Geology of the Tabakoto gold deposit, Kédougou-Kéniéba Inlier, West African Craton, Mali

In the past decade, numerous studies conducted on the southern part of the West African Craton (WAC) have led to its recognition as the foremost Paleoproterozoic gold province globally (Goldfarb et al., 2017). Paleoproterozoic gold deposits in the WAC (Fig. 1a) is marked by the three major tectonic episodes (Masurel et al., 2021): the Eoeburnean (2200 to 2135 Ma), the early Eburnean (2110 to 2095 Ma), and the late Eburnean (2095 to 2060 Ma), each characterized by complex deformation, metamorphic, and magmatic processes. Although several styles of gold mineralization developed during these episodes (Goldfarb et al., 2017; Masurel et al., 2021), by far the majority correspond to the orogenic deposit type (e.g., Milési et al., 1992; Béziat et al., 2008;Feybesse et al., 2006; Lawrence et al., 2013a; Baratoux et al., 2015; Markwitz et al., 2016; Goldfarb et al., 2017; Masurel et al., 2017a; Thébaud et al., 2020; Masurel et al., 2021). Deposits that have not been classified as orogenic include (i) paleo-placers associated with quartz pebble conglomerates in the Ashanti belt in Ghana (e.g., Milési et al., 1989; Hirdes and Nunoo, 1994), (ii) the porphyry-Cu type deposit at Gaoua in Burkina Faso (Le Mignot et al., 2017), (iii) the skarn type deposits of Ity and Tongon in Côte d’Ivoire (e.g., Béziat et al., 2016; Lawrence et al., 2017), and (iv) the intrusion-related gold systems at Morila in Mali (e.g., McFarlane et al., 2011) and Bonikro in Côte d’Ivoire (Masurel et al., 2019). This last type of deposit is distinguished by the metal association of Au-Bi-Te-As and is spatially and temporally related to moderately-reduced granitic plutonism (Sillitoe and Thompson, 1998; Thompson and Newberry, 2000; Hart, 2007).

The Paleoproterozoic Kédougou-Kéniéba Inlier (KKI; Fig. 1b) represents the second-largest gold endowment in the West African Craton after the Ashanti belt in Ghana (e.g., Goldfarb et al., 2017). Similarly to the rest of the WAC, most of the Paleoproterozoic gold deposits in the KKI are of the orogenic type. Examples include the Sabodala deposit (Senegal) (Fig. 1b) hosted in mafic and ultramafic rocks (Sylla et al., 2016), and the Massawa deposit (Fig. 1b) localized in shear zones within greenschist facies volcanoclastic rocks and metagreywacke, both in the Mako belt (Treloar et al., 2014). In the eastern reaches of the KKI, within the Kofi series (essentially in Mali), a number of well-studied deposits (Fig. 1b) occur in the vicinity of granitoid bodies, which has prompted researchers to inquire about the connection between magmatic activity and mineralization. These include the Gara deposit (Loulo district), hosted in tourmaline-rich folded metaturbidites (Dommanget et al., 1993, Lawrence et al., 2013b); the Sadiola deposit, hosted in detrital metasediments and impure marbles (Masurel et al., 2017a, 2017b); and the Gounkoto deposit, hosted in metagreywacke with sodic or phyllic alteration (Lambert-Smith et al., 2016a). However, most authors conclude that these deposits are also of orogenic type (e.g., Lambert-Smith et al., 2016b; Masurel et al., 2017b, 2017c; Allibone et al., 2020). Nevertheless, deposits of types other than orogenic are described in the KKI. These include the Alamoutala skarn deposit (Masurel et al., 2017b) and the Yatela deposit (Fig. 1b), latter been a residuum type deposit hosted in marble and diorite altered to saprolite (Hein et al., 2015; Masurel et al., 2016).

The Tabakoto gold deposit, located within the Kofi metasedimentary series of the KKI (Fig. 1b), has not been subject to scientific studies yet. Prospects have shown evidence for multiple mineralization events related to polyphased deformation, as well as a close association with several generations of intersecting magmatic dikes, which encouraged previous exploration campaignies (e.g. Nevsun Resources Ltd., Avion Resources Corporation, Endeavour Mining Corporation) to question the role of magmatism on the mineralization (e.g., Nielsen, 2002; Hein, 2008; De Hert et al., 2014).

In this paper, we present new structural, petrological, mineralogical and geochemical data from both mineralized ore and barren host rocks of the Tabakoto deposit, and establish the spatial and temporal relationships between deformation, hydrothermal alteration and gold mineralization phases. These data provide new insights to discuss the chronology and mechanisms of gold mobilization and deposition in the framework of the different tectonic-magmatic-hydrothermal phases, and put it into perspective at the scale of the KKI.

