An Exploration Study of the Kagenfels and Natzwiller Granites, Northern Vosges Mountains, France: A Combined Approach of Stream Sediment Geochemistry and Automated Mineralogy

Following a regional reconnaissance stream sediment survey that was carried out in the northern Vosges Mountains in 1983, a total of 20 stream sediment samples were collected with the aim of assessing the regional prospectivity for the granite-hosted base and rare metal mineralisation of the northern Vosges magmatic suite near Schirmeck. A particular focus of the investigation was the suspected presence of W, Nb and Ta geochemical occurrences in S-type (Kagenfels) and I-S-type (Natzwiller) granites outlined in public domain data. Multi-element geochemical assays revealed the presence of fault-controlled Sn, W, Nb mineralisation assemblages along the margins of the Natzwiller and Kagenfels granites. Characteristic geochemical fractionation and principal component analysis (PCA) trends along with mineralogical evidence in the form of cassiterite, wolframite, ilmenorutile and columbite phases and muscovite–chlorite–tourmaline hydrothermal alteration association assemblages in stream sediments demonstrate that, in the northern Vosges, Stype and fractionated hybrid I-S-type granites are enriched in incompatible, late-stage magmatic elements. This is attributed to magmatic fractionation and hydrothermal alteration trends and the presence of fluxing elements in late-stage granitic melts. This study shows that the fractionated granite suites in the northern Vosges Mountains contain rare metal mineralisation indicators and therefore represent possible targets for follow-up mineral exploration. The application of automated mineralogy (QEMSCAN®) in regional stream sediment sampling added significant value by linking geochemistry and mineralogy.


Introduction
Since 2010, the global drive for clean energy and industrial metals led to the delineation and shortlisting of a number of "critical" metals by the European Union [1]. Of the 26 metals listed in the 2017 report, high-tech metals, such as W, Nb and Ta, are of significant importance to the European manufacturing industry. Recent exploration for these metal deposits has been ongoing for more than two decades and has been predominantly focused around known prospective European Variscan belts, such as the Erzgebirge and Cornwall, and characteristic S-type granite provinces therein [2]. A recent regional geochemical reconnaissance sampling campaign, however, identified I-type granites of the lesser known central Vosges Mountains near Sainte-Marie-aux-Mines (France) to be prospective for W and Li-Cs-Ta [3]. Similarly, an investigation of Bureau de Recherches Géologiques et Minières (BRGM) public domain data [4,5] led to the delineation of a distinct W anomaly in the Natzwiller Granite, which is located in the northern domain of the Vosges Mountains. The northern Vosges Mountains comprise a series of late Devonian-Permian intrusions which show a distinct development from primitive mantle to highly fractionated peraluminous melts and the emplacement of S and I-S-type granites. Detailed studies into the petrology and geochronology of the magmatic suites allowed the reconstruction of a complex magmatic system and evolution of tectonic processes during the Variscan Orogeny [6,7]. The present study aims to investigate the granite-hosted W, Li, Nb and Ta mineralisation potential of the northern Vosges Mountains and provides a link between chemical and mineralogical analyses of regional stream sediment samples and magmatic processes outlined in previous studies. The application of automated mineralogical techniques (QEMSCAN ® ) to stream sediment samples furthermore aims to demonstrate the usability of this technique in early stage mineral exploration and therefore adds to the limited literature available on the subject [8][9][10]. Consequently, this paper provides an economic perspective to previous research carried out on the northern Vosges magmatic suite.

