Polymetallic veins and breccias and carbonate-replacement ore deposits in the Cyclades continental back arc, Greece, formed from a range of fluid and metal sources strongly influenced by the dynamics of the late Mesozoic-Cenozoic Hellenic subduction system. These complexities are recorded in the isotopic signatures of hydrothermal barite. We investigated 17 mineral occurrences on four Cycladic islands and from Lavrion on the mainland. Here, barite occurs in almost all deposit types of Miocene to Quaternary age. We used a multiple isotope and geochemical approach to characterize the barite in each deposit, including mineral separate analysis of δ34S and δ18O and laser ablation-inductively coupled plasma-mass spectrometry of 87Sr/86Sr and δ34S. Barite from carbonate-hosted vein and breccia Pb-Zn-Ag mineralization on Lavrion has a wide range of δ34S (2–20) and δ18O (10–15) values, reflecting a mix of magmatic and surface-derived fluids that have exchanged with isotopically heavy oxygen in the carbonate host rock. Sulfur (δ34S = 10–13) and oxygen (δ18O = 9–13) values of barite from the carbonate-hosted vein iron and barite mineralization on Serifos are permissive of a magmatic sulfate component. Barite from epithermal base and/or precious metal deposits on Milos has δ34S (17–28) and δ18O (9–11) values that are similar to modern seawater. In contrast, barite from vein-type deposits on Antiparos and Mykonos has a wide range of δ34S (16–37) and δ18O (4–12) values, indicating a seawater sulfate source modified by mixing or equilibration of the hydrothermal fluids with the host rocks. Strontium isotope ratios of barite vary regionally, with 87Sr/86Sr ≥ 0.711 in the central Cyclades and 87Sr/86Sr ≤ 0.711 in the west Cyclades, confirming the strong influence of upper crustal rocks on the sources of fluids, Sr, and Ba in the formation of ore.

The metamorphic core complexes of southern Greece host a wide range of base and precious metal deposit types throughout the Cycladic Islands from Attica, on the mainland, to the active volcanic arc in the south. Polymetallic mineralization in the Cycladic metallogenic belt (hereafter Cyclades mineral district) is strongly influenced by the dynamics of the subduction zone and regional variations in the underlying basement (Menant et al., 2018; Wind et al., 2020). Major detachment faults have been active in the region since at least the early Miocene, accommodating some of the most rapid extension of continental crust in the world (Pe-Piper and Piper, 2002), and created widespread fluid pathways in the brittle upper crust. The Cyclades mineral district includes several mineralization styles, such as low-grade porphyry Mo, carbonate-hosted replacement, vein and breccia Pb-Zn-Ag-Au, low- and intermediate-sulfidation (LS and IS) epithermal veins, Cu-Fe and Fe skarn, Fe- and Zn-rich gossan, epithermal-like Ag-Pb-Zn-Ba(-Au) ore, iron and barite veins, sediment-hosted Mn oxide ores, and polymetallic sea-floor massive sulfide deposits (Skarpelis, 2002; Kilias et al., 2013; Melfos and Voudouris, 2017; Menant et al., 2018; Voudouris et al., 2019; Wind et al., 2020, and references therein). In addition, well-known non-metallic deposits of bentonite, perlite, and kaolinite occur on Milos (Plimer, 2000).

Low-angle detachments formed at the ductile-brittle transition, where deep fluids (magmatic, metamorphic) have extensively mixed with surface-derived seawater and meteoric water (e.g., Marchev et al., 2005; Moritz et al., 2006; Menant et al., 2018; Scheffer et al., 2019). Polymetallic veins and replacement deposits occur in the metamorphosed footwall and unmetamorphosed hanging wall of the major detachment systems as well as in overlying Quaternary volcanic units (Berger et al., 2013; Menant et al., 2013, 2018; Ducoux et al., 2017; Wind et al., 2020). However, the relative importance of magmatic, meteoric, and seawater-derived fluids and their complex interplay is widely debated (e.g., Naden et al., 2005; Tombros et al., 2007; Bonsall et al., 2011; Alfieris et al., 2013; Spry et al., 2014; Kevrekidis et al., 2015; Scheffer et al., 2017). This is partly due to the wide range of mineral and isotopic systems and various paragenetic stages that have been studied in different deposits. In this paper, we examine the possible sources of fluids and their interaction with host rocks using hydrothermal barite, which occurs in the majority of deposits (Hauck, 1984; Salemink, 1985; Skarpelis, 2002; Bonsall et al., 2011; Tombros et al., 2015; Schaarschmidt et al., 2021). This approach allows a comparison of the same or similar mineralizing processes across the region because of the very restricted paragenesis and conditions of barite mineralization in each of the deposits.

Mining in the Cyclades mineral district probably started as early as the eighth to fourth millennia B.C. (Gale and Stos-Gale, 1981), but it remains a promising region for exploration of precious, rare, and critical metals (Melfos and Voudouris, 2017). The carbonate-replacement-, vein-, and breccia-style Pb-Zn-Ag ore deposits (Lavrion district) were mined in antiquity until the late 20th century and played a major role in the economic development of the region. The most important deposits in the Lavrion district are thought to have produced 13 million tonnes (Mt) of ore at 400 g/t Ag and 20% Pb, with resources of 4 Mt at 7% Pb-Zn remaining (Melfos and Voudouris, 2017). The Fe skarns and Fe-Ba vein mineralization on Serifos produced ~6.6 Mt of Fe ore (Salemink, 1985). On Antiparos, argentiferous galena was mined from polymetallic quartz-sulfide veins at Agios Georgios. Barite was exploited on Milos and Mykonos islands until the mid-1980s. Mining activity on the island of Mykonos involved barite exploitation and processing until 1984, and older small-scale occasional mining for argentiferous galena, cerussite, anglesite, and Fe hydroxides (Skarpelis, 2002). The barite vein ores on Mykonos produced ~2.3 Mt of barite with reserves of ~1 Mt (Tombros et al., 2015). The sulfide-sulfate deposits of Milos contained ~2 Mt of barite (Hauck, 1984). Barite-quartz veins on Milos are widely associated with epithermal systems, including the Au-Ag-Te mineralization at Profitis Ilias (current reserves of 5 Mt at 4.4 g/t Au, 43 g/t Ag) and Chondro Vouno (3.3 Mt at 4.2 g/t Au: Kilias et al., 2001; Alfieris et al., 2013). Breccia-style Ag-Pb-Zn-Ba(-Au) ores were mined at Triades-Galana on Milos from the late 19th to the early 20th century, firstly for Pb and Zn and later for Ag, when reserves were estimated at 10 Mt ore at 500 g/t Ag (Liakopoulos et al., 2001). Reported resources at Triades-Galana are 1.2 Mt at 1 g/t Au and 124 g/t Ag (Stewart and McPhie, 2003). The Cape Vani deposit on Milos operated as an Mn mine from 1886 to 1909 and from 1916 to 1928 producing 0.22 Mt of Mn (Plimer, 2000).

The deposits of the Cyclades mineral district have been extensively studied to determine the origin of metals and ore-forming fluids, with variable contributions inferred from magmatic, meteoric, and modified seawater sources (Kilias et al., 2001; Naden et al., 2005; Bonsall et al., 2011; Alfieris et al., 2013; Kevrekidis et al., 2015; Tombros et al., 2015; Scheffer et al., 2017, 2019; Wind et al., 2020). A wide range of investigations, mainly involving isotopic studies of sulfides, whole-rock geochemistry, and fluid inclusion microthermometry, have been carried out (Table 1). Part of the challenge of determining the dominant fluid sources is the complex paragenesis of the studied minerals and the lack of a common mineral or isotopic system to compare between the different deposits across the region. We previously demonstrated the use of Pb isotopes in identifying metal sources, taking advantage of the nearly ubiquitous presence of galena in all deposits, and showed the regional differences in the underlying basement in the north-central and western Cyclades (Wind et al., 2020). Herein, we explore the geochemical and isotopic signatures of barite, which is also present in nearly all of the deposits.

The properties of barite, especially its low solubility and the possibility of analyzing multiple isotope systems, make it an excellent geochemical archive for ore-forming processes. It is also an important proxy for different mineralizing systems involving diverse fluid types almost always involving mixing of reduced and oxidized fluids (Hanor, 2000; Griffith et al., 2018). In this study, we used a multi-isotope approach to investigate the composition of Cycladic barite from 17 mineral occurrences on four islands and from Lavrion on the mainland, spanning ~40,000 km2 of the mineral district and arc-back-arc system. We also demonstrate the value of combining isotopic studies of barite with analyses of trace elements, especially the Sr content and spatially resolved analyses of mineral zonation, as also shown in studies from other mineral systems (e.g., Jamieson et al., 2016; Magnall et al., 2016). In particular, we combined analyses for δ34S and δ18O by isotope ratio mass spectrometry (IRMS) with analyses of 87Sr/86Sr ratios and δ34S values by laser ablation-multicollector-inductively coupled plasma-mass spectrometry (LA-MC-ICP-MS) to trace the fluid and crustal sources in a range of mineralization styles. The combination of isotopic systems indicates that different fluids were tapped along the major detachment faults where highly variable basement rock types are exposed.

Cycladic barite can be grouped into two end-member systems: (1) seawater-dominated systems, with high δ34S values, low δ18O values, and radiogenic Sr with variations in the isotopic signatures and Sr content likely influenced by the underlying basement, and (2) magmatic-dominated systems, with low δ34S values, high δ18O values, and Sr derived from the Miocene intrusions. Figure 1 and Table 1 present an overview of the investigated mineral occurrences and their characteristics, and our interpretation is summarized in Figure 2 and in the following sections. We show that different mineral deposits in the region can be clearly distinguished by the isotopic signatures of their barite, providing a useful guide in future exploration for these types of deposits in Greece and in other continental arc-back-arc systems.

The extended continental back-arc domain in the Aegean formed by the northward subduction of the African plate underneath Eurasia, where subduction and accretion of successive oceanic and continental fragments occurred since the Mesozoic (Jolivet and Brun, 2010; Papanikolaou, 2013; Menant et al., 2016, and references therein). Accelerated slab retreat and rapid southward migration of the Hellenic trench since the Eocene-Oligocene caused extension and thinning of the back-arc domain, manifested in the southward migration of arc magmatism (Fytikas et al., 1984; Pe-Piper and Piper, 2002). Miocene exhumation of middle and lower crustal material as metamorphic core complexes (Gautier and Brun, 1994; Jolivet et al., 2004, 2015; Ring et al., 2010) occurred along four major low-angle, crustal-scale detachment systems (Fig. 1): the North Cycladic detachment system (top-to-NE: Jolivet et al., 2010), the Naxos-Paros detachment system (top-to-N: Gautier et al., 1993), the West Cycladic detachment system (top-to-SSW: Grasemann et al., 2012), and the Santorini detachment system (top-to-SSE: Schneider et al., 2018). Three major tectonostratigraphic units have been identified: the undifferentiated and (mostly) unmetamorphosed Pelagonian zone in the hanging wall of the major detachment systems, and the metamorphosed Cycladic basement and the Cycladic blueschist unit in the footwall (Fig. 1; Bonneau, 1984; Jolivet et al., 2015; Grasemann et al., 2018; Flansburg et al., 2019).