The Tabakoto gold deposit belongs to the western Malian gold district located in the eastern part of the Paleoproterozoic Kédougou-Kéniéba Inlier (Fig. 1b). The KKI consists of Birimian volcano-plutonic belts known as the Mako and Falémé greenstone belts, alternating with metasedimentary Dialé-Daléma and Kofi series (Fig. 1b), which were deformed and metamorphosed during the Eburnean orogeny (Bonhomme, 1962; Ledru et al., 1991). The NNE-SSW to N-S trending Mako belt (MB, Fig. 1b) is composed of basalt, locally displaying pillow structures, gabbro and ultramafic rocks of tholeiitic affinity interlayered with immature detrital sediments and a more or less differentiated sequence with calc-alkaline affinity (gabbro, basalt, andesite, dacite, and rhyolite) occurring as veins cross-cutting the tholeiitic rocks (Hirdes and Davis, 2002; Dioh et al., 2006; Labou et al., 2020). The Falémé belt (FB, Fig. 1b) is made of metamorphosed pyroclastic rocks, subordinate rhyodacite lavas, and andesite flows with preserved pillow structures, interbedded with metavolcanoclastics, metagreywackes and metacarbonates (Bassot, 1987; Hirdes and Davis, 2002). The protoliths of the Dialé-Daléma (DDS) and Kofi (KS) series (Fig. 1b) are dominated by sediments, largely derived from erosion of the Mako belt (Bassot, 1966; Koné et al., 2020). The metasedimentary rocks are intercalated with volcanoclastic rocks and rhyolite flows (Bassot, 1987; Hirdes and Davis, 2002). These features evoke an epicontinental paleoenvironment close to a magmatic arc.

The different belts and series of the inlier are intruded by numerous granitic to diorite plutons (Dioh et al., 2006; Gueye et al., 2008; Masurel et al., 2017c) and by post-Eburnean mafic dikes (Baratoux et al., 2019). The Kofi series is intruded by deformed syntectonic monzogranite plutons such as the Gamaye (Ga) and Yatea (Yt) plutons, the latter being located about 4 km to the North of the Tabakoto deposit (Fig. 1b).

The KKI displays a complex superposition of structures that has been attributed to three deformation phases with an early period of crustal thickening (D₁) followed by a period of transcurrent tectonics (D₂-D₃) characterized by ductile to brittle structures (Milési et al., 1989; Ledru et al., 1991; Diene et al., 2012; Diatta et al., 2017; Masurel et al., 2017c).

The D₁ phase was first proposed to be represented by inclined NNW-SSE to N-S trending folds with associated reverse faults accommodating ENE-WSW to E-W horizontal shortening best documented in the magmatic series of the Mako belt and the volcano-sedimentary series of the Dialé-Daléma (Milési et al., 1989; Gueye et al., 2008; Diene et al., 2012; Masurel et al., 2017c; Allibone et al., 2020). However, more recently, Diallo et al. (2020) identified a WNW-ESE to NW-SE trending S_0/1 foliation in the northern part of the Mako belt marked in regional magnetic data by magnetic lineaments and that locally correlates with isoclinal folds identified in the field.

The D₂ is manifested by regional-scale upright NNE-SSW trending folds, N-S to NNE-SSW trending thrusts and shear zones (Milési et al., 1989; Ledru et al., 1991, Dabo and Aifa, 2010;Diene et al., 2012; Lawrence et al., 2013a; Masurel et al., 2017c; Diallo et al., 2024). The D₂ is characterized by the transposition of earlier structures as identified in magnetic data by a regional deflection of WNW-ESE trending lineaments into NNE-SSW trending lineaments indicating an apparent dextral sense of shear (Diallo et al., 2020). Accordingly to previous works, D₂ marks a switch from a dominantly coaxial to a non-coaxial deformation regime (Diene et al., 2015; Lambert-Smith et al., 2016a; Masurel et al., 2017c).

The D₃ phase is marked by a transposition of N-S to NNE-SSW trending D₂ structures and localization of deformation along major shear zones such as the Senegalo-Malian Shear Zone (SMSZ; Bassot and Dommanget, 1986) and the Main Transcurrent Zone (MTZ; Ledru et al., 1991). D₃ is also accompanied by voluminous calc-alkaline magmatism (Gueye et al., 2008), circulation of hydrothermal fluids and gold mineralization along secondary structure of regional-scale shear zones and/or faults (Lawrence et al., 2013a; Treloar et al., 2014; Diene et al., 2015; Dabo et al., 2016;Lambert-Smith et al., 2016a; Masurel et al., 2017c).

As an alternative, based on the analysis of regional scale structures, Chardon et al. (2020) propose a deformation continuum characterized by anastomosed shear zones that accommodated distributed shortening and lateral flow of the orogenic lithosphere between the converging Kénéma-Man and Congo Archean provinces. In this work we will use the framework of deformation events discretized into several phases for practical purposes even if the Eburnean orogeny may also be considered as a continuum of deformation varying in intensity with time, as proposed by the latter author.

Systematic face mapping of the pit workings (Fig. 2) for Tabakoto North Pit (TNP) and underground galleries for TNP and Tabakoto Main Pit (TMP) was completed, in order to emphasize the relationship between deformation, emplacement of various dikes generations, hydrothermal alteration and gold mineralization. Logging of drill core sections was also undertaken. Samples investigated in this paper come from underground workings and drill core, and include country rocks and all types of ores. Structural data were reported as strike/dip/quadrant for the planar structures and plunge/azimuth for linear structures.

Polished thin sections were prepared from 18 representative samples at the laboratory of Geoscience Environment Toulouse (GET) and were used for petrographic observation, SEM (Scanning Electron Microscope) and EPMA (Electron Probe Micro-Analyser) investigations.

SEM analyses were performed with a JEOL JSM6360LV Microscope for SEM imaging. X-ray microanalysis (EDS: Energy Dispersive Spectrometry) were obtained with an X-Ray Microanalysis BRUKER System. This setup provided semi-quantitative chemical analyses. EPMA was used for in situ quantitative microanalysis of the chemical composition of sulfides, oxides and gold. These analyses were carried out at the Raimond Castaing micro-characterization center in Toulouse (University Toulouse III-Paul Sabatier) using a Cameca SX Five electronic microprobe composed of five wavelength dispersion spectrometers (WDS). The analysis conditions were 15 kV for the acceleration voltage, for currents of 10 or 20 nA depending on the resistance of the minerals to the electron beam. The acquisition times were 10 s at the peak and 5 s at either side of the peak for the continuous background.