Regional Geological Setting
The Vosges Mountains (NE France) are part of the European Variscan orogenic belt and consist of three primary geological domains: the Southern Paleozoic basin (Southern Vosges), the central high grade gneiss and granulite domain (Central Vosges), and the northern low-grade Palaeozoic sedimentary succession (Northern Vosges) [11] (Figure 1a). In this context, the Late Carboniferous Lalaye-Lubine suture, a dextral shear zone, represents the boundary between the Saxothuringian domain of the northern Vosges and the Moldanubian domain of the central and southern Vosges [12]. Throughout the Vosges, a complex suite of Carboniferous magmatic rocks intruded into the early Palaeozoic-Carboniferous basement, predominantly comprising clastic metasedimentary rocks. Over the last three decades, these distinct magmatic suites have instigated significant interest and research into the study of Variscan tectonics and magmatic evolution [13]. northern Vosges around Schirmeck has not been taken into consideration. On the other hand, Weil (1936) [23] noted the presence of adularia, beryl, fluorine and molybdenite in a pegmatite vein ("Grotte des Partisans") located in Kagenfels Granite between Schirmeck, Rothau and Natzwiller. The mineralogical observations confirm the peraluminous nature of S-type Kagenfels Granite and imply that late stage fractionation processes led, locally, to the enrichment of incompatible elements. Leduc (1984) [5] reported results of 905 regional stream sediment samples (data available from BRGM, 2019 [24]) collected across a 330 km 2 extensive area of the northern Vosges Mountains. This survey employed a 125 µm stream sediment fraction along with plasma emission spectroscopy analysis for 22 elements; however, the report does not specify details on the acid leach and analytical precision of the analyses, although these are the only data available for this area. The study resulted in the delineation of a number of polymetallic anomalies in the Fouday, Natzwiller and Kagenfels Granites. For example, distinctive anomalies-i.e., >98th percentile-of W (20-56 ppm) and Sn (27-35 ppm) can clearly be recognised in the eastern parts of the Natzwiller and Kagenfels Granites, respectively (Figure 2a,b, Table 1). Furthermore, elevated concentrations-i.e., >90th percentile-of Nb (41-57 ppm) are present in the Kagenfels Granite and streams draining the Kagenfels area to the north and east (Figure 2c). An analysis of the spatial distribution of Nb is not discussed in Leduc's report. Boron generally shows subdued concentrations of <11 ppm throughout the Natzwiller and Kagenfels Granites, whereas elevated concentrations of 22-63 ppm can be observed in the southeastern margin of the Kagenfels Granite, Ville and Steige schists ( Figure 2d). Table 1. Summary statistics of 579 out of 905 historic BRGM samples located in the study area ( Figure  1). The 579 samples were selected from the larger dataset so that the northern Vosges magmatic suite around Hohwald, Champ du Feu, Natzwiller, and imminently adjacent portions of the Bruche Unit and Schirmeck Volcanic Massif, were covered. In total, 22 elements were analysed using Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES); however, only selected available elements of the typical granite-related mineralisation suite are presented here. Note the distinct anomalies of Sn, W (>99th percentile) and Nb (>98th percentile).

Methodology
In order to obtain an independent dataset and a comprehensive geochemical analysis suite characterised by low detection limits, 20 stream sediment samples (samples [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] were collected from first and second-order streams in the Hohwald, Natzwiller, Schirmeck and Grendelbruch areas (Figure 1b). The sample locations were principally designed to test distinctive structural and geochemical trends identified from previous BRGM investigations [5,24] as well as to characterise the different intrusive units across the northern Vosges. The sampling strategy mirrors the workflow and orientation study described in [3]. Samples were collected from stream traps, such as in the lee of large boulders or on point bars, and sieved to retain the <2 mm fraction in the field. The resulting material yielded average weights of 500 g per sample. Samples were stored in plastic bags, zip-tied and labelled with sample ID, coordinates and elevation information. The fine fraction was allowed to settle in the bag before excess water was poured back into the stream. After each sampling location, the equipment was thoroughly cleaned to prevent cross-contamination. A detailed list of stream sample attribute data (colour, grain and mesh size, contamination, trap type, etc.) was recorded on an iPad using ESRI's 'Collector for ArcGIS' app (Version 19.0, ESRI, Redlands, CA, USA). Daily data quality checks and synchronisation with a master database ensured that the data quality was consistent throughout the sampling campaign. Detailed observations and comparisons of drainage sediment composition, outcropping adjacent lithologies and heavy minerals present in pans were noted and supported the lithological classification of samples, along with the identification of fractionated lithologies; for example, quartz or pegmatite-rich rock. Linking observations of stream sediments and adjacent outcrops confirmed that stream sediments appeared unweathered and accurately represent the overall bedrock geology of respective catchment areas, and therefore can be used for further representative lithogeochemical investigations. While glaciation occurred in centralwestern Europe and affected parts of the central-southern Vosges Mountains, the only glacial debris related to the last glacial event (Weichselian) are recorded in the Bruche River valley. No evidence of glacial sediment was encountered during fieldwork.
