Magnetite can also be associated with retrograde minerals, such as actinolite, in some of the ore-bearing skarns (Fig. 6e).Serpentinized marble also exists at a depth of ~650 m.Serpentinized marble here contains minor olivine crystals, which are surrounded by serpentine (Fig. 5j).

Geochemistry
The concentration of major elements, trace elements, and rare earth elements (REE) in samples from drill hole ZK302 are listed in Appendix  Table A. A comparison of the composition of marbles, skarns, and quartz monzonite is provided in Fig. 7.
The CaO and MgO concentrations, and LOI of the dolomite marble are 31.2wt%, 20.7 wt%, and 46.6 wt%, respectively (Appendix Table A), which are similar to near-pure dolostone.Compared to dolomite marble, brucite marble has slightly higher CaO and MgO, but lower LOI (Fig. 7d-f).The CaO, MgO and LOI contents in serpentinized marble are similar to or slightly lower than dolomite marble and brucite marble (Fig. 7d-f).Serpentinized marble has higher SiO 2 contents than dolomite marble and brucite marble (e.g., Fig. 7a), consistent with its greater abundance of SiO 2 -bearing minerals, such as serpentine, olivine, and phlogopite.
Quartz monzonite has a high-K calc-alkaline geochemical affinity based on the SiO 2 -K 2 O division of Peccerillo and Taylor (1976).Relative to the marbles, the quartz monzonite has higher SiO 2 , TiO 2 , Al 2 O 3 , and K 2 O, but significantly lower CaO, MgO, and LOI contents (Fig. 7a-f).The quartz monzonite and marbles contain similar MnO and P 2 O 5 concentrations, although few marble samples extend to higher concentrations (Fig. 7h, i).Major elements in the chloritized quartz monzonite exhibit a wider range of concentrations than unaltered quartz monzonite, with TiO 2 , Al 2 O 3 , K 2 O, Fe 2 O 3 , and MnO exhibiting the largest variations (Fig. 7a-i).Skarn-altered quartz monzonite has lower concentrations of SiO 2 and Fe 2 O 3 , and higher concentrations of TiO 2 , CaO, MgO, LOI, and MnO than unaltered quartz monzonite.
Skarns have SiO 2 contents intermediate to the marbles and quartz monzonite, except for one garnet skarn sample (ZK302-290), which has a SiO 2 content of 60.8 wt% due to a greater abundance of quartz in this sample.The concentration of TiO 2 in skarns is highly variable (Fig. 7a).Epidote skarn has the highest TiO 2 contents, whereas diopside skarn and phlogopite-serpentine skarn have the lowest TiO 2 contents.Garnet-diopside skarn is similar to quartz monzonite in terms of TiO 2 contents, whereas garnet skarn spans the complete range from marble to quartz monzonite (Fig. 7a).
In general, garnet and epidote skarns are characterized by higher Al 2 O 3 and lower MgO contents than garnet-diopside and diopside skarns (Fig. 7b, e), consistent with the relative abundance of garnet, epidote, and diopside in these skarn varieties.Garnet skarn has similar Al 2 O 3 , MgO, and LOI contents to quartz monzonite, but its CaO and K 2 O contents are more similar to the marbles (Fig. 7b-f).The Fe 2 O 3 , MnO, and P 2 O 5 contents of the skarns are greater than in the marbles and quartz monzonite (Fig. 7g-i).
Quartz monzonite is characterized by higher concentrations of the large ion lithophile elements (LILEs; e.g., Rb, Sr), high field strength elements (HFSEs; e.g., Zr, Nb), and Th than the marbles (Fig. 7j-n).Most of the skarns have similar Rb contents to the marbles (Fig. 7j), but are characterized by highly variable Sr contents (Fig. 7k).Notably, the Sr content in epidote skarn is almost 2000 ppm, which is approximately twice the concentration as in the quartz monzonite (Appendix Table A).Similar to Sr, the Zr, Nb, Th, and REE + Y contents of the different types of skarns are highly variable compared to the limited variability observed in the quartz monzonite and marbles (Fig. 7l-o).

Skarn classification
Skarns developed at Yeshan are metasomatically zoned, with a pyroxene-rich assemblage proximal to the carbonate wall rock and a garnet-rich assemblage distal to the carbonate wall rock (Fig. 5), similar to other skarns in the world (e.g., Chang et al., 2019;Einaudi et al., 1981;Zharikov, 1970;Leng et al., 2021).This zonation pattern has been explained by the varying migration capability among elements during magma-derived fluid unidirectional infiltrated into the wall rock and the distance of Si transferring is larger than Al (Korzhinskii, 1959(Korzhinskii, , 1970)).It is worth noting that skarn zonation occurring as intrusion-pyroxene skarn-garnet skarn-marble has been recently emphasized (e.g., Soloviev et al., 2021;Soloviev et al., 2022).This skarn zonation is opposite to that developed at Yeshan, and may result from counter-infiltration processes (Zharikov, 1970).