Several samples of magmatic rocks (dikes) were selected for whole-rock geochemical analyses, to obtain major and trace element data (Tab. 1). These were carried out at the Central Analytical Facilities, Stellenbosch University (South Africa). Major element concentrations were obtained by X-ray fluorescence spectrometry (XRF) while trace element concentrations were obtained by ICP-MS (Inductively Coupled Plasma Mass Spectrometer) with the following analytical conditions: Agilent 8800, Carrier gas: 0.8 L/min Ar + 0.004 L/min Nitrogen. Four geostandards (BHVO & BCR glass and BHVO & BCR powder) were used to check the reliability of the data, correct the measurements for instrumental drift and recalculate concentrations.

The most abundant lithology in the Tabakoto area are metasedimentary rocks of the Kofi series, mostly metagreywacke alternating with meta-argillite. These are intruded by mafic to felsic dikes and are affected to various extent by hydrothermal alteration (Fig. 2b).

The metasedimentary rocks consist of metagreywacke intercalated with meta-argillite, ranging from decameters to meters in thickness, which are interpreted to represent a turbiditic sequence (Figs. 3a, 3b). Transposition of the sedimentary bedding into a composite foliation is accompanied by greenschist facies metamorphism.

Metagreywacke consists essentially of arkose composed of angular to subrounded quartz and feldspar floating in a matrix (40–70%), which consists of carbonate, mica (muscovite or sericite, biotite), quartz, albite and chlorite (Figs. 3c, 3d). Locally, K-feldspar is more abundant, whereby the rocks are classified as fedspathic metagreywacke. Veins and breccia are localized in metagreywacke layers (Fig. 3a) and at the contact with magmatic dikes of the first generation. These features are interpreted to reflect heterogeneous deformation and strain partitioning of the metasedimentary series coeval with the circulation of a hydrothermal fluid marked by ductile deformation of the less competent metasedimentary unit and by brittle deformation of the more competent metagreywacke and dikes localizing the precipitation of veins (Figs. 3e, 3f).

Meta-argillite facies are finely layered and display a slaty texture (Fig. 3b). They are composed of quartz, feldspars, micas (muscovite, sericite), and carbonates within a finely recrystallized matrix.

A variety of dikes, up to several meters in thickness, intrude the metasedimentary rocks at Tabakoto. Cross-cutting relationships and chemical composition of the dikes allow the distinction of (i) a first generation of dikes with a composition ranging from basaltic to rhyolitic that are affected by deformation and metamorphism; (ii) a second generation of lamprophyre dikes, and (iii) a third generation of dolerite dikes (Figs. 2–6).

Metabasaltic to metarhyolitic dikes are relatively abundant throughout the deposit (Figs. 2b–6). They are steeply dipping and are transposed in the dominant N-S to NNE-SSW strike regional foliation.

The metabasalt dikes (samples T2 and T5) have a microlithic porphyritic texture with subhedral to euhedral phenocrysts of plagioclase strongly altered to albite and sericite (60–85%) (Figs. 4a, 4b). The matrix comprises a secondary paragenesis of albitized plagioclase, amphibole microliths and recrystallized quartz and carbonate, and very small amounts of rutile and hematite (≤1%).

The meta-andesite (samples T3, T8 and T12), metadacite (samples T1, T4 and T6), and metarhyolite (sample T13) dikes also have a microlithic porphyritic textures. Meta-andesite dikes (Figs. 4c, 4d) are characterized by quartz as microcrystals and isolated phenocrysts, plagioclases in microliths and phenocrysts destabilized into albite, sericite and calcite. Biotite occurs as microliths and euhedral phenocrysts, deformed and arranged parallel to the foliation. It is altered to calcite, chlorite, muscovite and Fe oxy-hydroxides. The matrix consists of albitized plagioclase and biotite microliths, recrystallized quartz, sericite, epidote, calcite, and chlorite (Fig. 4d).

Metadacite and metarhyolite dikes contain quartz, plagioclase and biotite phenocrysts in variable proportions included in fine-grained matrix composed of albite, quartz, sericite, and biotite (Figs. 4e, 4f). In the pit workings, the metarhyolite dikes are located in the southwestern part of the deposit. The metadacite dikes are more widespread, thinner and, are N-S to NE-SW striking with a steep dip (Fig. 2b). The metadacite and metarhyolite contain accessory minerals such as apatite, monazite, zircon, and rutile.

The lamprophyre dikes (sample T14;Fig. 6b) are up to 3 m wide. They are dark greyish color and are characterized by an aphanitic texture with small phenocrysts of biotite, amphibole and plagioclase. The matrix mostly consists of albite, calcite, ankerite, quartz, epidote, sericite and minor amounts of chlorite as well as tiny calcite veins, which indicates chloritization of the primary magmatic assemblage (Figs. 5a–5c). The dikes generally display a dominant NE-SW to E-W strike with a steep dip, and systematically intersect the metabasalt to metarhyolite dikes (Fig. 2b). These features indicate that they postdate the first generation dike and they were affected by metamorphism.