Samples were returned to Camborne School of Mines, University of Exeter (UK) laboratories and dried in laboratory ovens at a constant temperature of 40 °C. The samples were then sieved using a sieve stack and a Pascal Sieve Shaker to isolate the <75 µm fraction which was previously identified to be host to W, Li and Cu anomalies in the Vosges [3]. Individual fraction weights were determined to assess if sample loss had occurred. All samples underwent ICP-MS analysis using a standard four acid digest (HCl-HF-HNO3-HClO4) and Agilent 7700 ICP-MS instrument (Agilent Technologies, Santa Clara, CA, USA), so that, compared to historic studies, a wider range of trace and pathfinder elements, including Nb, Ta, Li, Hf, could be obtained, aiding the determination of fractionation trends ( Table 2). Quality control was ensured by inserting silica blanks, duplicate samples and OREAS 147 and 751-certified reference materials into the sample stream, with no accuracy issues noted outside +/− 1 standard deviation of the certified mean value. In a subsequent geochemical interpretation, following a previously published approach in the central Vosges [3] and southeast Ireland [30], the additional ICP-MS assays supported the usage of multi-element major and trace geochemistry in classifying lithological populations and petrogenetic and mineralisation processes. The classification of lithological units was achieved by delineating population clusters in bivariate geochemical plots. Each sample point was assigned a lithology, which was refined using geological observations in upstream catchment areas, outcrops and float in order to better represent subtle nuances in geochemical composition, such as "Natzwiller/Belmont Granite with visible cassiterite".
Of the 20 stream sediment samples (Samples 1-20) undergoing ICP-MS analysis, nine samples (2,3,14,15,16,17,18,19,20) of the <75 µm sample fraction were selected for mineralogical analysis. In addition, two rock samples (17A and 18A) were collected within a 10 m distance upstream of their corresponding stream sediment sample locations (17 and 18), and a mineralogical analysis (QEMSCAN ® ) was performed on uncovered polished thin sections of these samples. Mineralogical analysis was conducted on the fine <75 um stream sediment fraction, which is considered homogeneous and allowed a direct comparison between mineralogy and geochemistry. The samples were prepared into 30 mm diameter epoxy resin mounts and mixed with pure graphite powder to reduce settling bias and separate particles. The sample surface of the cured mounts was carefully ground to expose the particles and polished to a 1 micron finish using diamond media, then carboncoated to 25 nm thickness. Samples were analysed using a QEMSCAN ® 4300 [31][32][33][34] at Camborne School of Mines, University of Exeter, UK. Sample measurement used iMeasure version 4.2SR1 software for data collection and iDiscover 4.2SR1 and 4.3 software for data processing. The Particle Mineral Analysis (PMA) measurement mode was used to map particles at a resolution (pixel spacing) of 2 µm, field size of 600 µm (300 × 300 square, magnification of ×111), default of 1000 X-ray counts per analysis point and a target of 10,000 particles per sample. The final number of particles mapped per sample was higher than this (up to 14,556) due to the system completing the particles in the field it was on when it reached its 10,000-particle target. The number of analysis points per sample varied from 900,000 to 4 million.
The data collected during measurement were processed using a modified version of the standard LCU5 SIP (database), following and building upon details outlined in section 7 of Rollinson et al. (2011) [35]. Both mineral area-% and mineral mass-% (density weighted) data were produced, and it was decided to use the mineral mass-% data as they better reflect the economic mineral content of the samples. As these were stream sediments, the focus was on both the major minerals and trace or unusual minerals, with the SIP customised to reflect the mineralogy of samples. This included checking the mineral identification not just from the measured chemical spectra, but also against inhouse mineral reference standards. Moreover, checks were also completed for possible Li minerals following the method developed at Camborne School of Mines during the FAME EU Horizon 2020 project [36]. In the case of identified Ta-Nb minerals, independent SEM-EDS checks were also conducted. Table 2. Summary statistics of elements and elemental ratios obtained for the present geochemical study. The stream sediment grain size fraction analysed was <75 µm. All 20 samples were analysed by ICP-MS using a four-acid digest (HCl-HF-HNO3-HClO4).