As the structurally lowest unit, the Cycladic basement is exposed in the central part of the Cyclades (e.g., Antiparos, Delos, Ios, Naxos, Mykonos, Paros, Sikinos, Serifos, and Syros) and consists of Carboniferous orthogneisses, as well as paragneisses and mica-schists with a pre-Alpine history and peri-Gondwanan affinity (Henjes-Kunst and Kreuzer, 1982; Keay and Lister, 2002; Jolivet et al., 2010; McGrath et al., 2017; Flansburg et al., 2019). The nature of the contact between the Cycladic basement and overlying Cycladic blueschist unit is debated (local detachment planes, thrust faults, or reactivated detachment faults: e.g., Huet et al., 2009; Schneider et al., 2011; Grasemann et al., 2012; Augier et al., 2015; Ducoux et al., 2017), and recently a parautochthonous origin has been suggested (Poulaki et al., 2019). Regionally, carbonate platform sediments, which are variably described in the literature as Lower, Basal, or Gavrovo-Tripolitza units (Chatzaras et al., 2011; Schneider et al., 2011; Grasemann et al., 2018; Gerogiannis et al., 2019), occur stratigraphically below the Cycladic blueschist unit on the mainland in Lavrion as well as on Evia, Andros, Serifos, and Amorgos.

The Cycladic blueschist unit, exposed on the majority of the Cycladic islands, is a metamorphosed Triassic to Cretaceous volcano-sedimentary sequence with clastic metasedimentary rocks, marbles, and metabasites (Gärtner et al., 2011; Seman et al., 2017; Poulaki et al., 2019). Paleogeographically, the Cycladic blueschist unit has been considered to be derived from the subducted Pindos oceanic domain (Bonneau and Kienast, 1982; Pe-Piper, 1998; Jolivet and Brun, 2010; Fu et al., 2015). It has been divided into an Upper Cycladic blueschist nappe and a Lower Cycladic blueschist nappe, based on their peak metamorphic conditions, which are separated by the Trans-Cycladic thrust (Grasemann et al., 2018). Differences between the Upper and Lower Cycladic blueschist nappes also have been identified by the distinct Pb isotope compositions of galena in the west and north-central Cyclades mineral deposits (Wind et al., 2020). Deposits in the west Cyclades highlighted in this study include Villia, Plaka, Cape Kaparia, Mega Livadi, Cape Vani, Kalogries, Katsimouti, Galana, Triades, Profitis Ilias, Fourkovounia, and Kastana, which are located on islands along the West Cycladic detachment system (Serifos, Milos), and Lavrion on the mainland. These islands together with Kea, Kythnos, Folgeandros, Santorini, and Anafi include the main exposures of the Lower Cycladic blueschist nappe. The deposits in the central Cyclades highlighted in this study include Cape Charos, Cape Evros, Panormos Bay, Agios Georgios, and Prassovounia, which are located on the islands along the North Cycladic and Naxos-Paros detachment systems (Mykonos, Antiparos). These islands together with Evia, Andros, Tinos, Syros, Sifnos, Paros, Naxos, Sikinos, and Ios include the main exposures of the Upper Cycladic blueschist nappe. The investigated deposits in the central Cyclades, in the center of the metamorphic core complex, are proximal to exposures of the Cycladic basement (on Ios, Antiparos, Paros, Naxos, and Mykonos).

The Pelagonian zone in the hanging wall of the major detachment systems consists of klippen of Paleozoic basement and late Permian to Jurassic volcaniclastic rocks with ophiolite sequences and carbonates (e.g., Bonneau, 1984; Schenker et al., 2014). Rock types of the Pelagonian zone have a complex tectonomagmatic history and are thought to comprise mostly remnants of the Vardar ocean and continental Pelagonian terrane (Schenker et al., 2014; Lamont et al., 2020).

The uppermost unit in the Cyclades consists of Miocene to Pliocene marine to continental sediments deposited in the supradetachment basins that are preserved on Mykonos, Paros, and Naxos, chiefly in the hanging wall of the local detachments (e.g., Sánchez-Gómez et al., 2002; Bargnesi et al., 2013). Marine to continental sediments on Attica and Milos unconformably overlie the Cycladic blueschist unit (Stewart and McPhie, 2006; Krohe et al., 2010).

In the Cyclades, the southward-migrating volcanic arc was active since the early Miocene as a suite of syntectonic intrusions, the Cycladic granitoids, emplaced within the deformed upper crust (Fig. 1; Altherr et al., 1982; Altherr and Siebel, 2002; Iglseder et al., 2009; Bolhar et al., 2010; Schneider et al., 2018). Emplacement of Cycladic granitoids in the northeast started at ca. 17 Ma (e.g., at Ikaria and Tinos: Altherr et al., 1982; Bolhar et al., 2010), whereas in the northwest it started only after 10 Ma (e.g., at Lavrion and Serifos: Iglseder et al., 2009; Liati et al., 2009). Metaluminous to slightly peraluminous I-type intrusive bodies show a systematic regional variation in their composition with increasing distance from the trench (Altherr et al., 1982; Schliestedt et al., 1987; Altherr and Siebel, 2002). In addition to the I-type intrusions, smaller S-type intrusive bodies occur on Serifos, Tinos, Paros, Naxos, and Ikaria (Altherr et al., 1982; Altherr and Siebel, 2002, and references therein), which generally predate the I-type intrusions (Altherr et al., 1982; Bolhar et al., 2010). Mafic to felsic dikes of Miocene age occur in Lavrion and on Serifos (Altherr et al., 1982; Skarpelis et al., 2008). At the Miocene-Pliocene boundary, bimodal volcanism occurred in the central Cyclades (e.g., rhyolites in the south of Antiparos; Innocenti et al., 1982). The initiation of bimodal magmatism is thought to be related to thinning and melting of the continental crust above a slab tear (Pe-Piper and Piper, 2007). The geochemistry and isotopic compositions of the Miocene Cycladic granitoids are typical for crustal-derived magmas. Strontium and Nd isotopes indicate a mix of a radiogenic metapelitic graywacke-type paragneiss and a less radiogenic mafic component as melt sources (Stouraiti et al., 2010; McGrath et al., 2017). In contrast, Miocene to Quaternary volcanism of the volcanic arc is indicative of mantle-derived magmas with variable degrees of metasomatism and crustal contamination (Pe-Piper, 1994; Ersoy and Palmer, 2013; Francalanci and Zellmer, 2019). Since the Pliocene, the modern Aegean volcanic arc lies from Sousaki in the northwest, across Milos and Santorini toward Nisyros in the southeast (Innocenti et al., 1981; Francalanci et al., 2005; Francalanci and Zellmer, 2019).

Regional metallogeny

High heat flow and a permeable brittle upper crust promoted a wide range of magmatic-hydrothermal activity throughout the Cyclades. Some workers have linked the mineralization in the Aegean to regional transtension or extension where crustal-scale detachments and associated high-angle normal faults facilitated the ascent of metal-bearing fluids (Berger et al., 2013; Menant et al., 2013, 2018; Ducoux et al., 2017). Both seawater and meteoric water, as well as direct magmatic contributions, have been implicated in the formation of the majority of precious and base metal-rich deposits in the Cyclades (Table 1).

In the Lavrion district, mineralization occurs principally in marbles of the Lower Cycladic blueschist nappe (Fig. 1) and Upper and Lower marbles of the Basal unit that define the footwall of the local Lavrion detachment, which is part of the West Cycladic detachment system (Fig. 1). This district comprises porphyry Mo style, Cu-Fe skarn, stratabound high-temperature carbonate-replacement (mantos and chimneys), vein- and breccia-style Pb-Zn-Ag mineralization (i.e., high-grade Ag in base metal sulfides that are cemented by fluorite and carbonate gangue), and Fe- and Zn-rich gossan (Voudouris et al., 2008a, b, 2019; Skarpelis and Argyraki, 2009; Bonsall et al., 2011; Berger et al., 2013; Spry et al., 2014; Scheffer et al., 2017, 2019). The carbonate-replacement Pb-Zn-Ag mineralization is temporally and spatially associated with the emplacement of the Plaka granodiorite at 8.3 Ma (Altherr et al., 1982) and several mafic and felsic dikes at 9.4 Ma (Skarpelis et al., 2008). Multiple stages of mineralization have been identified throughout, including a late-stage remobilization of the Pb-Ag-Zn deposits resulting in the formation of supergene iron ore (Skarpelis and Argyraki, 2009; Bonsall et al., 2011; Scheffer et al., 2017). Porphyry and sulfide breccia porphyry-style mineralization, as well as skarn mineralization, occur within or adjacent to the Plaka granodiorite (Leleu et al., 1973; Voudouris et al., 2008b), whereas carbonate-replacement deposits and vein- and breccia-style mineralization are confined to a zone structurally below the Lavrion detachment (Berger et al., 2013; Scheffer et al., 2017, 2019). A magmatic origin for the fluids that formed the porphyry, skarn, and high-temperature carbonate-replacement deposits has been well established (e.g., Voudouris et al., 2008a, b; Bonsall et al., 2011; Spry et al., 2014). However, Scheffer et al. (2019) argued that the polymetallic vein- and breccia-style mineralization was related to mixing of meteoric water with evaporated seawater and concluded that two different and independent hydrothermal systems caused ore deposition in the Lavrion district. Barite, although not abundant at Lavrion, occurs in several deposits as open-space filling (Bonsall et al., 2011; Scheffer et al., 2019; Wind et al., 2020). Bonsall et al. (2011) and Scheffer et al. (2019) suggested that barite formation in the Lavrion district occurred late in the hypogene paragenesis during veining, brecciation, and carbonate-replacement Pb-Zn-Ag mineralization, postdating the higher-temperature porphyry-style and skarn- and carbonate-replacement-forming events. Based on the S isotope composition of carbonate-replacement-associated barite, Bonsall et al. (2011) suggested a Miocene seawater source of sulfate.