Dolerite dikes have an E-W strike and are steeply dipping. They were observed only in underground workings (Fig. 5). These dolerite dikes do not show any evidence of deformation, regional metamorphism, nor hydrothermal alteration. These dikes intersect all other lithologies in the area and clearly represent the last magmatic event in the Tabakoto deposit (Fig. 5d). It is possible that they correspond to the N090° dikes of the Sambarabougou swarm in the western part of the KKI dated at 1521 ± 3 Ma (Baratoux et al., 2019).

The geochemical analyses carried out on the magmatic rocks of the Tabakoto deposit are reported in Table 1. Metabasalts (samples T2 and T5) and lamprophyre (sample T14) show an alkaline affinity respectively in the SiO₂vs. Zr/TiO₂ diagram (Fig. 6a) of Winchester and Floyd (1977) and K₂O-SiO₂ diagram (Fig. 7b) of Raeisi et al. (2019), while meta-andesites, metadacites and metarhyolites show a calc-alkaline affinity in the AFM diagram (Fig. 7a) of Irvine and Baragar (1971).

Alkaline metabasalts show highly fractionated patterns (LaN/YbN = 24.9–16.1) marked by significant LREE (Light Rare Earth Elements) enrichment from 292 to 145 times the chondrites and clear HREE (Heavy Rare Earth Elements) depletion from 10 to 7 times the chondrites with positive (Eu/Eu* = 1.5) and negative (Eu/Eu* = 0.9) europium anomalies (Fig. 7c). The positive europium anomaly reflects the observed plagioclase accumulation. Lamprophyre also shows a highly fractionated pattern (LaN/YbN = 15.6) similar to those of alkaline metabasalts, with a pronounced negative europium anomaly (Eu/Eu* = 0.68). In the expanded diagram of trace elements normalized to the primitive mantle (Fig. 7d), all samples display a dispersion of LILEs (Cs, Rb and Ba), probably due to the various impacts of post-magmatic alteration, as well as positive anomalies in U and negative anomalies in Nb, Sr and Zr. The lamprophyre is an exception, showing a positive Sr anomaly.

With respect to chondrites, calc-alkaline rocks are highly enriched in LREE (229 to 105 times), and very depleted in HREE (9 to 1 times) with very steep patterns (La_N/Yb_N = 139.8–16.4) (Fig. 7c). Europium shows negative anomalies in meta-andesites (Eu/Eu* = 0.73–0.69) and metarhyolite (Eu/Eu* = 0.88), and positive anomalies in metadacites (Eu/Eu* = 1.35–1.01). The calc-alkaline rocks at Tabakoto also show a gradient of REE enrichment from metarhyolite through metadacite to meta-andesite, suggesting that these different rocks derive from the same parental magma that evolved by fractional crystallization. In the expanded diagram of trace elements normalized to the primitive mantle (Fig. 7d), calc-alkaline rocks show a dispersion of LILEs, positive U and Sr anomalies and very pronounced negative Nb anomalies.

The Tabakoto deposit displays a complex superposition of structures, characterized by a succession of three deformation phases. The structural position of the different vein populations of the deposit allowed us to establish the timing of their formation relative to the deformation events (Figs. 8–14). The nomenclature used in this paper refers to the local deformation of the Tabakoto deposit denominated D_Tn, where T stands for Tabakoto, and n is the phase of deformation identified at the deposit scale. Structures such as foliations (S), folds (F) and veins (V) are denoted by numbers (i.e., S_n, F_n and V_n). Note that we are agnostic at this stage of the relationship between the first phase recorded at the Tabakoto deposit (D_Tn) and the regional-scale deformation event D_n.

The metasedimentary rocks of the Tabakoto deposit display a composite foliation marked by the alternation of quartz-rich and mica-rich layers that corresponds to transposition of bedding S₀ into a S₁ schistosity during F₁ folding (Fig. 2b). This structure is also marked by the preferred orientation of micas and the deformation of quartz grains. The S₁ schistosity is steeply dipping (70° to 88°) generally towards the southeast but locally to the northwest (Figs. 8a and 9). Rootless F₁ isoclinal fold hinges are locally preserved (Fig. 9). This foliation is cross-cut by dikes of the first generation (Figs. 2b and 11c). The F₁ folding event, as well as the development of the S_0/1 foliation are classified as the D_T1 deformation event.

The S_0/1 and the first dike generation are affected by upright folds (F₂) with an NNE-SSW to NE-SW trending subvertical axial planar S₂ schistosity (Figs. 2b, 8–10) oriented NNE-SSW with an average dip of 78° (Fig. 8a). The S₂ schistosity is pronounced in non-competent rocks but refracts in competent rock units (i.e., in metagreywacke and dikes, Fig. 11c) or resolves as a spaced-fracture schistosity. High strain zones parallel to the S₂ schistosity marked by tight folding of the S₁ schistosity (Fig. 11a, c) are localized in meta-argilites. The first dike generation locally transposed into the S₂ schistosity. The F₂ folds are also identified at the regional scale and affect the Kofi metasedimentary series. The superposition of F₁ and F₂ folds results in hook-type interference features (Figs. 9, 11a). The F₂ folding event, as well as the development of the S₂ schistosity, record an E-W shortening accommodated by a vertical stretching and is classified as the D_T2 deformation event (Fig. 14).