Given the absolute immobile elements are difficult to constrain, the transfer variation among elements represented by the element ratios may offer the potential to characterize the skarn developed into different protoliths at Yeshan.The rationale is shown below.Without any specific knowledge about the skarn evolutionary stages, the zonation pattern at Yeshan can be reconciled with a simple fluid flow process (Meinert et al., 2005).Aluminum, Si, and other elements, such as Fe and Ti, in fluids would be sequestered into skarn minerals (e.g., garnet and pyroxene) during skarn alteration of carbonate wall rock at Yeshan.Similar to Al, Ti also has a lower mobility than Si in fluids (Rollinson, 1993).Aluminum and Ti are more compatible in garnet crystal structure than pyroxene (Deer et al., 1978, Deer et al., 1982); the proportion of Al and Ti in fluids decreased to a greater extent than that of Si-during garnet formation.Based on material balance theory (Hofmann, 1972), it may be, therefore, reasonable to expect that with increasing distance from the intrusion-carbonate contact zone, the SiO 2 /Al 2 O 3 and SiO 2 /TiO 2 ratio in the magma-derived fluid while migrating into the carbonate wall rock would increase.
It is critical to compare the solubility of a component to its original abundance in the rock (i.e., relative solubility index, RSI); the higher the abundance of a particular element in a rock, the higher the fluid/rock ratios required to alter it (Leitch and Lentz, 1994).Due to the negligible SiO 2 , Al 2 O 3 , and TiO 2 contents of the carbonate wall rock at Yeshan (Appendix Table A), the variable SiO 2 /Al 2 O 3 and SiO 2 /TiO 2 signatures of the magma-derived fluids could be fingerprinted by skarns formed from the carbonate wall rock.
Experimental studies show that the partition coefficients (D) of Si, Al, and Ti between fluids and melts range from 0.0005 to 0.01 in the magmatic-hydrothermal systems (P < 0.3GPa, Keppler, 1996;Reed et al., 2000), and the D values of Si are significantly higher than Al and Ti.These results suggest that 1) the concentrations of Si, Al, and Ti in the magma-derived fluid are at least two orders of magnitude lower than these of magmatic rocks, assuming that the residual melt composition after fluid exsolution is approximately equal to that of intrusion at the Yeshan Fe deposit; and 2) the SiO 2 /Al 2 O 3 and SiO 2 /TiO 2 ratios in the magma-derived fluid are higher than these of intrusion.Considering no Si-Ti-Al-bearing strata, except the cogenetic intrusion, exists in the Xiaomiaochen area at Yeshan (Fig. 4a), the Si, Al, and Ti contents of magmatic fluids may not be successively enriched by extensive fluid/ rock interactions, although the recent results of experimental studies and elemental analysis of fluid inclusions support that argument that these elements, particularly Ti traditionally considered as fluidimmobile (Rollinson, 1993), could have high contents (100 -1000 ppm) in the magmatic fluids at high temperature and high salinity (Chen et al., 2019;Mysen, 2019;Rapp et al., 2010).Moreover, the concentrations of Si, Al, and Ti in the magma-derived fluids are roughly calculated to be 3000 ppm, 42 ppm, and 18 ppm, respectively, at the Yeshan Fe deposit, through multiplying these elemental contents of the intrusion (Appendix Table A) and the relative D values between fluids and melts.The D values are 0.01 and 0.0005 for Si and Al, respectively, at 800 • C and 2.0 kbar with fluids of 1.1 m Cl − (Reed et al., 2000).The D value for Ti is approximately 0.005 at 1040 • C and 3.0 kbar with fluids of 5.0 m Cl − (Keppler, 1996).It is worth noting that geochemical data are compositional data, which are traditionally defined as constrained data and carry relative rather than absolute information (Aitchison, 1986).Thus, elemental ratios are indeed the robust representation of the compositional properties of varying lithological units at the Yeshan Fe deposit.Given the low Si, Al, and Ti contents in the magmatic fluids discussed above, the SiO 2 /Al 2 O 3 and SiO 2 /TiO 2 ratios of the igneous rocks would not be significantly modified by its cogenetic crystallized magma-derived fluid during skarn formation at Yeshan.Taken together, the variability in SiO 2 /Al 2 O 3 and SiO 2 /TiO 2 ratio should be able to distinguish endoskarns from exoskarns in the Yeshan skarn system, with endoskarns being characterized by lower overall SiO 2 /TiO 2 and SiO 2 / Al 2 O 3 ratios than exoskarns.