On Serifos island, numerous magnetite-rich exo- and endoskarns and iron and barite (Fe-Ba) orebodies occur predominantly in Cycladic basement marbles in the footwall of the Meghalo Livadi detachment (Salemink, 1985; Ducoux et al., 2017). The formation of the Serifos skarn was associated with the intrusion of the 11.6 to 9.5 Ma syntectonic Serifos granodioritic pluton and extensional deformation along the West Cycladic detachment system (Fig. 1; Salemink, 1985; Iglseder et al., 2009; Ducoux et al., 2017). Most of the mined Fe came from the hydrothermal Fe-Ba mineralization, which occurs in late, banded veins consisting of hematite-limonite and barite, with minor fluorite and calcite, and hematite-barite breccias. These mined Fe-Ba veins abound in the western part of the island chiefly within calcitic marbles; however, they cut across all rock types of the island above and below the Meghalo Livadi detachment and clearly postdate the skarns (Salemink, 1985; Ducoux et al., 2017). Based on fluid inclusion studies, Salemink (1985) estimated that deposition of the Fe-Ba veins occurred at temperatures of 100° to 210°C. Ducoux et al. (2017) suggested that these veins were deposited late in the evolution of the detachment system by surface-derived (meteoric) fluids causing remobilization of the Fe from the Fe-rich skarns. Distal skarn and polymetallic (including native Bi) marble-replacement mineralization occurs in small lenses in the Moutoulas prospect in the north; polymetallic sulfide-bearing quartz veins (up to 100 m long and 0.5 m wide) also crosscut the schists and marbles of the Moutoulas prospect (Salemink, 1985; St. Seymour et al., 2009; Fitros et al., 2017; Wind et al., 2020). Fitros et al. (2017) suggested that native Bi mineralization at Moutoulas was a product of Bi-bearing magmatic fluids interacting with the wall rock and mixing with meteoric water.

On Milos, a recently emergent volcanic island (<2 Ma), economic mineralization occurs in the western part of the island in the Quaternary calc-alkaline submarine and submarine to subaerial volcanic rocks. Mineralization includes sediment-hosted exhalative Mn, Fe, and barite (Cape Vani), epithermal Pb-Zn-Ag veins (Kondaros-Katsimouti), epithermal-like barite-quartz breccia Zn-Pb-(Ag-Au) sulfide (Triades-Galana), and epithermal Au-Ag-Te vein-type (Profitis Ilias-Chondro Vouno) deposits. The deposits in this densely mineralized area have been investigated extensively (Hauck, 1984, 1988; Hein et al., 2000; Kilias et al., 2001, 2020; Liakopoulos et al., 2001; Stewart and McPhie, 2003; Naden et al., 2005; Marschik et al., 2010; Alfieris et al., 2013; Papavassiliou et al., 2017; Chi Fru et al., 2018; Smith et al., 2018; Miles, 2021; Schaarschmidt et al., 2021). Manganese oxide and barite deposits at the abandoned Cape Vani Mn mine, and the area around it, formed in a shallow-water paleohydrothermal system predominantly from modified seawater and leaching of the metamorphic and volcanic host rocks (Hein et al., 2000; Liakopoulos et al., 2001). Ivarsson et al. (2019) confirmed barite formation from seawater sulfate that was isotopically modified through bacterial sulfate reduction (BSR). Kilias (2012) and Kilias et al. (2020) further suggested that biologically mediated Mn2+ oxidation played a significant role in Mn deposition. The polymetallic epithermal-type vein mineralization at Kondaros-Katsimouti crosscuts the dacitic lavas and volcaniclastic rocks and is thought to be related to a buried granite at depth (Papavasiliou et al., 2016), although evidence of this model is not obvious. Schaarschmidt et al. (2021) proposed that the NNW-SSE–trending Kondaros-Vani fault zone represents a vertical profile through the boiling zone of a shallow submarine hydrothermal system. Mineralization at Kondaros-Katsimouti as well as Triades and Galana is spatially related to the NE-SW–trending Kontaro fault. Mineralization at Triades-Galana has been identified as either Kuroko-type volcanogenic massive sulfide (VMS) (Hauck, 1984, 1988; Vavelidis and Melfos, 1998) or seawater-dominated, intermediate- to high-sulfidation epithermal mineralization (Alfieris et al., 2013; Marschik et al., 2010). Smith et al. (2018) present a hybrid model with an epithermal-like vein system superimposed on a shallow-sea-floor sulfide-sulfate deposit whose development was limited by mass wasting events. Barite, as well as sulfides, at Triades is thought to be a result of rapid precipitation at the hydrothermal-seawater interface by boiling hydrothermal fluids (Christanis and St. Seymour, 1995; Smith et al., 2018; Schaarschmidt et al., 2021). At Triades and Galana, barite formed continuously over multiple stages including pre-, syn-, and post-ore (Smith et al., 2018). The vein mineralization at Profitis Ilias-Chondro Vouno is also interpreted to be the product of a hybrid VMS-continental epithermal system with mixing of meteoric water, seawater, and volcanic gases as volcanism changed from submarine to subaerial (Naden et al., 2005; Kilias et al., 2001). Post-ore-stage barite occurs in the veins together with quartz and rarely fine pyrite, sphalerite, chalcopyrite, and galena (Kilias et al., 2001). Epithermal Au mineralization at Profitis Ilias is concentrated above the base metal zone and spatially related to boiling (Kilias et al., 2001). Mining of barite during the 20th century was concentrated in the east of Milos in the area around Kastana, and the different occurrences have been described in detail by Hauck (1984). In addition, small occurrences of barite mineralization can be found at Fourkovouni in the central north of the island. The widespread barite occurrences on Milos have been studied by Hauck (1984, 1988), Hein et al. (2000), Naden et al. (2005), Marschik et al. (2010), Ivarsson et al. (2019), and Schaarschmidt et al. (2021). These authors all suggested a seawater sulfate source with Sr derived from the country rocks. In the western part of the island, the barite occurs in association with abundant sulfides, but in the east of the island the barite veins are mainly barren of sulfides.

On Mykonos, the mined barite mineralization occurs in subvertical NW-SE–trending, extensional, polymetallic banded veins that are concentrated in the northeastern part of the island (Cape Evros and Cape Charos). The veins cut the upper part of the Mykonos monzogranitic laccolith, the metabasites of the Pelagonian zone, and the Miocene sediments and are associated with the Livada and Mykonos detachments that belong to the North Cycladic detachment system (Fig. 1; Lahti and Govett, 1981; Skarpelis, 2002; Lecomte et al., 2010; Menant et al., 2013). The mined veins are >3 m wide, can be traced for 5 km, and consist predominantly of barite, with Fe hydroxide, argentiferous galena, pyrite, sphalerite, chalcopyrite, and quartz (Skarpelis, 2002; Menant et al., 2013; Tombros et al., 2015; Wind et al., 2020). At Panormos Bay, in north Mykonos, the banded veins cut the exposed marine to terrestrial Miocene conglomeratic sequence (Sánchez-Gómez et al., 2002; Menant et al., 2013; Wind et al., 2020). Menant et al. (2013) have shown that the barite vein mineralization was formed from 11 to 10 Ma when the Mykonos monzogranite was exhumed and crossed the ductile to brittle transition below the North Cycladic detachment system. Tombros et al. (2015) proposed that magmatic Ba-rich fluids derived through hydrothermal alteration of the Mykonos monzogranite mixed with sulfate-rich Miocene seawater to produce the barite and sulfide veins. Brittle deformation, including fault zones and cataclasites adjacent to the Mykonos detachment, is thought to have allowed incursion of sulfate-rich Miocene seawater into the pluton. Subeconomic Au-bearing (1 g/t) mineralized siliceous caps and silicified breccias occur within the monzogranitic cupola along the Mykonos detachment at Cape Evros, Profitis Ilias, and Mavro Vouno in the northeasternmost part of the island (Skarpelis, 2002; Skarpelis and Gilg, 2006). The breccias are cemented by hydrothermal minerals, including epithermal-style quartz and barite, and relicts of primary sulfides (pyrite, chalcopyrite, sphalerite, galena) now oxidized to hematite and goethite. The breccias are cut by N-trending normal faults that are filled with barite ± fluorite and marcasite (Tombros et al., 2015). Skarpelis (2002) suggested that the siliceous cap and silicified breccia are remnants of a deeply eroded precious metal epithermal system, and Skarpelis and Gilg (2006) suggested a two-stage hydrothermal event, with epithermal-style Au quartz deposition followed by brittle faulting and baritesulfide deposition from basinal brines.

On Antiparos, subvertical, epithermal-style crustiform quartz veins, up to 80 cm wide, occur in the gneiss and schist of the Cycladic basement and marble of the Upper Cycladic blueschist nappe at Agios Georgios in the south of the island and in the western part near Prassovounia (Gale and Stos-Gale, 1981; Skarpelis, 2002; Kevrekidis et al., 2015; Wind et al., 2020). Barite has been described in crustiform-banded quartz veins with smithsonite and fluorite at Agios Georgios (Kevrekidis et al., 2015) and with galena and Fe-poor sphalerite in the Prassovounia area (Wind et al., 2020). Argentiferous galena (containing 800–2,000 ppm Ag: Kevrekidis et al., 2015), sphalerite, pyrite, and chalcopyrite occur in the veins. Kevrekidis et al. (2015) proposed that the mineralization is related to magmatic fluids from the leucogranitic Paros pluton that mixed with meteoric water. However, Voudouris et al. (2019) and Wind et al. (2020) suggested that a connection to the Pliocene rhyolites (5–4 Ma: Innocenti et al., 1982) exposed in the south of Antiparos is more likely.

We sampled 17 mineral occurrences from Miocene to early Pleistocene age throughout the Cyclades mineral district. Despite the wide range of mineralization styles, barite occurs as an abundant mineral in all of the deposits. On each island several prospects were sampled (App. Table A1). In the western Cyclades, samples were taken from the Lavrion district on the mainland and on Serifos and Milos islands. In Lavrion, barite samples were collected underground near Plaka (Fig. 3A), and one sample was taken from the Villia adit in the northern part of the district, where barite occurs as late-stage open-space filling of the carbonate-replacement deposits. On Serifos, hematite-barite (Fe-Ba) mineralization was sampled from Mega Livadi (Fig. 3B) and near Cape Kaparia (Fig. 3C). Sampling on Milos included the range of deposit styles in the western part of the island at Cape Vani, Kalogries, Katsimouti (Fig. 3D), Galana, and Triades (Fig. 3E, F), and on top of Profitis Ilias, as well as Fourkovouni (Fig. 3G) and Kastana (Fig. 3H) in the eastern part of the island. In the central Cyclades samples were taken from the vein-style mineralization at Cape Evros (Fig. 3I), Cape Charos (Fig. 3J), and Panormos Bay (Fig. 3K, L) on Mykonos and at Prassovounia and Agios Georgios on Antiparos. Sampling of galena at a number of these sites (Lavrion, Serifos, Milos, Tinos, Mykonos, and Antiparos) was previously reported in Wind et al. (2020). Mineral paragenesis was investigated in thin sections using reflected- and transmitted-light microscopy and is described below.