Vertical NE-SW to E-W trending lamprophyre dikes cross-cut the S₂ schistosity and thus postdate D_T2 deformation. A set of mutually cross-cutting conjugate subvertical dextral NE-SW and sinistral NW-SE trending ductile-brittle faults affect the metasedimentary rocks and the first and second generations of dikes (Figs. 2b, 8b and 12). The NE-SW faults strike from N55° to N70° with dips ranging from 70° to 88° and the NW-SE faults strike from N115° to N158° with dips ranging from 55° to 87° (Figs. 8b and 12).

A set of minor E-W normal faults and/or fractures (<10 cm in length) with dips ranging from 73° to 88° (Fig. 8b), cross-cut the metasedimentary rocks. These structures are attributed to D_T3 deformation and mark a change from ductile-brittle to brittle structures, which is consistent with an E-W shortening direction and N-S stretching (Figs. 12 and 14).

Based on orientation, structural features and mineral paragenesis, we classify four type of veins at Tabakoto (Figs. 10–14): (i) barren folded milky quartz veins (V₁); (ii) gold-bearing sheared smoky quartz-pyrite veins (V₂); (iii) S-shaped fault-fill lenses of gold-bearing quartz − carbonate veins (V_3a), and (iv) barren fracture-fill quartz veinlets (V_3b).

The V₁ veins are characterized by coarse-grained milky quartz. They are localized in metasedimentary rocks, and do not show alteration halos (Fig. 10d). Quartz crystals show undulose extinction, and serrated grain boundaries, pointing to intracrystalline deformation and dynamic recrystallization (Fig. 15a). In zones of limited D_T2 deformation, V₁ veins cross-cut the S_0/1 foliation and display a distribution as conjugate sets of shallow-dipping tension gashes with a NE-SW and WNW-ESE direction (Fig. 10d). The V₁ veins commonly affected by F₂ folds, are boudinaged and transposed into the S₂ schistosity (Figs. 10b, 10d). These features indicate that they formed after the onset of D_T1 but before D_T2 (Figs. 10d and 14). They are interpreted to record an hydrothermal event that predated the ore stage.

The V₂ veins are generally parallel to the N-S to NNE-SSW trending S₂ schistosity (Figs. 2b and 11c) and occur in the metasediments, the first dike generation, and along their margins. In the metasediments, they are generally localized in zone of high strain, where they are defined as shear veins (Fig. 10b). These veins are thus interpreted as shear veins synchronous to D_T2 event.

The V₃ veins display various sizes and comprise: (i) S-shaped fault-fill lenses of gold-bearing quartz-carbonate veins (V_3a) and (ii) fracture-fill quartz veinlets (V_3b). The V_3a are hosted in NE-SW and NW-SE oriented D_T3 faults (Figs. 12a–12e) and commonly display quartz and albite crystals with a fibrous and/or crack-seal texture (Figs. 16a, 16b). The V_3a veins present quartz with undulose extinction and subgrains that attest for dynamic recrystallization (Fig. 16a). The V_3a veins result from a succession of brittle deformation events alternating with precipitation during the movement of the D_T3 faults (Figs. 12–14).

Millimetric to centimetric V_3b barren veinlets cross-cut the V₁ and V₂ veins. (Fig. 12f). V_3b veinlets are E-W trending and subvertical, which corresponds to the plane orientation of stress and are interpreted as tension gashes emplaced during D_T3 event (Figs. 13 and 12).

Two main ore trends are identified at the scale of the Tabakoto deposit. The first corresponds to gold included in sulfides disseminated in the first dike generation and in V₂ veins, hosted by the metasedimentary units and the first dike generation (Figs. 13, 17a, b, d, e). The ore trend cuts across structural features of the D_T1 deformation event. In this ore trend, grades range between 3 and 10 g/t Au.

The second ore trend is hosted in the V_3a veins localized in the NE-SW and NW-SE trending and steeply dipping D_T3 faults (Figs. 12 and 13). Average grades are estimated between 0.8 to 3 g/t. The intersection between the V₂ and V_3a veins displays grades up to 10 g/t.

V₂ veins of first gold mineralization are marked by an alteration halo that is more pervasive in the first dike generation than the metasediments. Mineralogically, V₂ veins are characterized by quartz, albite, carbonate (dolomite and ankerite), and a relatively high proportion of Fe-As sulfides. Metasomatic alteration consists of (i) albitization, manifested as the nearly complete replacement of plagioclase in the matrix by albite (Fig. 15b); (ii) K-feldspar alteration are developed in the dikes as a replacement of the albite minerals (Fig. 15c); (iii) carbonate alteration (dolomite and ankerite), variably developed (5–60%), and, like the potassic alteration, superimposed on albitization with distinct episodes of carbonation (Fig. 15d) also evidenced by multi-stage vein infilling (Fig. 4d).