Univariate Anomaly Analysis of Ore and Incompatible Elements
Graduated point symbol maps of incompatible elements ( Figure 3) show an overall enrichment above the 95th percentile of W (99.92 ppm), Nb (197.99 ppm), Ta (11.72 ppm) and Be (33.71 ppm) in Kagenfels Granite. Compared to average concentrations of W (2 ppm), Nb (20 ppm), Ta (3.5 ppm) and Be (5 ppm) in granites [37], the Kagenfels Granite is locally characterised by an enrichment factor of more than 50 (W), 10 (Nb), 3.4 (Ta) and 6.7 (Be). The anomalous samples are located approximately 600-900 m north of the Natzwiller-Struthof camp at confluences with the Barembach stream (samples 13-19, Figure 1b) as well as the Magel stream to the east (samples [10][11]. The anomalous zones principally correlate with regional NNE-SSW trending fault zones, where this study additionally located occurrences of 2-3 m wide, partially overburden-covered quartz-feldspar pegmatitic granite stringers trending parallel to these fault zones, particularly at sample location 18. In addition, the Kagenfels Granite is characterized by distinct magmatic fractionation ratios of 64.5 < K/Rb < 110, 13.98 < Nb/Ta < 17.75 and 19.12 < Zr/Hf < 37.91. On the other hand, the southwestern part of the Natzwiller Granite displays an enrichment of Li (105.1 ppm), W (34.81 ppm) and B (21.99 ppm) corresponding to enrichment factors of 3.5 (Li), 17.4 (W) and 1.5 (B), respectively. Fractionation ratios of 101 < K/Rb < 128.8, 12.35 < Nb/Ta < 12.7, 33.85 < Zr/Hf < 36.45 are slightly more elevated than the Kagenfels Granite, indicating generally less fractionation. Sample 20, located on a fault-controlled boundary between the Natzwiller Granite and Belmont Granite Suite, shows visible cassiterite in a stream pan, but only limited concentrations of Sn (17.98 ppm) and the highest K/Rb (273.2) ratio of the sampling campaign. The presence of coarse cassiterite visible in a pan and limited concentrations of Sn in the < 75 um fraction suggest a strong partitioning of Sn by size fraction. Cassiterite was not abraded and comminuted to finer size fractions due to the limited fluvial transport and therefore is thought to originate from a proximal source. The Belmont Granite Suite, comprising the Fouday, Waldersbach, Champ du Feu North Granites along with three samples collected in the older Neuntelstein Diorite and Hohwald Granodiorites, is typically characterised by an absence of notable concentrations of incompatible elements, whereby fractionation ratios of 113.5 < K/Rb < 237.1, 37.42 < Zr/Hf < 45 imply a less fractionated nature than the Natzwiller and Kagenfels Granites. The calculated Nb/Ta ratio of 8-13.1, however, is considerably lower than the Natzwiller and Kagenfels Granites.

Geochemical Classification of Principal Rock Units and Magmatic Fractionation Using Stream Sediment Data
The analysis of trace element concentrations, supported by geological observations in outcrops and stream catchment areas, led to the classification of four major regional lithology types ( Figure 4). Stream sediment geochemistry was classified per lithology unit, and the link between the stream sediment geochemistry and lithology was investigated. The lithologies principally reflect the regional BRGM geological map of the northern Vosges Mountains. Major and trace elements and ratios used in this study are predominantly Ti, Th, B, K/Rb, Nb/Ta, and Zr/Hf. A combination of these elements supports the determination of significant petrological and fractionation processes characteristic for each lithology of the northern Vosges magmatic suite.