Selected samples were imaged, and the compositions of barite determined, with a JEOL 6610LV scanning electron microscope (SEM) and JEOL JXA-8230 electron microprobe at the University of Ottawa, Canada. Twelve samples were analyzed for the following elements by electron microprobe analysis (EMPA): S, Ba, Al, Sr, Ca, K, Pb, Fe, and Si. The analytical detection limits ranged from <0.01 wt % for Al, Ca, K, and Si to ≤0.04 wt % for Sr and Pb. For mineral separate analyses, samples were crushed, and single barite crystals were carefully selected under a binocular microscope. Fine powder of barite was analyzed for S and O isotopes using a Elementar Isotope Cube elemental analyzer (EA) interfaced to a Thermo Finnigan DeltaPlusXP isotope ratio mass spectrometer at the Ján Veizer Stable Isotope Laboratory at the University of Ottawa. Epoxy grain mounts of barite from 21 samples were also prepared for analysis using a New Wave UP193 laser attached to an AXIOM multicollector-inductively coupled plasma-mass spectrometer at GEOMAR Helmholtz Centre for Ocean Research Kiel, Germany. Prior to the LA-MC-ICP-MS analysis, individual crystals were imaged under transmitted and reflected light and with the SEM to identify inclusion-free areas for analysis of 87Sr/86Sr ratios and δ34S. Analyses and data reduction of 87Sr/86Sr ratios and δ34S followed the approach described by Fietzke et al. (2008) and Jamieson et al. (2016). Concentrations of a suite of 54 elements in four barite crystals were measured using an Excimer laser ablation system (GeoLasPro, Coherent) coupled to a double-focusing, high-resolution magnetic sector mass spectrometer (AttoM, Nu Instruments) at GEOMAR following the method of Fietzke and Frische (2016). Selected samples were also processed for measurement of S isotopes by EA-IRMS in coexisting galena. A detailed description of the analytical methods is provided in Appendix 1 together with the complete data sets in Appendix Tables A2 and A3.

Following is a brief description of the main characteristics of Cycladic barite. A complete list of the sample locations and paragenesis is included in Appendix Table A1.

Carbonate-hosted barite, Lavrion and Serifos

Two carbonate-hosted Pb-Zn-Ag orebodies of the Lavrion district were sampled for this study. In the Villia deposit (adit no. 123), in the north of Lavrion, several centimeter-scale, bladed, clear barite crystals occur as open-space filling in carbonate-replacement mineralization (mantos and chimneys). At the Plaka deposit (adit no. 145), massive barite and barite flakes occur as open-space filling in the breccia-style mineralization (Figs. 3A, 4A). The analyzed barite in this study was deposited during a late stage (stage II) that followed the main carbonate-replacement and breccia ore formation (stage I). At the Plaka deposit, stage I consists mainly of pyrite, Fe sphalerite, galena, chalcopyrite, and minor arsenopyrite and pyrrhotite with carbonate gangue (Wind et al., 2020). The barite associated with the stage II Pb-Zn-Ag mineralization occurs with carbonates, Fe-poor sphalerite, pyrite, and minor galena (Fig. 5A). We interpret the late barite-rich open-space filling and the main carbonate-replacement (Villia) and breccia-style (Plaka) Pb-Zn-Ag mineralization as having formed during the same event. Cavities filled by euhedral barite, indicating open-space crystallization and growth, contain similar sulfide minerals as the main ore-forming stages (Fig. 5A), consistent with formation at the same time during ongoing regional deformation. Coarse barite, intergrown with calcite and Fe oxides, was also observed in gossanous material close to the mine entrance (Fig. 5B). On a crystal scale, multiple growth zones are visible in the barite (Fig. 6B).

In the southwestern part of Serifos, carbonate-hosted barite was sampled from the mined Fe-Ba mineralization that occurs as veins, open-space fillings, and breccia lenses (Fig. 3B, C). At Mega Livadi, the veins are concentrated in marbles and schists in the footwall of the Meghalo Livadi detachment (Salemink, 1985; Ducoux et al., 2017). Euhedral barite flakes (up to 3 cm in length) occur in crustiform-banded veins a few tens of meters in length, consisting of subparallel layers (≤2–3 cm wide) of carbonate, hematite, and rarely magnetite, limonite, and barite, with barite infilling open spaces in the veins (Fig. 3B). Although sulfides are mostly absent in the veins, fine pyrite and minor pyrrhotite inclusions are found in the barite (Fig. 5C). Similarly at Cape Kaparia, breccia veins and lenses several meters wide occur in marbles in the footwall of the Meghalo Livadi detachment. Euhedral bladed barite crystals (up to 1 cm in length) with radial growth or snowflake structures fill the open spaces between colloform-crustiform-banded clasts of hematite and limonite breccia lenses (Figs. 3C, 4B). Late calcite often forms a corona around barite. Small inclusions of Fe oxides (Fig. 5D) and minor pyrite occur in the barite, which also shows very fine growth zoning patterns (Fig. 6C).

Epithermal, sea-floor, and subseafloor barite mineralization, Milos

In the different deposits on Milos, hydrothermal barite occurs in veins crosscutting altered lava domes and pyroclastic/volcaniclastic sequences (Fig. 3D), in horizons within the volcanic and sedimentary strata (Fig. 3E), and as pods and impregnations in altered volcaniclastic sediments (Fig. 3F-H).

At Cape Vani, previous work cited above combined with observations in this study indicate barite was deposited throughout the paragenesis of Mn oxide mineralization. It occurs as disseminated crystals or cement and in pervasive hydrothermal alteration (barite, quartz, K-feldspar, illite) of the volcaniclastic sandstone that hosts the Mn ore, as well as in subvertical quartz-barite (Mn oxide, chalcedony, chert, K-feldspar) hydrothermal veins and stockworks in the andesites/dacites and overlying Mn-mineralized sediments. Quartz-barite veins are a few centimeters wide and contain barite crystals, which vary from a few millimeters to 3 cm in length. In places quartz-barite veins that crosscut Mn oxide mineralization contain minor sulfides (pyrite, galena). Barite also occurs as bedding-conformable layers, underlain by mineralogically similar discordant pipe-like bodies. These layers contain small (<5 mm), acicular, and milky barite crystals, with Mn oxides ± Fe hydroxides (oxidized sulfides) and chert that fill open spaces and relicts of K-feldspar crystals (Fig. 5E, F). Scanning electron microscopy-energy dispersive spectrometry (SEM-EDS) analyses indicate that Mn oxide minerals associated with pyrite in barite belong to the hollandite supergroup. Mn oxide minerals identified within the Vani Mn oxide deposit include todorokite, manjiroite, vernadite, and ramsdellite (Liakopoulos et al., 2001; Kilias et al., 2020). Rarely small inclusions of oxidized pyrite, galena, and (Pb-)Mn oxides were observed in barite.

Mineralization near Kalogries formed contemporaneously with the Cape Vani deposit and has a similar mineralogy and paragenesis. Quartz-barite veins hosted by manganiferous glauconitic volcaniclastic sandstone/sandy tuff were sampled for this work. The veins, which trend north-northeast to south-southwest, vary in thickness from a few centimeters to 10 cm and consist predominantly of barite, quartz, and minor Mn oxides, with relict K-feldspar crystals and glauconite clasts. Barite crystals preserve radial growth, with primary growth zones in the core and diffuse secondary zones at the rim and in smaller crystals (Fig. 6D).

At Kondaros-Katsimouti, samples were obtained from polymetallic epithermal quartz and barite veins that cut the dacitic lavas and overlying volcaniclastic sandstone. The veins are steeply dipping, subparallel, northeast-southwest trending, and up to 20 cm wide and extend over several tens of meters at Katsimouti (Fig. 3D). Barite occurs together with galena, pyrite, Fe-poor sphalerite, and minor tetrahedrite. Late-stage silicification, including quartz or chalcedony pseudomorphs after barite (Fig. 4C), has significantly reduced the sulfide and barite content of the veins.

At Galana and Triades, samples were taken from multiple barite- and quartz-cemented breccia domes and pod-like veins and lenses at the base of the breccia domes. Quartz and barite gangue are associated with galena, Fe-poor sphalerite, pyrite, and minor chalcopyrite, tetrahedrite-tennantite, and enargite (Fig. 5G); rare covellite and bornite of probable supergene origin also occur. Samples from the barite-filled pods contain euhedral bladed barite showing void-filling textures (Fig. 3E, F). Vein barite is massive to bladed and commonly has corroded rims due to silicification (Fig. 5G, H). Inclusions of galena, sphalerite, Pb oxides, and euhedral quartz were observed in barite (Fig. 5G).

At Profitis Ilias, barite was sampled from subvertical interconnected quartz veins that are up to 3 m wide and extend to a depth of at least 300 m (Kilias et al., 2001). Veins at the surface cut silicified rhyolitic tuffs and are several centimeters wide, banded, and composed mainly of quartz and barite; sulfides were not observed in the studied samples, but small inclusions of Fe oxides were observed. Coarse-grained bladed barite occupies the central portions of the veins (Fig. 4D); the early formed margins of the veins are mainly colloform with microcrystalline quartz belonging to the Au-bearing ore stage (Kilias et al., 2001).

Samples of barite from Fourkovouni, in the northeast of Milos, were collected from veins and open-space fillings that occur in the altered and silicified tuffaceous rock (Fig. 3G). Here, the volcaniclastic host rock is highly silicified and stained red, possibly indicating oxidized pyrite. Barite crystals are acicular to tabular and milky and show pronounced growth zoning (Fig. 6E).

At Kastana, also in the east of Milos, we sampled barite veins up to 5 cm wide and stockworks, as well as barite ± quartz impregnations in the volcaniclastic material (Fig. 3H). Here, massive to tabular barite occurs together with quartz, as well as barite clasts and barite laths in a matrix of microcrystalline quartz (Fig. 5I) and granular barite with abundant quartz inclusions (Fig. 5J). Rarely fine inclusions of K-feldspar, muscovite, and oxidized pyrite were also observed in barite.

Vein barite, Mykonos and Antiparos

At Cape Evros and Cape Charos on Mykonos, samples were collected from subvertical barite veins, up to 10 m wide and over kilometers long, which cut the metabasites of the Pelagonian zone and Miocene conglomeratic sequence in the hanging wall as well as the monzogranite in the footwall of the local detachment systems. Barite veins and stockworks (Fig. 3I, J) crosscut and postdate silicified breccias, which have been previously described by Skarpelis (2002) and Skarpelis and Gilg (2006). Samples of the silicified breccia from this study contain euhedral pyrite, minor chalcopyrite, and covellite (Figs. 4E, F, 5M). Banded barite veins, which crosscut the siliceous breccia, consist mainly of barite and Fe oxides/ hydroxides. Barite samples include massive to bladed barite crystals, ranging in size up to 7 cm (Fig. 3J). Sulfides (e.g., galena, Fe-poor sphalerite, pyrite, chalcopyrite: Wind et al., 2020) occur often interstitially between large barite laths but also as microinclusions, with pyrite and chalcopyrite as well as supergene covellite being the most common inclusions (Fig. 5K). Veinlets of pyrite and marcasite, at rims often oxidized to hematite, are intersecting barite veins and occur also between large barite laths (Fig. 5L). Late-stage quartz and hematite, as well as goethite of probable supergene origin, were observed in all studied samples. At Panormos Bay on Mykonos, banded veins consisting of massive and bladed barite (Fig. 3K, L) cut the clastic sediments of the Miocene conglomeratic sequence. Barite crystals vary in size from <1 to 5 cm in the different bands of the vein, whereas veinlets of fine-grained barite occur in the later stages of vein formation intergrown with quartz and galena with late supergene hematite ± anglesite at rims (17MYP3 in Figs. 3L, 4G top). In the center of the banded veins, polymetallic sulfides, including galena, tetrahedrite(-tennantite), chalcopyrite, and supergene covellite (Wind et al., 2020), together with quartz ± carbonates, fill spaces between large barite crystals (Fig. 5N), which show distinctive growth zoning (Fig. 6F).