Given the low migration capability of Al and Ti compared to Si in hydrothermal fluids on a large transport scale (>20 m), their concentrations in distal skarns may be very low, resulting in anomalously high SiO 2 /Al 2 O 3 and SiO 2 /TiO 2 ratios.For convenience, the natural logarithm of both ratios is used to gauge the mobility difference among Si, Al, and Ti during the infiltration of magma-derived fluid into the carbonate wall rocks.Fig. 8 schematically illustrates the theoretical basis for classifying endoskarn and exoskarn using ln(SiO 2 /Al 2 O 3 ) and ln (SiO 2 /TiO 2 ) at Yeshan.It should be noted that the variations in SiO 2 / Al 2 O 3 and SiO 2 /TiO 2 ratio may not be suitable for discriminating skarns formed proximal to the contact zone between the intrusion and the carbonate wall rock (i.e., garnet-pyroxene skarns); this topic will be discussed separately.
Using the framework of Fig. 8, two diopside skarn samples (ZK302-240 and ZK302-250) that are proximal to marbles have higher ln(SiO 2 / Al 2 O 3 ) and ln(SiO 2 /TiO 2 ) values than the garnet skarn samples and are classified as exoskarns (Fig. 9d, e).The ln(SiO 2 /Al 2 O 3 ) values of two garnet skarn samples (ZK302-270 and ZK302-290) are distinctly higher than those in the igneous rocks, but lower than those in the diopside skarns (Fig. 9e).These garnet skarn samples match in the ln(SiO 2 /Al 2 O 3 ) and ln(SiO 2 /TiO 2 ) values well the proximal garnet skarns developed at the wall rocks in Fig. 8 and are considered exoskarns.The remaining garnet skarn samples with ln(SiO 2 /Al 2 O 3 ) values similar to the igneous rocks are classified as endoskarns.However, one garnet skarn sample ZK302-450 has a ln(SiO 2 /TiO 2 ) value closer to these of proximal garnet skarns (Fig. 9d); the final judgment of this sample would require other geochemical indexes.
Four garnet-diopside skarn samples (ZK302-471, ZK302-471.5,ZK302-510, and ZK302-686) have higher ln(SiO 2 /Al 2 O 3 ) and comparable ln(SiO 2 /TiO 2 ) values to the igneous rocks (Fig. 9d, e).These samples characterized by low garnet/diopside abundance ratios commonly form in relatively reduced environments, such as the igneous-wall rock contact zone where the effects of the oxidized magma may be lessened or completely masked by very reduced carbonate wall rocks (Meinert et al., 2005).Accordingly, due to the relatively small migration distance, the difference in mobility among Si, Al, and Ti may not be manifested at the contact zone.Again, other geochemical indices are, therefore, required to classify the garnet-diopside skarn as either endoor exoskarn.
The REE + Y content of the igneous rocks is significantly higher than the marbles at Yeshan (Fig. 7o).Based on the partition coefficient of REE between fluids and melts (Reed et al., 2000;Wen et al., 2020), the REE + Y contents in the exsolving magma-derived fluids are crudely calculated to be about 4.5 ppm following the similar calculation procedure for Si, Ti, and Al in fluids.The REE + Y content of skarns formed at the contact zone (e.g., garnet-diopside skarn) is expected to be more similar to their protolith.Accordingly, garnet-diopside skarn samples ZK302-471.5 and ZK302-510, which are characterized by REE + Y contents similar to or greater than the igneous rocks, likely formed from the igneous rocks and are classified as endoskarns (Fig. 9f).The other two garnet-diopside skarn samples (ZK302-471 and ZK302-686) have REE + Y contents that are more similar to the marbles and are classified as exoskarns (Fig. 9f).The classification of these garnet-diopside skarn samples at Yeshan as exoskarns suggests that TiO 2 may be able to be transferred in significant quantities into the wall rocks along with SiO 2 and Al 2 O 3 on a small transport scale (<5 m); this is supported by the ilmenite mineralization observed in the garnet-diopside exoskarn along the diorite-marble contact (Altunbey and Sagiroglu, 2003).The garnet skarn sample ZK302-450 exhibits the REE + Y content similar to the marbles (Fig. 9f), and is more likely to form into the carbonate wall rock.