Kagenfels Granite was macroscopically distinguished in outcrop and stream sediments from other lithologies by the presence of muscovite, biotite, ilmenite, and minor beryl. The lithology forms distinct population clusters in Ti vs. Th, K/Rb vs. Th and Ti vs. B bivariate plots with concentrations of Th (38.33-86.53 ppm), Ti (2170-5683 ppm), B (9.9-13.85 ppm), 64.5 < K/Rb < 110, and 19.12 < Zr/Hf < 37.91. Whilst the K/Rb and Zr/Hf ratio of samples 10-11, 13-14, and 18 represent the lowest, and therefore most fractionated, magmatic fractionation indicator values in the studied samples, Nb/Ta ratios of 13.98-17.75 are unexpectedly high. The Belmont Granite Suite comprising the Fouday, Champ du Feu and Waldersbach granites [6,7] forms a distinct population cluster in Ti vs. Th, Ti vs. B, and K/Rb vs. Th bivariate plots, and can be clearly distinguished from the Kagenfels and Natzwiller granites by lower concentrations of the aforementioned elements, particularly Th (8.87-24.45 ppm), and has K/Rb and Zr/Hf values ranging from 105-165 and 37.42-45, respectively. The Hohwald Granodiorite (samples 5-6) and the Neuntelstein Diorite (sample 7) plot within this lithogeochemical population have therefore been included in the present Belmont Granite suite classification.
A principal component analysis (PCA) of the geochemical dataset was carried out using a log10 transformation (to eliminate the closure effect) of B, Be, Cu, Li, Nb, Rb, Sn, Ta, Th, Ti, W and Zr input variables (Table 3, Figure 5). The PCA revealed that PC1 to PC4 represent 85.68% of the total variance of the dataset, whereas the remaining eight PCs can be attributed to random processes or noise. The eigenvector table and combined variable-sample analysis (RQ) plots, in which samples plot as points and variables as vectors with the length of the vectors proportional to the variability of the two displayed principal components, outline two major correlations and elemental trends. Firstly, previously classified Kagenfels and Natzwiller Granites are principally characterised by negative RQ1 loadings, whereas Natzwiller Granite can be distinguished from Kagenfels Granite by positive RQ2 and RQ4 loadings. On the other hand, there is a noteworthy variability of the Kagenfels Granite samples 10-11, 13-14 and 18, expressed by generally negative RQ2 and RQ4 loadings of Rb and Zr, respectively. In the K/Rb vs. Zr/Hf plot (Figure 4e), these samples represent the most fractionated population with K/Rb < 120 and Zr/Hf < 32. Therefore, the PCA trends are distinctive for the previously outlined fractionated lithogeochemical populations and characterised by a Ta-Nb-Be-Rb-Th-W and Zr geochemical association. In particular, whilst PC1 broadly characterises fractionated lithologies of the Kagenfels and Natzwiller Granites in the study area, PC2 and PC4 allow the determination of local variation of fractionation in the Kagenfels Granite. In contrast, the Belmont Granite Suite and the Natzwiller/Belmont Granite with visible cassiterite (sample 20) show fundamentally different trends with positive RQ1 loadings, and a Cu-Ti and Li geochemical association. Distinctive negative and positive variability in RQ2 and RQ4, respectively, allow the determination of potential sub-populations, such as samples 6 (Hohwald Granodiorite) and 7 (Neuntelstein Diorite), which are generally less evolved and have intruded approximately 10 Ma earlier than the Belmont Suite [6]. In summary, the analysis of petrogenetic indicator elements in bivariate plots along with principal component analysis provide a robust tool to determine different litho-geochemical populations and element associations in the study area.