At Agios Georgios on Antiparos, the vein system consists of two generations of steeply dipping quartz veins: an older NE-SW–trending generation filled with milky quartz and a younger NW-SE–trending set with clear quartz (Kevrekidis et al., 2015). Previous work and our own observations show barite occurs together with clear quartz, smithsonite, fluorite, argentiferous galena, minor native Au and Ag, and covellite. We sampled milky, commonly broken, and brecciated blades of barite in a vein surrounded in a matrix of Pb- and Mn-bearing Fe oxides/hydroxides, Pb-Mn oxides, and minor quartz (Fig. 4H), although small inclusions of galena, a few micrometers in size, are still preserved in the barite (Fig. 5O). At Prassovounia, NE-SW–trending milky to clear quartz veins cut the gneisses and schists of the Upper Cycladic blueschist nappe and contain low concentrations of base metal sulfides. We sampled bladed barite in association with Fe-poor sphalerite, galena, and minor pyrite in a matrix of microcrystalline quartz (Fig. 4I), where the sulfides are commonly oxidized and barite has partially replaced crystal rims, likely related to the late-stage silicification. We also sampled NE-SW–trending barite-carbonate veinlets that cut the marble of the Upper Cycladic blueschist nappe. Barite crystals are up to 4 cm in length, show radial growth, and are intergrown with coarse carbonates and Fe oxides (Fig. 5P). Inclusions of pyrite, now mostly oxidized, were observed in both samples of barite from Prassovounia.

Geochemistry of barite

Results of the semiquantitative analysis by SEM in EDS mode and analyses by EMPA and LA-ICP-MS are provided in Appendix Tables A2 and A3 and summarized in Table 2. Compositional zoning in barite was observed in samples from all of the investigated deposits, mainly recorded by variations in the amount of Sr in the crystal lattice (Figs. 6, 7). In backscattered electron (BSE) microscopy images, the brightness of the zones is correlated with the Sr content, which ranges from below detection limit of the SEM-EDS and EMPA (<0.04 wt %) to 5.2 wt % (Fig. 6A; Table 2). The lowest concentration of Sr measured by LA-ICP-MS is 435 ppm in sample 17L145_1 (crystal b, rim) from Lavrion. The strong negative correlation between BaO and SrO (Fig. 7) reflects the complete solid solution between barite and celestine (Hanor, 2000). These variations are linked to primary growth zones and secondary depletions often at the crystal rim (Fig. 6B-F). The majority of samples analyzed by SEM-EDS are from Milos, which also records the highest Sr concentrations (Fig. 6: Cape Vani, Kalogries, Katsimouti). However, barite from Cape Kaparia, Serifos, has higher average Sr contents than almost all barite samples from Milos (Fig. 6A), and the range of the sample is also greater than that at any individual deposit on Milos. The highest concentrations of Sr were measured by EMPA in two samples from Kalogries and Fourkovouni (Fig. 7; Table 2) with concentrations of 4.2 and 5.2 wt %, respectively. Overall, barite from barren submarine volcanic breccias shows a broad range and high concentrations of Sr (e.g., Figs. 6D, E, 7), whereas barite in association with sulfides from Milos, Lavrion, Mykonos, and Antiparos has a narrower range of Sr concentrations (0.8–2.7 wt %: Fig. 7).

The following elements were detected by LA-ICP-MS in nearly all analyzed samples: S, Ca, Zn, Cu, Sr, Ba, La, Ce, Pr, Gd, and Hf. However, it should be noted that because of unavoidable interferences (e.g., Sr2+ on Ca, S-O cluster on Zn and Cu, or BaH on La) concentrations of some trace elements may be overestimated. In samples associated with galena mineralization (e.g., Plaka, Lavrion; Katsimouti, Triades, and Galana, Milos; Cape Evros and Panormos Bay, Mykonos; Agios Georgios, Antiparos), Pb was detected by SEM-EDS analysis and concentrations of up to 0.64 wt % were measured by EMPA. In contrast, LA-ICP-MS analyses showed low concentrations of a few parts per million up to 20 ppm in samples lacking any association with galena mineralization (e.g., Mega Livadi, Serifos). Barite from Plaka, Lavrion, contains elevated Ca and Pb, which likely represent carbonate and galena inclusions, respectively. Rarely, increased Ca correlates with Sr-rich growth zones (e.g., traverse 17MAR2 from Fourkovouni, Milos: App. Fig. A1). Sodium concentrations were measured by SEM in barite grain mounts and concentrations of a few parts per million up to 52 ppm were found by LA-ICP-MS in the samples 17MAR3 (Milos) and 17MYP1 (Mykonos), potentially related to Na-rich fluid inclusions. Concentrations of other trace elements and rare earth elements (REEs) are generally at or below detection limits of EMPA and LA-ICP-MS (App. Tables A2, A3).

Stable isotope data of mineral separates

Sulfur isotopes were determined by EA-IRMS in 33 representative samples of barite and 24 samples of galena coexisting with barite or in associated veins. The results are reported in standard δ notation as per mil () relative to the Vienna Canyon Diablo Troilite (VCDT) standard. Barite has a wide range of δ34S values, between 9.3 and 31.5 (Table 2; App. Table A2; Fig. 8). Significant variation in δ34S values was observed between the different mineralization styles. Barite from the carbonate-hosted mineralization at Lavrion has δ34S values from 9.3 to 19.8. At Plaka, two samples taken underground from adit no. 145 with coarse tabular barite have the lowest δ34S values (9.3 and 11.7), whereas a sample with small barite flakes and barite associated with carbonate in gossan material near the mine entrance have higher values (19.5 and 19.8, respectively). Barite from Serifos has a narrow range of δ34S values from 10.8 to 13.5, with slightly higher δ34S values at Cape Kaparia (mean: ~12.9) than at Mega Livadi (mean: ~11.1). Barite from the different deposits on Milos has δ34S values of 21.2 to 24.6, except for two samples of vein barite from Katsimouti, which have lower δ34S values of 17.9 and 18.5. High δ34S values were measured in vein barite from Mykonos; barite from Cape Evros has δ34S values between 25.2 and 31.5 and shows three clusters around ~25.3, ~27.8, and ~31.1. Vein barite from Cape Charos has δ34S values of 25.4, and barite from Panormos Bay has values between 25.7 and 26.2. The δ34S values for barite from Antiparos vary between 19.6 and 23.5, with barite from Prassovounia (mean: 22.4) having higher values than barite from Agios Georgios (19.6).

Galena δ34S values range between –13.9 and 4.3 (Table 2; App. Table A2; Fig. 8). The majority of δ34S values are between –1.8 and 4.3, except for galena from Panormos Bay on Tinos, which has significantly lower values of –13.9 and –13.1. Galena from Plaka, Lavrion, has a narrow range of δ34S values from 3.0 to 3.6, contrasting with the wide range of δ34S values of barite (9.3–19.8) from the same site. Galena from Kamariza and Soureza at Lavrion has slightly lower δ34S values of –1.2 to 2.9 and 1.8 to 2.6, respectively. On Serifos, galena from the polymetallic mineralization at Moutoulas has δ34S values between 1.7 and 3.0. Galena from the epithermal deposits at Katsimouti, Galana, Triades, and Profitis Ilias on Milos has δ34S values between –1.1 and 2.2, whereas δ34S values of galena from Katsimouti are only positive. Galena from Mykonos has distinct δ34S values for the vein systems at Cape Evros and Panormos Bay. Galena from Cape Evros has δ34S values of 3.9 and 4.3 from the same samples where barite has high δ34S values (~31.1). Galena from Panormos Bay has δ34S values of –1.7 to –1.0 compared to 25.7 to 26.2 in the coexisting barite. Galena from Prassovounia on Antiparos has δ34S values between –1.8 and –0.2 compared to 23.5 in the coexisting barite.

Oxygen isotope ratios were determined by EA-IRMS in barite from 31 samples and are reported in δ notation as per mil () relative to the Vienna standard mean ocean water (VSMOW) standard in Table 2 and Figure 9, and the complete data set is reported in Appendix Table A2. The δ18O values range between 3.7 and 15.3, with significant variation between the different deposit types. Barite from the carbonate-hosted deposits at Lavrion has high δ18O values between 10.5 and 15.3, where lower δ18O values correlate with higher δ34S values. Barite from Serifos has δ18O values between 8.8 and 12.8. Here, δ18O values of barite from Cape Kaparia are higher δ18O values of barite from Mega Livadi. Barite from the different deposits on Milos has a narrow range of δ18O values from 9.7 to 11.3. Barite from Triades-Galana on Milos has the lowest δ18O values, and barite from Profitis Ilias has the highest. Samples of vein barite on Mykonos and Antiparos have the widest range of δ18O values, from 4.4 to 11.9 and 3.7 to 13.8, respectively. In general, we observed a negative covariation between the δ34S values and δ18O values of barite (Fig. 9), which is further discussed below.


In this study, we were able to make sequential measurements of 87Sr/86Sr and δ34S on the same ablation spot in barite by LA-MC-ICP-MS. In total 165 ablations were performed on 51 barite crystals from 21 samples. The results are summarized in Table 2 and shown in Figure 10, and the complete data set is reported in Appendix Table A3. For samples with replicate analyses, δ34S and 87Sr/86Sr were reproduced within analytical error, except in Sr-poor barite zones (<7 ×107 cps, <0.2 wt % Sr). Ablations in Sr-poor barite zones (light colors on Fig. 10) often yielded Sr isotope ratios outside the range of values obtained on Sr-rich zones of the same sample and crystal.

Strontium isotope ratios vary significantly from 0.7089 to 0.7144 (n = 155), but samples from the same deposit commonly show only minor variations, and 87Sr/86Sr ratios are remarkably homogeneous among all occurrences on one island (Fig. 10A, B). Except for a sample from Agios Georgios on Antiparos (App. Fig. A2), no significant correlation between the Sr concentration and the 87Sr/86Sr ratio could be found (Fig. 10B). The carbonate-hosted barite at Lavrion has 87Sr/86Sr ratios from 0.7093 to 0.7107, and that on Serifos from 0.7089 to 0.7112, though the majority of analyzed barite from Serifos has 87Sr/86Sr ratios of 0.7099 ± 0.0002 (Fig. 10A). Barite from seven mineral occurrences sampled on Milos has a narrow range of 87Sr/86Sr ratios of 0.7091 to 0.7104, with the least radiogenic barite from Galana and the most radiogenic from Kastana (Figs. 10, 11). The 87Sr/86Sr ratios of vein barite from Mykonos fall into two distinct groups (Figs. 10, 11), with more radiogenic Sr in barite from Cape Evros than in barite from Panormos Bay (0.7139–0.7144 and 0.7120–0.7125, respectively). Vein barite from Antiparos has 87Sr/86Sr ratios of 0.7110 to 0.7124, with generally higher and more homogeneous values from Prassovounia than from Agios Georgios (Fig. 11).