Two epidote skarn samples (ZK302-530 and ZK302-550) have ln (SiO 2 /Al 2 O 3 ), ln(SiO 2 /TiO 2 ) values, and REE + Y contents comparable to the igneous rocks, suggesting that their protoliths are likely the igneous rocks; they are, therefore, classified as endoskarns.The Fig. 8. Idealized skarn zonation pattern modified after Einaudi et al., (1981) and Zharikov, (1970) illustrating a general variation of whole-rock ln(SiO 2 /Al 2 O 3 ) and ln(SiO 2 /TiO 2 ) from intrusion, endoskarn, and exoskarn at Yeshan.The black arrows show the direction and distance in which Si and Al from the magma-derived fluid transfer into the wall rock.Noting that the horizontal widths of the skarn zones are artificially scaled.classification of all skarns in this study is shown in Appendix Table A. In summary, endoskarn is more abundant than exoskarn at Yeshan.

Element mobility during endoskarn formation
Based on the extent of alteration and mineral paragenesis, two representative endoskarn samples (ZK302-670 and ZK302-390) from the Yeshan Fe deposit were selected to constrain mass transfer during endoskarn formation.Samples ZK302-886 and ZK302-870 are unaltered igneous rocks and chosen as the endoskarn protolith.
Sample ZK302-670 characterized by the coexistence of diopside and igneous biotite (Fig. 5k, 6f-g) represents the earliest stage of alteration of the igneous rocks.Nearly half of the feldspar crystals were replaced by diopside, with no new Al-bearing minerals having formed (Fig. 5k); Al 2 O 3 was, therefore, likely lost to the fluids during this early stage of skarn alteration.The slope of the isocon line defined by P 2 O 5 , TiO 2 , Ba, Sc, and Cr is 1.66 (Fig. 10a), indicating a total mass loss of 40% (Appendix Table B).MgO, CaO, MnO, and volatiles (expressed as LOI) were added, whereas SiO 2 , Al 2 O 3 , Fe 2 O 3 , Na 2 O, and K 2 O were removed from the igneous rock during this stage of skarn formation.The added MgO, CaO, and MnO were taken up by newly formed diopside, whereas the lost SiO 2 , Al 2 O 3 , Na 2 O, and K 2 O may be related to the partial destruction of feldspar.These lost elements could have migrated into the wall rock and precipitated Al-Si-bearing minerals, such as garnet, diopside, and spinel (e.g., sample ZK302-686; Fig. 5l, 6 h).The increased sericitization, chloritization, and carbonation of the igneous rocks is likely a result of the addition of volatiles.
Trace elements were mobilized at this stage of alteration, with most having been removed from the igneous rock.Zirconium, Hf, Th, and U exhibit similar losses (Fig. 10a), which may be due to the dissolution of   zircon during the skarn alteration or the zircon heterogeneity in the igneous protolith.
Sample ZK302-390 is chosen as an example to illustrate element mobility during garnet endoskarn formation as it is characterized by a high abundance of garnet (~80% in volume) and relatively low abundance of retrograde minerals, particularly epidote (Fig. 5f), simplifying the interpretation of element mobility during the prograde skarn stage.
In contrast to sample ZK02-670, skarn minerals in garnet endoskarn are Al-bearing minerals (garnet).The Al 2 O 3 -constant isocon is chosen and has a slope value of 0.83 (Fig. 10b), which corresponds to a 20% increase in the total mass of the rock (Appendix Table B).In this scenario, the formation of garnet endoskarn resulted in a minor gain in MgO and a moderate gain in TiO 2 (Fig. 10b).The soluble of TiO 2 in hydrothermal fluids has been experimentally demonstrated (Rapp et al., 2010).
CaO, MnO, Fe 2 O 3 , P 2 O 5 , and volatiles show significantly gained during garnet endoskarn formation, and K 2 O, Na 2 O, and LILEs (e.g., Rb, Cs, Ba, and Sr) lost.Nb, Ta, and Th were added during this stage of skarn formation (Fig. 10b).Experimental studies have demonstrated that the chloride complexes are not important for mobilizing Nb and Ta in hydrothermal fluids (Kotova, 2015;Akinfiev et al., 2020).The stability of Th-chloride complexes at temperatures above 250℃ is poorly understood (Nisbet et al., 2018).Other ligands (e.g., fluorine, phosphate, hydroxide, carbonate) should be, therefore, considered in the future for the transportation of Nb, Ta, and Th at Yeshan.

Element mobility during exoskarn formation
Carbonate is sensitive to elemental mobility during exoskarn formation because of its relatively simple chemical composition.Sample ZK302-11 is chosen as the protolith for the exoskarns as it has the greatest abundance of dolomite and is least altered (i.e., the lowest proportion of brucite).Given the relatively simple mineralogy, samples ZK302-240 (~90% diopside with subordinate magnetite and apatite) and ZK302-270 (~90% garnet and ~10% calcite) are chosen to evaluate the geochemical changes that accompanied exoskarn formation.