Correlation between Outcrop and Stream Sediment Sample Mineralogy
In order to ascertain the representativity of stream sediment samples and previously mapped lithological units, a comparison between the modal mineralogy of stream sediment and adjacent outcrop samples has been undertaken. Stream sediment samples 17 and 18 are located in the northern part of the Kagenfels Granite unit and are enriched in W (102.56 ppm and 6.94 ppm) and Nb (63.67 ppm and 92.76 ppm), respectively. Visual field mapping and description of drainage areas and sediments confirm the presence of a muscovite-biotite(-tourmaline) granite, along with sporadically occurring pods of a finer, hornblende-bearing variety, which was previously described as the less evolved "rhyolite facies" of the Kagenfels Granite [14]. Abundant outcrops of the granite are present several hundred metres upstream of the sampled locations, whilst pods of the "rhyolite facies" are encountered in the vicinity of the stream sources along the margins of the granite. Sample 18 yielded minor columbite grains during routine pan inspection. The QEMSCAN modal mineralogy of stream sediment samples 17 and 18 (Table 4) confirms the peraluminous monzogranitic nature of the samples containing tourmaline (1.43%, 0.82%), muscovite (1.34% and 1.37%), and spessartine (0.14% and 0.65%) as abundant accessory Al-rich mineral phases. Furthermore, wolframite (0.03% in 2060) and ilmenorutile (0.01% and 0.05%) represent the hosts for W and Nb-rich accessory mineral phases, which correlate with the geochemical anomalies outlined in this study.    The mineralogical analysis of the collected samples confirms that major silicate phases (e.g., quartz, feldspars, micas, and tourmaline) of the −75 µm fraction are not or very weakly weathered and generally do not show weathering rims or the complete replacement of the mineral grains with kaolinite ( Figure 6). This implies a rapid erosion of outcrops and (re)deposition in stream traps. Variable amounts of Fe-Mn oxides are present in the samples and indicate that the weathering of primary Fe-rich minerals can to some extent affect stream sediments of the study area.
The outcropping muscovite-biotite(-tourmaline) host granite bedrock samples 17B and 18B display a comparable monzogranitic composition and variably contain more abundant K-Feldspar (10-14%) and quartz (5-10%) than the stream sediments. Importantly, both rock samples are characterised by similar, yet slightly subdued, amounts of key accessory minerals, such as tourmaline (0.29% and 0.22%), muscovite (0.46% and 0.42%), ilmenite (0.09% and 0.18%) and rutile (0.04%). Due to local variations in bedrock geology and the nature of sample origin and preparation techniques (i.e., a thin section of a singular outcrop chip vs. well-mixed stream sediment samples), trace and heavy minerals will naturally be more abundant in samples obtained from stream traps where the accumulation of denser mineral grains is more pronounced. Conversely, minerals with a lower specific gravity, such as quartz, mica and feldspar, will likely be less abundant in stream sediments than in outcrop. The mineralogical analysis of stream sediments and outcropping bedrock, however, demonstrates that the mineralogical composition of stream sediments is generally indicative of its corresponding source lithology. As a result, it can be assumed that the largely unweathered stream sediment samples reflect the overall lithological composition and are representative of the stream catchments.

Petrogenetic Indicators
Other samples obtained and analysed from the Kagenfels Granite (Appendix A) show similar modal mineralogical compositions compared to stream sediment samples 17 and 18 and therefore provide evidence for peraluminous two-mica monzogranite. In particular, stream sediment samples [14][15][16]19 (Figure 9), which, together with a high geochemical Nb/Ta ratio of 14, implies that Nb-rich mineral phases are distinctively more abundant than Ta phases in the western part of the Kagenfels Granite. Sample 18 ( Figure 10) shows that chlorite is associated with tourmaline and muscovite and therefore represents evidence for granite-related magmatic-hydrothermal alteration [38]. However, due to the size of the fragments and the absence of quartz, it is not entirely conclusive whether this mineral association is related to veins or wall-rock alteration.
Stream sediment samples 2 and 3, obtained from the two principal drainages on the southwestern flank of the Natzwiller Granite, generally show a similar monzogranitic composition as the Kagenfels Granite. Accessory Th-and Ti-bearing mineral phases in Natzwiller and Kagenfels Granites (Table 4) indicate a comparable content of Th-rich mineral phases, with the highest content (1.91 mass-%) of Ti-phases in sample 3. This generally corresponds to the Th vs. Ti bivariate plot (Figure 5a), where the Natzwiller Granite samples have Ti concentrations of >8640 ppm. However, given the notable enrichment of Ti in the geochemical analyses, a higher abundance of Ti mineral phases would be expected in the mineralogical analysis. However, this may be explained by the occurrence of trace Ti occurring in other minerals, such as micas [39]. A distinctive feature of samples 2 and 3, on the other hand, is the abundance of tourmaline (8.48% and 3.39%) and chlorite (13.61% and 6.88%), which in Figure 5b Table 4. Modal mineralogy (mineral mass-%) of selected stream sediment and outcrop samples of the Kagenfels (17A and 18B) and Natzwiller Granites (2 and 3). The four stream sediment samples show a broadly similar monzogranitic composition; however, the Natzwiller Granite has generally more abundant B and Ti-rich mineral phasese.g., tourmaline and rutile. Outcropping rocks and stream sediments of the Kagenfels granite have a comparable mineralogical composition; however, with a subdued heavy mineral signature. Please refer to Appendix 1 for a complete list of sample mineralogy tables.