Barite δ34S values from mineral separates (EA-IRMS) and from LA-MC-ICP-MS spot analyses are very similar with a somewhat larger range for spot analyses of single crystals (Fig. 8). Sulfur isotope compositions obtained by LA-MC-ICP-MS spot analyses vary from 2.1 to 37.2 (Table 2), emphasizing a strong zoning in δ34S values at the crystal scale (e.g., Lavrion and Mykonos). In some cases, lower δ34S values might be related to microinclusions of sulfides in barite, often at the crystal rim (App. Fig. A3). Significant zoning in δ34S often correlates with a variation in the Sr concentration (Figs. 10, 12, 13); however, for most crystals the Sr isotope ratios did not vary.

The isotopic compositions (δ34S, δ18O, and 87Sr/86Sr) of barite vary distinctly between the different deposit types across the Cyclades and allow us to compare the source of barite-forming fluids, where a wide range of fluid sources has been previously suggested. In the extensively studied Lavrion district and on Milos, a range of mineral and fluid inclusion assemblages have been analyzed (Table 1) corresponding to multiple stages of mineralization (e.g., Naden et al., 2005; Bonsall et al., 2011; Scheffer et al., 2019) and different fluid sources (magmatic, meteoric, and seawater). Fluid mixing has been widely interpreted as a fundamental control on the mineralization in many of the hydrothermal systems (Table 1), but the nature of the fluids involved has been difficult to determine. We have examined this record in hydrothermal barite, which occurs in all of the deposits and almost always reflects different mixing processes. The results are discussed in the following sections according to the regional patterns of S sources, sulfide-sulfate relationships, O isotopes, and Sr isotopes. For details of the individual deposits, readers are referred to the summaries provided in Tables 1, 2, and 3 and in the synoptic Figures 2 and 14.

Sulfur sources in Cycladic barite

It is well established from experimental results that there is no isotopic fractionation of S between barite and dissolved sulfate at temperatures of 110° to 350°C (Kusakabe and Robinson, 1977); thus, the composition of barite can be used to trace the isotopic composition of sulfate from the parent fluid. Sulfur isotope compositions of barite in the Cyclades mineral district vary significantly between the different deposit types (Table 2; Fig. 8). In general, we observed that barite samples from the same deposit have only small variations in δ34S values, except for carbonate-hosted barite from Lavrion (2.1–24.5: Fig. 8). High δ34S values from analyses of mineral separates of barite from Plaka and Villia at Lavrion are similar to those previously reported (17.2–23.7 at Villia, Kamariza, Kiafa Mariza, and Megala Pefka: Bonsall et al., 2011). However, two samples from Plaka yielded low δ34S values (9.3–11.7), and by LA-MC-ICP-MS spot analyses we found significant variability in δ34S of individual barite crystals ranging from 2.4 to 20.4 (Fig. 13, further discussed below). These strong variations in S isotope ratios of barite in our samples from Plaka suggest that sulfate was derived from at least two distinct S sources, most probably magmatic and seawater-derived sulfate. High δ34S values (18.2–24.6) of barite at Lavrion have been interpreted as evidence of the involvement of Miocene seawater sulfate (Bonsall et al., 2011); however, lower values (2.1–7.4) reported in this study are more consistent with a magmatic source for sulfate or remobilization of S from stage I of ore mineralization (Figs. 8, 9). Our findings support the observations by Bonsall et al. (2011) and Scheffer et al. (2019) that barite formed late in the paragenetic sequence, and further suggest a shift in the dominant sources from magmatic to seawater-derived fluids, which is recorded in barite and is further supported by the δ18O values.

The small variation in S isotope values for the Fe-Ba mineralization at Serifos (10.0–13.5) likely indicates that sulfate is derived from one dominant sulfate source. Since the S is significantly lighter than Miocene seawater sulfate (Fig. 8; δ34SSO4 = 21.2–22.6: Burdett et al., 1989; Paytan et al., 1998), magmatic sulfate might be the source of sulfate for the barite-hematite deposits on Serifos. In a few barite crystals, we observed small inclusions of pyrite (Fig. 4C), which are often oxidized to hematite, suggesting that lower δ34S values might also have been produced by the oxidation of reduced magmatic S. Ducoux et al. (2017) suggested that surface waters (oceanic or meteoric) circulated along the detachment and penetrated into the footwall, leading to remobilization of Fe from high-temperature magnetite skarns providing an alternative (i.e., leached magmatic) source of S. However, the spatial relationship of the Fe-Ba deposits to the high-temperature skarns and rather homogeneous δ34S values of barite support a juvenile S source for barite mineralization.

The δ34S values of barite from the epithermal veins on Milos tend to overlap with the present Mediterranean seawater value (δ34SSO4 = 20.7: Böttcher et al., 1998) and the isotopic composition of coastal seawater in the Palaeochori Bay (δ34SSO4 = 21.2: Gilhooly et al., 2014). Our S isotope values of barite from Milos are similar to the values from Hauck (1984), Hein et al. (2000), Marschik et al. (2010), Ivarsson et al. (2019), and Schaarschmidt et al. (2021), and all authors proposed a seawater-dominated hydrothermal system for the mineralization (Table 1). Whereas δ34S values obtained by IRMS analysis for the different deposits on Milos are generally within the range of seawater sulfate, LA-MC-ICP-MS spot analyses yielded δ34S values of up to 27 within individual barite crystals (e.g., Galana and Kastana; Fig. 8; Table 2). One explanation for the high δ34S values might be that the sulfate was shifted isotopically in response to BSR, as previously suggested by Ivarsson et al. (2019) for the Cape Vani deposit. However, for most barite occurrences on Milos, BSR seems unlikely because of the high fluid temperatures of 100° to 300°C estimated by fluid inclusion work (Table 1) and calculated temperatures for barite-galena equilibrium (this study). These high δ34S values of barite might also reflect a mixture of seawater sulfate and sulfate leached by interaction with the wall rock, as indicated by the Sr isotope ratios. However, whole-rock δ34S values of the basement rock types on Milos are lacking, and whole-rock δ34S values of the metamorphic basement (phyllites, Cycladic blueschist unit) from Santorini are generally low (0.5–18.2, majority <2.5: Hubberten et al., 1976). A similar range was reported for whole-rock δ34S values of total S in Santorini lavas (0.1–17.5, majority ≤6: Hubberten et al., 1975). The absence of a source of high δ34S values makes the interpretation of δ34S values in barite that are higher than that of Mediterranean seawater uncertain. We found significantly lower δ34S values of 16.5 to 22.6 in vein barite from Katsimouti, similar to the values reported by Schaarschmidt et al. (2021), indicating either oxidation of reduced S in the hydrothermal fluids or a direct contribution of oxidized magmatic S to the hydrothermal system. Schaarschmidt et al. (2021) suggested that boiling caused the lower δ34S values at Katsimouti; however, boiling was an important process at most mineral occurrences on Milos (Table 1; e.g., Christanis and St. Seymour, 1995; Kilias et al., 2001; Papavasiliou et al., 2016; Smith et al., 2018; Schaarschmidt et al., 2021) without significantly influencing the δ34S values of barite. Similar, low δ34S values (~17) have been reported from pore-water sulfate at hydrothermal vents in Palaeochori Bay, where an input of secondary sulfate from sulfide oxidation was suggested (Gilhooly et al., 2014). Miles (2021) reported low δ34S values of jarosite (–7.6 to 9) from Triades and alunite (5.9) and jarosite (6.9) from Katsimouti and linked these to steam-heated alteration above the groundwater table. In general, however, S isotope values of barite from Milos indicate seawater-dominated hydrothermal fluid sources for the past 3 m.y. (Hein et al., 2000; Marschik et al., 2010; Ivarsson et al., 2019). Sulfide S in galena from the Milos deposits has δ34S values closer to H2S in the free vent gases and pore waters (~2.5 and 2.7, respectively: Gilhooly et al., 2014) collected from Palaeochori Bay. These are at the low end of δ34S values reported for native S and fumarolic vapor from Nisyros and Milos (1.0–6.7: Hauck, 1984; Marini et al., 2002) and elemental S of Nea Kameni, Santorini (0.0–10.0: Hubberten et al., 1975) and are discussed in more detail in the following section.

High δ34S values of barite were found in the veins from Mykonos. All of the δ34S values are higher than that of Miocene seawater sulfate and higher than values reported by Tombros et al. (2015). Pronounced variations were observed in individual barite crystals from Cape Evros on Mykonos (28.1–37.2, sample 17MYV2_1), with higher δ34S values corresponding to lower Sr concentrations. In the barite from Cape Evros, we observed three clusters of δ34S values: δ34S values >30 in coarse bladed barite associated with sulfides that occur as inclusions (Fig. 5K) or interstitially; δ34S values ~28 in finer-grained barite (<1 cm) together with quartz (Fig. 4E barite vein in the center); and δ34S values ~25 in fine to coarse bladed barite rich in Fe oxides/hydroxides (Figs. 4F, 5L, M). In contrast, mineral separates of barite from the different zones of the vein at Panormos Bay have uniform δ34S values and are similar to those from Cape Charos and the lowest δ34S values of barite from Cape Evros (25–26). However, on the crystal scale the δ34S values of barite from Panormos Bay vary from lower than Miocene seawater sulfate to 27.8 (Fig. 12). This was most likely due to preferential reduction of 32SO4 in pore fluids, leading to residual enrichment of 34S in sulfate. Partial reduction of the porewater sulfate might have been caused by ferrous Fe in the hydrothermal fluid (Chiba et al., 1998; e.g., as suggested for the barite-gold orebodies on Wetar Island, Indonesia: Scotney et al., 2005). In the absence of a reservoir of high δ34S values, δ34S values in barite that are higher than that of Miocene seawater might also imply a source in the wall rocks that was oxidized and leached, but this source has not been identified. Variation in the δ34S values of barite on the crystal scale might also be an indication of local isotopic fractionation of S during precipitation (e.g., during boiling, which has been suggested for several deposits: Table 1). Isotopic fractionation of S at the crystal scale during boiling has been documented in sulfide minerals in other epithermal vein systems (e.g., McKibben and Eldridge, 1990; Arehart et al., 1993).

Vein barite from Antiparos also has a range of δ34S values from higher than that of Miocene seawater sulfate (20.3–26.3 at Prassovounia) to significantly lower (15.4–22.4 at Agios Georgios). The heavier S in barite from Prassovounia can be best explained by a modified seawater sulfate source or leaching of heavy S from the wall rocks, whereas at Agios Georgios a mix of seawater sulfate and sulfate derived from magmatic fluids is also possible. This is in contrast to the model by Kevrekidis et al. (2015), who proposed a mix of magmatic and meteoric fluid sources. The fluid inclusion work of Kevrekidis et al. (2015) showed large temperature and salinity variations (130°–400°C, 4.8–15.7 wt % NaCl equiv) in ore-stage vein quartz, which they considered to be evidence of dominantly magmatic fluids. However, we note that δ34S values of barite are more indicative of a dominantly seawater related system. These data suggest mixing of two fluid sources or possibly that barite postdates the main ore-forming event and records a shift in the dominant fluid source of the system.