In the isocon diagram for diopside exoskarn formation, the isocon line defined by CaO, V, Li, Cs, Sr, Y, and rare earth elements has a slope of 0.87, suggesting a mass gain of 14%.The formation of diopside exoskarn resulted in a significant increase in SiO 2 and Fe 2 O 3 , and a significant loss of volatiles; CaO and MgO contents were not significantly affected (Fig. 11a).The addition of SiO 2 and Fe 2 O 3 is consistent with the presence of diopside and magnetite.TiO 2 , Zr, and Th were gained during this stage of exoskarn formation (Appendix Table A; Fig. 11a), which is consistent with the presence of ilmenite and Th-Zr-bearing minerals (Fig. 6a).The water/rock ratio for diopside exoskarn formation is calculated to be approximately 540 based on a SiO 2 concentration in the hydrothermal fluid of 20 mmol per kilogram H 2 O (Li et al., 2020).SiO 2 , Al 2 O 3 , TiO 2 , and Fe 2 O 3 , which are required for garnet formation, were added during the formation of garnet exoskarn, whereas MgO and volatiles were significantly lost.Rare earth elements, Y, Nb, U, and V compatible in the crystal structure of garnet, were, likely sequestered into garnet.Lithium, however, is highly incompatible into garnet (Brenan et al., 1998).Considering the low Li content of sample ZK302-270, it is inferred that the Li content of the wall rocks was not significantly modified by alteration.Lithium is, therefore, used as a reference framework to constrain mass transfer during garnet exoskarn formation, which corresponds to a total mass loss of 47% (Fig. 11b; Appendix Table C).Given the negligible concentration of Nb, Ta, Th, and Ga in the wall rocks (Appendix Table A), the addition of these elements in the garnet exoskarn provides robust evidence for their mobility in fluids.
Based on the solubility of microcline-muscovite-quartz and albite-paragonite-quartz (Woodland and Walther, 1987;Walther and Woodland, 1993), the calculated water/rock ratio for garnet exoskarn formation is approximately 560.This value is comparable with the estimated water/rock ratio for grandite exoskarn formation at the Mochito Zn deposit (Williams-Jones et al., 2010).

Conclusions
The Yeshan Fe deposit is a typical skarn Fe deposit in Eastern China.Alteration and mineralization at Yeshan are associated with the emplacement of Early Cretaceous quartz monzonite into Sinian dolostone.No textures of protoliths preserve in skarns.To develop a lithogeochemical technique for discriminating between endo-and exoskarns, and quantifying element mobility at Yeshan, a drill hole through the Yeshan Fe deposit was targeted that intersects the complete stratigraphic sequence of the skarn system.Underpinned by the skarn zonation model, a criterion that the SiO 2 /Al 2 O 3 and SiO 2 /TiO 2 ratios in exoskarn are higher than endoskarn has been proposed and used to characterize the protolith of varying skarns formed far from the contact zone at Yeshan.For skarns formed near the contact zone (e.g., garnetdiopside skarn), their protoliths could be identified using whole-rock REE + Y contents, given the significant difference in REE + Y content between the igneous and the carbonate wall rocks.Classification results show that endoskarns are more abundant than exoskarns at Yeshan.Isocon analysis suggests that almost all elements could be mobile at the Yeshan Fe deposit.

Fig. 3 .
Fig. 3. Geologic map of the Yeshan Fe deposit showing the distribution of orebodies.

Fig. 4 .
Fig. 4. (a) Geological map of the Xiaomiaochen area.(b) Cross-section A-B showing the spatial relationship among the marble, the intrusion, the skarns, and the orebodies in the Yeshan deposit, as well as the location of samples used in the study.

Fig. 10 .
Fig. 10.Isocon diagrams used to characterize element mobility during skarn alteration of quartz monzonite in the Yeshan Fe deposit.Element concentrations are arbitrarily scaled to avoid stacking.The original composition of quartz monzonite used for calculation is defined by the average of samples ZK302-870 and ZK302-886.The black line represents the isocon.Elements plotting above the isocon indicate enrichment/gains, whereas those below indicate depletion/losses.Open and solid circles denote the light rare earth elements (LREE) and heavy rare earth elements (HREE), respectively.(a) Skarn-altered quartz monzonite (ZK302-670).(b) Garnet skarn (ZK302-390).
S.Zhao et al.