Discussion
The graduated point symbol and bivariate fractionation plots demonstrate a distinctive endowment of base metals and high field strength elements (HFSE) in the northern Vosges magmatic suite, with a particular emphasis on the Natzwiller and Kagenfels Granite suites. The geochemical trends observed in these plots are comparable with previous regional reconnaissance sampling [5] and whole-rock geochemical data presented in Tabaud et al. (2014) [6], particularly for Th, Sr, Rb and Ti, which provide a tool to distinguish intermediate and felsic magmatic rocks. The application of univariate anomaly mapping and the lithogeochemical classification of catchment sediments, therefore, supports the delineation of areas of increased enrichment of economically sought-after metals and corresponding magmatic lithologies. In particular, the present data confirm the known presence of Nb and Be and outline additional, previously unrecognised W anomalies in Kagenfels Granite.
Earlier empirical mineral and whole-rock geochemical studies have successfully demonstrated the use and application of magmatic fractionation ratios in defining late-stage magmatic melts prospective for Sn-W and Li-Cs-Ta mineralization [27,28,[40][41][42][43]. Decreasing K/Rb, Nb/Ta, and Zr/Hf ratios indicate the increasing fractionation of the granitic melt and a transition to hydrothermal alteration [27]. Numerical changes in these ratios during late-stage magmatic fractionation are a result of the substitution of K with Rb in micas and feldspars [44], fractionation of Nb over Ta due to secondary muscovitisation and hydrothermal sub-solidus reactions enriching Ta in F-rich residual melts [29], and increasing Kd values of Hf in zircon, which are only weakly influenced by secondary fluid-related processes [28,45,46]. Recent geochemical studies of the Leinster Granite (Ireland) and Central Vosges Mg-K granites [3,30], along with mineralogical results of this study, have shown that these petrogenetic ratios are equally applicable to determine highly fractionated lithologies using geological materials affected by secondary dispersion processes, such as stream sediments. The lack of significant weathering of K, Rb, Sn, Nb and Zr-bearing silicate mineral phases in the analysed stream sediment fraction that were collected as well as the representativity of these stream sediments in relation to mapped and sampled outcrops of the catchment area confirm that these petrogenetic ratios can be employed to fingerprint fractionation patterns in the study area. In this context, the Natzwiller Granite shows fractionation ratios of 101 < K/Rb < 128.8, 12.3 < Nb/Ta < 12.7, and 33 < Zr/Hf < 36 (Figure 3i ). In addition, despite a relatively low concentration of Sn (17.98 ppm) being measured in sample 20, automated mineralogical techniques identified multiple cassiterite grains in the stream sediment sample. This evidence suggests that the Natzwiller Granite has experienced, at least locally around NE-SW trending fault zones, the introduction of a highly fractionated melt enriched in fluxing elements, which allows incompatible and HFS elements to remain in late-stage, low-temperature melts [29]. Tabaud et al. (2014) [6] previously described the Natzwiller Granite as sub-aluminous to weakly peraluminous in nature, resulting from partial melting of an enriched mantle source and subsequent interaction with subducted metasedimentary and metaigneous crustal source material. Therefore, despite the comparably lower fractionation grade, the Natzwiller Granite was able to retain incompatible elements of possible economic interest.