The foremost sulfide minerals of the V₂ veins consists of pyrite, followed by arsenopyrite and pyrrhotite with lesser amounts of chalcopyrite, sphalerite and traces of galena and loellingite (Fig. 17). Pyrite displays various morphologies. It can form single crystals which commonly contain abundant inclusions of gangue minerals as well as other sulfides (i.e., chalcopyrite, galena, pyrrhotite), plus gold (Fig. 17a). Pyrite also forms aggregates where it is intermingled with euhedral to subhedral arsenopyrite, and less commonly pyrrhotite (Figs. 17a, 17b). Two main types of pyrite are identified namely (Tab. 2) (i) pyrite with no or very little As and (ii) pyrite with As content up to 1.7 wt%. Arsenopyrite forms either anhedral to subhedral inclusions in pyrite or occurs as euhedral crystals again associated with pyrite (Fig. 17b). Pyrrhotite mostly occurs as subhedral to anhedral inclusions in pyrite and arsenopyrite. Locally, euhedral arsenopyrite is surrounded by pyrrhotite (Fig. 17c). Chalcopyrite may be observed intergrown with pyrrhotite and as inclusions within pyrite and arsenopyrite. Galena is found in arsenopyrite and pyrite as fine-grained inclusions or filling fractures in arsenopyrite. Small grains of loellingite (As > 60 wt%; Tab. 2) occur locally as inclusions in pyrrhotite or in the matrix close to pyrrhotite (Fig. 17d). Some sphalerite, gersdorffite, pentlandite and scheelite were also found in the first dike generation and V₂ veins. Small grains of native bismuth and silver were detected within arsenopyrite (Fig. 17e). In the V₂ veins, gold grains are typically <15 µm and occur mostly as (i) inclusions in the sulfides (pyrite and arsenopyrite, more rarely pyrrhotite) and (ii) at the contact between two different sulfides (Figs. 17a, 17b, 17f). Microprobe analyses of these gold grains indicate Ag contents between 18 and 22 wt% (Tab. 3).

In the V_3a veins, in addition to quartz and carbonate, albite is common, and, with muscovite, occurs in strain shadows around sulfides (Figs. 16c, 16e). K-feldspar occurs along the edges or in fractured sulfides (Figs. 18b, 18c). Carbonates consist of dolomite and ankerite, with rare calcite and siderite (Figs. 16d). Phyllic alteration is broadly expressed around these V_3a veins and is represented by chlorite, muscovite and sericite. These alteration zones are pervasive, in which case sericite predominates. Chlorite is found in zones with intense hydrothermal alteration and in fractures of sulfides with muscovite (Figs. 18a, c, d). Chlorite is also developed in the strain shadows around pyrite and arsenopyrite, in which case it is associated with ankerite and/or albite (Fig. 16e). Muscovite is frequently found in fractured sulfides associated with gold (Fig. 18c). Muscovite is also developed as a rosette structure associated with chlorite and carbonate (Fig. 16f).

Pyrite and arsenopyrite are the most common sulfide minerals in the V_3a veins. They form massive anhedral crystals, which are commonly broken up and fractures are filled by albite, muscovite and sericite. Other sulfides such as pyrrhotite, galena, sphalerite and chalcopyrite complement the paragenesis. In the mineralization associated with the V_3a veins, gold occurs as (Fig. 18): (i) late infill or discrete gold veinlets filling cracks in pyrite and arsenopyrite in association with K-feldspar, chlorite and muscovite and (ii) as free native gold grains in the gangue closely associated with arsenopyrite and/or pyrite and chlorite alteration. It forms grains of size up to 30 µm and contains silver values between 11 and 13 wt% (Tab. 3). Locally, gold grains were also observed at the contact between the veins and the host rock.

New mapping of the north pit workings (Fig. 2b) and structural analysis of the whole Tabakoto deposit (TNP and TMP, Fig. 2a) reveals that it is dominated by a metasedimentary unit cross-cut by different generations of dikes with an alkaline to calc-alkaline signature (Figs. 7a, 7b). The pronounced negative Nb anomalies and the enrichment and fractionation in the REE spectra of the calc-alkaline rocks are consistent with magmas emplaced in subduction zones. Alkaline rocks with a negative Nb anomaly can be generated when a lithospheric mantle is stimulated by a hot spot. Synchronous emplacement of alkaline basalts and calc-alkaline andesites has been described in Baja, California, México (Benoit et al., 2002), where tearing of the slab opened an asthenopsheric window and lets through lower mantle melts with adakitic signatures.

The structural record of the metasedimentary units and dikes portrays a polyphased tectonic-metamorphic evolution characterized by deformation under ductile to brittle regimes (Figs. 14 and 19). Folding of the metasedimentary series and associated bedding-parallel foliation S₁ are consistent with the early phase of crustal thickening. In the western part of the KKI, the Dialé-Daléma metasedimentary series yield detrital zircon grains with U-Pb (LA-ICPMS) ages ranging from 2200 to 2100 Ma, which provide a maximum depositional age for the sediments (Hirdes and Davis, 2002; Koné et al., 2020). V₁ veins, emplaced and then deformed during crustal thickening D_T1 event, are poor in sulfides and barren of mineralization. At the regional scale, the D_T1 deformation event of the Tabakoto deposit correlates with the regional D₁ deformation event (Gueye et al., 2008; Diene et al., 2012; Masurel et al., 2017c).

The first dike generation is not cross-cut by V₁ veins. During subsequent D_T2 deformation, these dikes were affected by localized ductile to brittle shearing along their contacts (Fig. 10a). The first generation dikes are folded by F₂ but do not show F₁/F₂ interference patterns. These features are consistent with an emplacement of the first dike generation between D_T1 and D_T2. The N-S to NNE-SSW-trending subvertical S₂ penetrative schistosity in the axial plane of upright F₂ folds records thickening of the metasedimentary unit during E-W shortening attributed to D_T2 deformation (Figs. 8a and 14). The V₂ veins were emplaced within the metasediments, along the contacts between the metasediments and the first dike generation, and also cross-cut the dikes of first generation (Figs. 10, 11 and 19). These veins are interpreted to be the result of hydrothermal fluids circulating during the syn-D_T2 deformation. At the regional scale, the D_T2 deformation event of the Tabakoto deposit correlates with the regional D₂ transpresive tectonic event defined elsewhere in the West African Craton (e.g., in KKI, Milési et al., 1989; Ledru et al., 1991; Dabo and Aifa, 2010; Diene et al., 2012; Masurel et al., 2017a; in southern Mali, McFarlane et al., 2011; in other regions of the WAC, Pouclet et al., 2006; Hein, 2010).