Sulfide-sulfate relationships

The δ34S values of galena coexisting with barite in the different deposits plot close to magmatic S values from the Aegean arc (Fig. 8). Galena was analyzed in the same samples as barite from Plaka at Lavrion, Galana and Triades on Milos, Cape Evros and Panormos Bay on Mykonos, and Prassovounia on Antiparos (Table 2). Galena and barite from Katsimouti and Profitis Ilias on Milos are from different samples but collected at the same site and therefore temperatures for the barite-galena equilibrium were not calculated for these samples.

At Lavrion, S isotope values of galena are mostly positive in the carbonate-replacement mineralization at Plaka and Soureza and range from negative to positive values at Kamariza, similar to previously reported values (Table 1; Skarpelis et al., 2007; Bonsall et al., 2011). Bonsall et al. (2011) reported slightly lower δ34S values for sulfides from Kamariza, possibly reflecting a temperature-dependent fractionation or disproportionation of magmatic SO2. Our δ34S values of galena from Plaka show almost no variation (3.0–3.6), in contrast to the δ34S values of barite from the same samples, which range between 2.1 and 24.5 (Fig. 8), demonstrating that much of the barite and galena analyzed in this study were not in isotopic equilibrium. δ34S values of galena are similar to the lowest δ34S values of barite, indicating a similar source of S for those samples (e.g., from a magmatic contribution or remobilization and inherited from stage I sulfides during barite formation). Skarn-associated galena from the polymetallic mineralization at Moutoulas on Serifos similarly shows only minor variations in δ34S values (1.7–3.0), consistent with values for pyrite and sphalerite reported by St. Seymour et al. (2009) but slightly lower than the values reported by Fitros et al. (2017). Those studies suggested a dominantly magmatic source for the S. Galena from the epithermal mineralization on Milos has δ34S values from –1.1 to 2.2, whereas galena from epithermal veins at Katsimouti has only positive δ34S values, and galena from Triades, Galana, and a drill core sample from Profitis Ilias has mainly negative values. Our δ34S values for galena from Triades and Galana are slightly lower than the values obtained for sulfides by Hauck (1984), Marschik et al. (2010), and Schaarschmidt et al. (2021; Table 1). Marschik et al. (2010) suggested that the Δ34S(sulfate-sulfide) between sulfide and barite at Triades and Galana was due to sulfate reduction from a seawater-dominated fluid under conditions of isotopic equilibrium. Galena from vein-type mineralization at Panormos Bay on Tinos was previously reported to have strongly negative δ34S values (Table 1; Tombros et al., 2007) that were explained by leaching of S from the underlying basement by a hydrothermal fluid derived from meteoric water. The low δ34S values were attributed to metamorphic pyrite with δ34S of –11.4 from a schist in the Cycladic blueschist unit (Table 3). Our sample of galena from a quartz vein at Panormos Bay on Tinos confirms the low δ34S values, but there are no data for coexisting barite. At Cape Evros on Mykonos, Tombros et al. (2015) reported a wide range of δ34S values for sulfides from 4.0 to 25.1, but the high δ34S values of the sulfides were not found in our study. Galena in our study shows significant local variations with positive δ34S values (3.9–4.3) at Cape Evros, similar to the magmatic S values, and negative δ34S values (–1.7 to –1.0) at Panormos Bay. The range of values closely match the variation in δ34S values of barite for the two locations with a consistent Δ34S of ~27. Tombros et al. (2015) proposed an early magmatic fluid source and later seawater-dominated system, whereas Skarpelis and Gilg (2006) suggested that the barite veins were related to basinal brines. Our data suggest a seawater-dominated fluid that interacted extensively with the basement, based on Sr isotopes of the whole rock and veins leading to important local isotopic variations on Mykonos. Galena from two vein systems at Prassovounia on Antiparos have slightly negative δ34S values (–1.8 to –0.2). Kevrekidis et al. (2015) proposed a link between the Miocene intrusion on Paros and the ore mineralization based on O and H isotopes from vein quartz, and a magmatic S component is supported by the low δ34S values of barite from Agios Georgios in this study.

Temperatures were calculated for barite-galena equilibrium using fractionation factors from Ohmoto and Rye (1979) reported in Seal (2006) and S isotope data from the different deposits of the Cyclades. Some authors have suggested that sulfide and sulfate minerals were coprecipitated under conditions of isotopic equilibrium (e.g., Triades and Galana on Milos; Marschik et al., 2010). Calculated temperatures for barite-galena equilibrium in these deposits vary between 259° and 281°C (samples 17MGA03 and 17MT03) and are slightly higher than temperatures estimated by Smith et al. (2018) from fluid inclusion studies in barite, quartz, and sphalerite but within the range of values of Vavelidis and Melfos (1998) from fluid inclusion data in barite of Triades (Table 1). In barite veins from Mykonos, inclusions of galena in barite, and vice versa, indicate these phases were in equilibrium in 17MYP1 and 17MYP4 from Panormos Bay and 17MYV2_1 from Cape Evros. Our data give equilibrium temperatures of 237° to 246°C at Panormos Bay and 246°C at Cape Evros. These are within the range of formation temperatures of hydrothermal chlorite in the cataclasites of the local detachment (203°–258°C: Menant et al., 2013) and within error of the fluid inclusion trapping temperatures in quartz of the mineralized breccia (~230°C: Skarpelis and Gilg, 2006). Intergrowths of barite and galena in sample 17AP2 from Prassovounia on Antiparos also suggest coprecipitation (Fig. 4I). The calculated temperature for barite-galena equilibrium in this sample is 263°C, which is within the range of temperatures reported from fluid inclusion microthermometry in vein quartz from Agios Georgios on Antiparos (130°–400°C: Kevrekidis et al., 2015). For the carbonate-hosted deposits (Lavrion) barite-galena pairs give unreasonably high temperatures, indicating that sulfide and sulfate minerals were not precipitated under equilibrium conditions and consistent with reversed fractionations calculated from previously published isotopic data (Skarpelis et al., 2007; Bonsall et al., 2011).

Oxygen isotopes in barite

At hydrothermal temperatures, the O isotope composition of sulfate in barite equilibrates with water according to experimentally determined isotopic fractionation (e.g., Kusakabe and Robinson, 1977; Chiba and Sakai, 1985; Seal et al., 2000). However, the O isotope exchange rate of dissolved sulfate and water depends strongly on the temperature and pH of the solution, with small fractionation factors at temperatures >200°C (Seal et al., 2000).

We observe a negative correlation between the δ18O values and δ34S values of barite, in general, and within individual deposit clusters (e.g., Mykonos and Antiparos: Fig. 9) that reflects the range of end-member fluids involved in the different mineralizing systems. Ore-forming fluids for the Cyclades mineral deposits acquired distinct O isotope signatures depending on the fluid source and the extent of interaction with wall rocks of different compositions (δ18O = 10–30; Figs. 9, 14; Table 3), which we define in the following. Although Turchyn and Schrag (2004) have documented marine barite δ18OSO4 excursions in the past 10 m.y. (8–13), Markovic et al. (2016) suggest a more steady δ18OSO4 of 6 to 7 since 4 Ma. The present-day δ18OSO4 of surface seawater from Palaeochori Bay on Milos is 9 (Gilhooly et al., 2014). The O isotope compositions in Cycladic barite are distinguishable from a pure meteoric water source (–8 ± 7; Table 3). Fumarolic vapors from Nisyros island have low δ18O values from 3.0 to 4.9 (Marini et al., 2002). The majority of wholerock δ18O values of the Miocene intrusions on Serifos and Mykonos are between 9.2 and 11.5 (Altherr et al., 1988; Altherr and Siebel, 2002), and whole-rock δ18O values of Quaternary volcanic rocks from Milos are generally between 8.0 to 9.5, with values up to 11.6 in glass (Briqueu et al., 1986). The basement rock types in the Cyclades have wholerock δ18O values that are generally higher than those of magmatic fluids, although the data are limited (Table 3).

The δ18O values of barite from the carbonate-replacement and breccia-style deposits at Lavrion and carbonate-hosted vein-type deposits of Serifos range from 8 to 16, consistent with fluid-rock interaction with an isotopically heavy source like the carbonate host rocks. Unaltered carbonates in the Cyclades have δ18O values of >20 (Table 3), whereas calcite and dolomite from the deposits have δ18O values of ~15 (Bonsall et al., 2011; Berger et al., 2013), consistent with mixing of heavy O from the carbonaceous wall rocks with lighter O in a magmatic fluid and/or surface-derived water (seawater ± meteoric). The δ18O values of barite from Serifos are within the range of values reported by Salemink (1985), who also proposed a magmatic fluid source with wall-rock interaction. The δ18O values of barite from the epithermal deposits on Milos range from 9 to 11 and are similar to the values reported by Marschik et al. (2010) of barite from Triades and Galana. Ivarsson et al. (2019) reported a broader range of δ18O values of barite from Vani and linked these to a seawater sulfate source modified through BSR. The narrow range of δ18O values of barite from this study are close to the values of modern marine sulfate of the Palaeochori Bay on southern Milos (Fig. 9). In contrast to the δ34S values, δ18O values of barite from Katsimouti are not distinguishable from those of barite at other occurrences on Milos, although variable degrees of boiling between the deposits should have resulted in differences in the δ18O values of barite (Huston, 1997). Similarities in the O isotope composition of barite and Quaternary volcanic rocks indicate that magmatic fluids could also have played a role in ore formation at Katsimouti. The polymetallic veins on Mykonos and Antiparos have δ18O values of barite that range from 4 up to 12. High δ18O values are consistent with a mix of seawater ± magmatic-hydrothermal fluids. Low δ18O values from our study (Cape Evros, Mykonos) are higher than the δ18O values of barite reported by Tombros et al. (2015). Tombros et al. (2015) reported low δ18O values (<4) for vein barite and slightly higher values for silica-cap barite (~5) and vein quartz (6.4–9.5) at Cape Evros. Wholerock δ18O values of the Miocene intrusion from Mykonos vary between 4 and 11.5 (Altherr et al., 1988), with the lowest values in altered rocks near the contact with metapelites and gneisses of the Cycladic blueschist unit and Cycladic basement, which could explain the low values from our study and those reported by Tombros et al. (2015). Low δ18O values (<6) for some barite on Mykonos and Antiparos might also be related to a meteoric water component or basinal brines, as suggested by Skarpelis and Gilg (2006). However, this contrasts with the observations based on the δ34S values, which indicate a modified seawater source with wall-rock equilibration. This is further supported by the data on Sr isotopes in barite, discussed below.