In contrast, the peraluminous S-type Kagenfels Granite typically displays low values of 64.5 < K/Rb < 119, 14 < Nb/Ta < 17.7, and 19 < Zr/Hf < 38 (Figure 3i-k), and therefore demonstrates a high degree of magmatic fractionation and secondary muscovitisation, characteristic of peraluminous Stype granites [27,42]. Geochemical fractionation trends ( Figure 5), PC analysis ( Figure 6) and mineralogical evidence in the form of tourmaline, muscovite, chlorite, wolframite, cassiterite, columbite and ilmenorutile imply a peraluminous evolution and confirm a highly fractionated and locally hydrothermally altered nature of the Kagenfels Granite. Strong fractionation of the melt, along with a predominant NE-SW structural control in the Kagenfels Granite, led to the emplacement of pegmatitic quartz-feldspar (-beryl) veins at "Grotte des Partisans" [23] and the Barembach stream confluences observed in the present study. The same process also resulted in characteristic elemental concentrations of As (25.52-92 ppm), Cu (7.88-166.61 ppm) and incompatible elements, such as Be (10-21 ppm), Sn (7.19-26.29 ppm), Li (12.13-52.61 ppm), and W (6.94-102.56 ppm). However, distinctive high values of Nb/Ta > 17 and the general absence of Li concentrations of >45 ppm across the western part of the Kagenfels Granite (Figure 3h,j) suggest that, locally, magmatic fractionation and hydrothermal alteration processes were not as pronounced as in the eastern part of the Kagenfels Granite, which yields higher concentrations of Be (24-34.22 ppm), and Li (49. 35-54.77 ppm), at 16.4 < Nb/Ta < 16.8, 64.5 < K/Rb < 93, and 19 < Zr/Hf < 25 (Figure 3h-k). The comparatively higher Nb/Ta ratios in the western Kagenfels Granite are a result of the predominant occurrence of Mn-columbite, euxenite-(Y), and Nb-rich titanium oxide/ ilmenorutile as evidenced in stream sediment sample 17 and 18. Consequently, the western part of the Kagenfels Granite predominantly produced a mineral assemblage with Nb > Ta-rich minerals, indicating the preferential fractionation of Nb over Ta-rich minerals and, therefore, the locally lower fractionation of the granitic melt. In a regional context, these observations imply that the Kagenfels Granite is the most fractionated granite suite in the northern Vosges Mountains, and consequently represents a prime target to explore for granite-hosted mineralisation.
The application of automated mineralogical techniques, such as QEMSCAN ® and manual SEM-EDS, supported the routine collection of stream sediment samples in a mineral exploration context. These techniques were not only able to identify the bulk mineralogical composition of the stream sediment samples and therefore link geochemical signatures to source mineralogy, but also provide potential information about element deportment characteristics (ilmenorutile and columbite as principal Nb hosts, wolframite as principal W host, cassiterite as principal Sn host) for mineral processing-related studies. Therefore, this study, along with previous investigations into heavy mineral ilmenite deposits in India [8,9], demonstrates the usefulness of automated mineralogical techniques in early-stage exploration campaigns, which often predominantly involve the routine collection and analysis of stream sediment, soil and till samples and do not necessarily involve the link between sample geochemistry and mineralogical response and element deportment.

Conclusions and Implications for Mineral Exploration
This regional follow-up stream sediment exploration study has outlined geochemical and mineralogical evidence to support the previously established presence of Sn-W mineralization [5] and has provided new insights into this mineralisation suite, particularly regarding potential Li-Cs-Ta-Nb mineralisation in the S-type Kagenfels and I-S-type Natzwiller Granite suites of the northern Vosges Mountains. The occurrence of late-stage, incompatible elements in strongly fractionated lithologies and minerals implies the presence of a local mineralisation system. The Natzwiller and Kagenfels granites of the northern Vosges therefore warrant further exploratory work, particularly in areas of known structural control and occurrence of pegmatitic quartz-feldspar (-beryl) vein systems. The study has shown that automated mineralogical techniques can routinely be used in conjunction with univariate, multivariate and litho-geochemistry to determine the nature of stream sediments, source rock and generated exploration targets. A combined geochemical and mineralogical approach in early-stage grassroots exploration is therefore beneficial and useful when screening large areas during routine surveys.  Table A2. Raw geochemical data for the present geochemical study. The stream sediment grain size fraction analysed was <75 µm. All 20 samples were analysed by ICP-MS using a four-acid digest (HCl-HF-HNO3-HClO4).