Structures of D_T1 and D_T2 are cross-cut by NE-SW and NW-SE-striking faults and gold-bearing S-shaped quartz − carbonate veins (V_3a), which indicate emplacement of the latter during the D_T3 deformation event (Figs. 12, 14 and 19). These structures are consistent with E-W shortening and N-S stretching. Similar faults to those of the D_T3 are also described elsewhere in the KKI (e.g., Dabo et al., 2016; Masurel et al., 2017a; Diallo et al., 2024). Hence, D_T3 faults and fractures and associated veins (V_3a, V_3b) at Tabakoto may be correlated with regional D₃ and/or D₄ deformation events described in Senegal (Dabo et al., 2016; Diatta et al., 2017) and in Dialafara area in Mali (Diallo et al., 2024).

The Tabakoto deposit, like typical orogenic gold deposits in the WAC and elsewhere, is not clearly associated with a shear zone, as is the case for instance for the Sadiola and Loulo district deposits. Rather, the Tabakoto deposit is characterized by multiple overlapping stages of magmatic intrusions and hydrothermal alteration during distinct regional-scale deformation events. To better constrain the relationships between mineralization and magmatism, an estimate age of the absolute ages of the dikes in the Tabakoto deposit can be obtained from the literature. Directly to the west, in the Loulo district, and to the north at Sadiola, compositionally and texturally similar dikes to those of the first generation at Tabakoto provided ages within an interval between 2090 and 2060 Ma (Masurel et al., 2017a,b,c; Allibone et al., 2020; Lambert-Smith et al., 2020). This age span is also reported by Allibone et al. (2020) as the age of the discontinuous, small-displacement, sinistral shear zones that host the neighboring Yaléa and Gounkoto deposits. We thus suggest that the magmatic and possibly related hydrothermal events involved at Tabakoto probably occurred during the same age interval between 2090 and 2060 Ma. Thus, the Tabakoto deposit can be described as formed during the Eburnean orogeny marked by E-W shortening first associated with crustal thickening accommodated by ductile deformation (D_T2 event) followed by N-S horizontal stretching (D_T3 event) marked by ductile to brittle deformation.

The first mineralization phase at the Tabakoto deposit occurred as disseminated gold-bearing sulfides in the first dike generation and during progressive deformation related to the emplacement of an array of quartz-sulfide hydrothermal veins (V₂), preferentially located in the first dike generation (metabasalt, meta-andesite, metadacite and metarhyolite) and in the metasedimentary units (Fig. 19).

In addition to sulfides commonly present in gold deposits in the WAC, these veins contain reduced phases such as pyrrhotite and loellingite as well as bismuth and scheelite. Loellingite formation has been described to generally occur in two different settings (Barnicoat et al., 1991; Neumayr et al., 1993; Tomkins and Mavrogenes, 2001 and references therein) depending on the metamorphic degree: (i) during prograde metamorphism associated with the desulfurization of arsenopyrite with the formation of loellingite and pyrrhotite (Barnicoat et al., 1991); and (ii) during retrograde metamorphism associated with the sulfurization and destabilization of loellingite and pyrrhotite, leading to the formation of arsenopyrite and, locally, a new generation of pyrrhotite (Neumayr et al., 1993). The association of these minerals (po-lo-bi-sch) associated with gold is not common in typical orogenic settings. and in other gold deposits of the WAC, the presence of loellingite linked with gold mineralization has only been reported at the Morila deposit, in southern Mali, where mineralization was interpreted to be intrusion related (McFarlane et al., 2011).

At Tabakoto, loellingite is found as inclusions in pyrrhotite or in contact with it. Pyrrhotite also includes gold (Fig. 17d) and arsenopyrite at equilibrium (Figs. 18e, 18f). This assemblage seems to correspond to retrograde reactions (e.g., Neumayr et al., 1993) suggesting that, at Tabakoto, relatively hot conditions (upper greenschist facies) may have been attained locally during D_T2. The atomic-percent-of-As-in-arsenopyrite geothermometer of Sharp et al. (1985), applied to the stable arsenopyrite-pyrite-pyrrhotite assemblage of the V₂ alteration halo gives a temperature of ∼430–450 °C, which is consistent with the presence of biotite in the metasediments. Chlorite overgrowths on biotite records cooling following this metasomatic event. This cooling would also represent favorable condition to trigger the precipitation of As-rich pyrite and arsenopyrite as well as gold (e.g., Perfetti et al., 2008), which would have been easily trapped in these sulfides.

Another characteristic of the Tabakoto deposit is the presence of lamprophyre dikes, which have not yet been described in the KKI. Similar dikes are also reported in the Syama deposit, southern Mali (Traoré et al., 2016), a deposit mostly hosted in metabasalt and meta-andesite. However, the lamprophyres cross-cut the S₂ shistosity and related mineralized V₂ veins and V_3a veins. These would imply that they were emplaced after the major mineralization events, and are considered here as post- mineralization.

The second mineralization phase is related to the formation of the NE-SW and NW-SE trending faults that formed during the D_T3 deformation event. Field relations suggest that these faults acted as channel driving the fluid flow which formed a large number of quartz − carbonate veins (V_3a). This episode seems to have occurred at lower temperature conditions (greenschist facies), more typical of other orogenic deposits in the WAC, as suggested by brittle structures, common sulfides such as pyrite and arsenopyrite, and alteration dominated by albite, sericite and carbonate. In these veins, gold occurs mostly as fracture filling, suggesting that it precipitated at a later stage than the sulfides, possibly during successive reactivation of these veins, during progressive deformation.

Data reveal that gold mineralization of the Tabakoto deposit is associated with both an episode of dike intrusion and a later phase of hydrothermal event. The association between gold mineralisation and a range of calc-alkaline magmatism have been documented in the WAC at Morila deposit in Mali (McFarlane et al., 2011) and at Bonikro in Ivory Coast (Masurel et al., 2019), which are defined as intrusion-related gold system. However, at Tabakoto deposit, the bulk of the gold mineralization is associated with two sets of quartz veining developed in high strain zones (D_T2) and faults (D_T3) affecting both the dikes of different nature (metabasalt to metarhyolite) and metasedimentary units. In this case, the spatial association of gold mineralization with dikes at Tabakoto highlights the role of magmatic dikes as competent heterogeneities within metasedimentary units. The dikes acted as favorable sites for fluid flow due to competence contrast with the host rock and their brittle rheology. Accordingly, the brittle rheology of the dikes compared to the metasedimentary units is a key parameter that controlled the spatial distribution of the mineralization at Tabakoto. However, the fluids of the first gold mineralization may be part of magmatic origin because of the presence of minerals such as Au-Bi-Sch-Po-Ag-As, which are common in intrusion-related gold systems (e.g., ; Thompson and Newberry, 2000; Hart, 2007). Here, we can point to a possible source of fluide from a hidden intrusion at depth or in the vicinity of the Tabakoto deposit, which could be the Yatea pluton located to the north of the deposit at 4 km.

In the light of lithological and mineralogical particularities at Tabakoto, two scenario can be highlighted: (i) an intrusion-related gold system (Thompson and Newberry, 2000; Hart, 2007) with the contribution of fluid mobilized during the intrusive dike event or (ii) an intrusion-hosted orogenic gold system with an important auriferous quartz veins, which cut across the intrusive dikes and metasedimentary units where the mineralized veins are localized within shear zones and faults of the more competent units. At Tabakoto, we propose a combination of the two system. At first, the intrusions of dikes with low sulfides and low gold grade, which is portray by disseminated gold-bearing sulfides. This was followed by the bulk gold values associated with the new gold input marked by the late hydrothermal fluids and/or remobilization of gold in dikes due to the late brittle deformation. This process is also documented in other African deposits (e.g., Pampe in Ghana, Kalana in Mali, Edikan in Ghana, Bonikro in Ivory Coast, Moboma in Central African Republic, Brothers project in south amrica; Salvi et al., 2016a, b; Tourigny et al., 2018; Masurel et al., 2019; Kpeou et al., 2020; Combes et al., 2024).

Based on detailed mapping of the Tabakoto deposit, analysis of the litho-structural data as well as our interpretation, we can highlight the following conclusion:

• (i)

  The bulk of the gold mineralization is associated to hydrothermal fluid circulation affecting both the dikes of different nature and the metasediments under ductile to brittle conditions.

• (ii)

  The litho-structural architecture at Tabakoto is consistent with an intrusion-hosted orogenic gold system, where the dikes acted as favorable sites for fluid flow.

• (iii)

  A possibly early magmatic fluid from hidden intrusion or in the vicinity of the deposit portray by the presence of Au-Bi-Sch-Po-Ag-As minerals.

• (iv)

  The D_T2 deformation event at Tabakoto can be correlated with the D₂ deformation phase at the regional scale, as defined by Milési et al. (1989), Gueye et al. (2008), Masurel et al. (2017c), and Allibone et al. (2020).

• (v)

  The D_T3 phase at Tabakoto can be correlated with the regional Eburnean D₃ and/or D₄ deformation widely observed in neighbouring Senegal (Dabo et al., 2016; Diatta et al., 2017) and in Mali (Diallo et al., 2024).

• (vi)

  The Tabakoto deposit is an example of structural-magmatic-hydrothermal gold deposition during a continuous deformation with E-W shortening from vertical stretching during D_T2 to horizontal stretching during D_T3 from the climax of the Eburnean orogeny marked by ductile deformation, to subsequent lateral flow of the orogenic crust under ductile to brittle conditions.

This study was achieved within the framework of the PhD thesis of the first author, funded by the French Ministry of Research through SCAC/Mali and by the West African Exploration Initiative (WAXI) project. AMIRA International, the industry sponsors, and sponsors in kind are gratefully acknowledged for their support of the WAXI-3 project (P934B). Endeavour Mining Mali is gratefully acknowledged through Julien Baumard (Exploration Manager) and John Barry (Technical Services Manager) who are no longer there. Special thanks to Moussa Kéita (currently at Syama), Hassan Said, and Oliver Benford for their help at Tabakoto.

The authors acknowledge the Raimond Castaing micro-characterization center and the scanning electron microscope (SEM) center in Toulouse (University Toulouse III-Paul Sabatier). Philippe and Fabienne de Parseval as well Thierry Aigouy and Sophie Gouy are thanked for their excellent scientific assistance during the different analyses.