Strontium isotopes and crustal sources

In Figure 11 we compiled the Sr isotope compositions of potential source rocks in the Cyclades, as well as the composition of Miocene to present-day seawater. Since 12 Ma, the 87Sr/86Sr composition of seawater has increased from 0.7088 to 0.7092 (e.g., Veizer, 1989; Hodell et al., 1991; Farrell et al., 1995). Values for the intrusions and volcanic rocks are recalculated to their initial 87Sr/86Sr ratios based on their reported ages, which are further described in Appendix 1. Miocene Cycladic granitoids have 87Sr/86Sr ratios of 0.709 to 0.712. Stouraiti et al. (2010) suggested a crustal source for Sr in the intrusions, derived from a mix of radiogenic gneisses and less radiogenic metabasites and marbles. The underlying Carboniferous to Cretaceous basement rocks across the Cyclades also have a wide range of Sr isotope ratios, with generally low 87Sr/86Sr in metabasites and marbles (≤0.709), high 87Sr/86Sr (0.716–0.766) in the gneisses of the Cycladic basement, and intermediate 87Sr/86Sr (0.708–0.739) in clastic metapelites of the Cycladic blueschist unit. Quaternary volcanic rocks along the active volcanic arc, in contrast, have low 87Sr/86Sr consistent with a mantle source at Milos and neighboring Santorini and Kolumbo (0.704–0.707).

The range of Sr isotope ratios of barite from carbonate-hosted deposits at Plaka, Lavrion (0.7092–0.7108) indicates a mix of different upper crustal Sr sources. The 87Sr/86Sr ratios of the Plaka granodiorite (Altherr et al., 1988; Stouraiti et al., 2010) and barite overlap, indicating that they could be a source of Sr for the associated barite (Fig. 11). However, a spatial relationship to pegmatitic dikes, which are enriched in Ba and Sr (>700 ppm and >250 ppm, respectively; Skarpelis et al., 2008; Bonsall et al., 2011; Berger et al., 2013) suggests that they may have been a more likely source, although no Sr isotope data are available for these rocks. Barite with 87Sr/86Sr lower than the Plaka granodiorite could be explained by Miocene seawater mixing with the barite-forming fluids or interactions with the immediate carbonate host rock (marble, Fig. 11), which is supported by the δ18O values of barite, whereas higher 87Sr/86Sr values are indicative of an interaction of the hydrothermal fluid with the radiogenic metapelites of the Cycladic blueschist unit. Barite from the carbonate-hosted deposits on Serifos has a narrow range of Sr isotope ratios (0.7097–0.7108), except for some outliers. The 87Sr/86Sr compositions are close to those of the nearby granitoid intrusions (Serifos granodiorite and dikes in Fig. 11; Altherr et al., 1988; Altherr and Siebel, 2002; Stouraiti et al., 2010, 2018), suggesting little or no input from the immediate carbonate host rock.

The range of Sr isotope ratios of barite in epithermal deposits on Milos can be explained by a mixture of a seawater source and an upper crustal source of radiogenic Sr (e.g., Hein et al., 2000; Naden et al., 2005; Marschik et al., 2010; Schaarschmidt et al., 2021). Quaternary volcanic rocks and Neogene sediments from Milos have much lower whole-rock 87Sr/86Sr values than barite (Fig. 11). Altered volcanic rocks show a significant shift in 87Sr/86Sr closer to the values for local geothermal waters (Naden et al., 2005). Schists, glaucophane schists, and eclogites of the crystalline metamorphic basement (Cycladic blueschist unit) on Milos have a wide range of 87Sr/86Sr (0.7033–0.7151; Briqueu et al., 1986; Naden et al., 2005; Schaarschmidt et al., 2021). We observed small variations in Sr isotope ratios of the different barite occurrences on Milos. These are potentially related to the amount of mixing between seawater and this type of hydrothermal fluid or the degree of interaction of the hydrothermal fluid with different basement rock types. Local variations in the 87Sr/86Sr composition of barite can also be caused by reaction of the hydrothermal fluid with different proportions of the metamorphic basement and the Quaternary volcanic rocks as proposed by Schaarschmidt et al. (2021). Lower 87Sr/86Sr values of sphalerite from Milos (Fig. 11; Marschik et al., 2010; Schaarschmidt et al., 2021) compared to barite support the interaction of the hydrothermal fluid with the metamorphic basement and Quaternary lavas. Strontium isotope ratios of barite at Galana, and to a lesser extent at Triades, are permissive of Sr being partially derived from the volcanic rocks and the metamorphic basement, whereas other barite occurrences on Milos (e.g., barite from Profitis Ilias, Fourkovouni, and Kastana) require a more radiogenic crustal source in the crystalline metamorphic rocks.

The Sr isotope ratios of vein barite from Mykonos and Antiparos are completely different from Milos, Serifos, and Lavrion (Fig. 11). Barite from Cape Evros on Mykonos is isotopically homogeneous, with 87Sr/86Sr values averaging ~0.7141. Tombros et al. (2015) reported a wider range of 87Sr/86Sr for barite from a banded barite vein at Cape Evros (0.7124–0.7146) and one sample from the siliceous cap with 0.7095. Barite from Panormos Bay, with 87Sr/86Sr values averaging ~0.7122, has distinctly less radiogenic Sr than barite from Cape Evros (Fig. 11). Significant local variations in the isotopic composition of the mineralization at Panormos Bay and Cape Evros were previously reported by Wind et al. (2020), based on the Pb isotope composition of galena. In general, however, the Sr in barite from Mykonos is too radiogenic to have been derived from the local Miocene granitoid (Mykonos granite and dikes in Fig. 11). Altherr et al. (1988) suggested metapelites and gneisses of the basement were a likely source of radiogenic Sr for the granitoid. In contrast to the values from Altherr et al. (1988), Tombros et al. (2015) reported significantly higher 87Sr/86Sr ratios for a sample of a K-feldspar-biotite monzogranite and silicified monzogranite (0.7303 and 0.7138, respectively) from Cape Evros. The possible role of basement rocks, as well as metabasites, and Miocene conglomeratic sequences in the hanging wall cannot be assessed owing to the lack of whole-rock 87Sr/86Sr measurements of those rocks.

The Sr isotope ratios of vein barite from Antiparos similarly indicate a radiogenic source with some variations between the different deposits on the same island. The 87Sr/86Sr ratios of barite from Agios Georgios and Prassovounia are within the range of values for the Pliocene rhyolite on Antiparos (Fig. 11; Innocenti et al., 1982). However, a trend toward more radiogenic Sr in barite, which is accompanied by an increase in Sr content and δ34S values at Agios Georgios (App. Fig. A2), likely points to a change in fluid composition related to leaching of Sr from the radiogenic basement. In general, the Sr concentrations of the basement rock types (8–246 ppm: gneiss and schist; Table 3) are higher than in the Pliocene rhyolite (≤10 ppm: Innocenti et al., 1982), although S-type intrusions on Paros have Sr isotope ratios similar to those of the Pliocene rhyolite and higher Sr and Ba (42–383 ppm and 87–992 ppm, respectively: Stouraiti et al., 2010). Kevrekidis et al. (2015) have suggested that these intrusions were a potential source of the mineralizing fluids for the polymetallic veins on Antiparos. However, no age data exist to link the mineralization to the intrusions. Therefore, the Pliocene rhyolites that are spatially related to the veins appear to be a more likely source of Sr and S.

To summarize, for barite with a dominant seawater source (e.g., Milos, Mykonos, and Antiparos) 87Sr/86Sr ratios are homogeneous within deposits but vary between the different deposits on the same island, suggesting interaction of the hydrothermal fluid with the immediate host rock. However, in general, the broadly similar Sr isotope composition of Cycladic barite across the sampled region suggests that the basement may have a larger influence than the immediate host rocks as the dominant source of Sr. Strontium isotope ratios of barite define different crustal sources for the central Cyclades (87Sr/86Sr ≥0.711) and the western Cyclades (87Sr/86Sr ≤0.711) consistent with the regional basement influence described in Wind et al. (2020), and the isotopic signatures of the mineral occurrences in the west Cyclades are more homogeneous than in the central Cyclades.

Geochemical and isotopic zoning in barite

Strontium concentrations in Cycladic barite are high in comparison to ore-associated barite in many other deposit types (e.g., Griffith et al., 2018), although barite in VMS deposits has similar concentrations of Sr (e.g., Jamieson et al., 2016; Lajoie et al., 2020). In general, higher concentrations of Sr are found in barite from deposits with an obvious seawater component in the hydrothermal fluids (e.g., Milos). However, isotopic data suggest that most of the Sr was leached from the basement rocks and not derived solely from seawater. The narrow range of SrO (<1–3 wt %: Fig. 7) in barite associated with sulfides indicates mostly uniform hydrothermal fluid concentrations of Sr during barite formation. Variations in Sr content in individual barite crystals are either due to small changes in the fluid compositions, mixing of fluids, or the temperature dependence of Ba-Sr partitioning (Hanor, 2000, and references therein). Later silicification and/or circulation of post-ore hydrothermal fluids also can cause a decrease in Sr concentrations, which is often observed at the crystal rims (Fig. 6D, E). Concentrations of other trace elements are generally very low, similar to values reported for barite from ancient and modern hydrothermal vents (e.g., Hanor, 2000; Jamieson et al., 2016; Lajoie et al., 2020).

On a sample and crystal scale we commonly observed geochemical and isotopic variations that indicate a change in fluid composition during barite precipitation. Variations in the isotopic composition of Cycladic barite are often accompanied by changes in the Sr concentration. Barite from vein systems at Panormos Bay on Mykonos shows geochemical variations that can be best explained by fluctuations in temperature. In single crystals from these samples, a positive correlation between the Sr content and δ34S values was observed (Fig. 12B), while the 87Sr/86Sr ratios remained constant (Fig. 12C). One possible explanation is crystallization in a closed system with a decrease in temperature. Temperature-controlled variations in the Sr content of barite were previously suggested by Jamieson et al. (2016) and Lajoie et al. (2020). Another possible explanation is that hotter hydrothermal fluids have the ability to leach more Sr out of the surrounding host rock, potentially modifying the fluid composition.

In barite samples from the carbonate-hosted deposits at Plaka, Lavrion, we observed an increase in δ34S values with a strong decrease in the Sr content (Fig. 13C) and an increase in 87Sr/86Sr. The pronounced zoning is indicative of a change in fluid sources, consistent with models of the mineralization at Lavrion (e.g., Bonsall et al., 2011; Scheffer et al., 2017, 2019). We propose that the low δ34S values and less radiogenic Sr observed in barite cores (Fig. 13A-E) point to a magmaticdominated fluid with interaction with the carbonaceous wall rock, whereas the high δ34S values and higher 87Sr/86Sr in the rims point to a fluid dominated by seawater and modified by additional leaching of Sr from a radiogenic basement source. A comparison of all samples from Plaka, Lavrion (Fig. 13E), also supports the suggestion of a positive correlation between δ34S values and 87Sr/86Sr highlighting a shift in the dominant fluid source.

We also observed the reverse trend in some barite (e.g., Agios Georgios on Antiparos: App. Fig. A2) where δ34S values decrease with decreasing Sr content and decreasing 87Sr/86Sr. Changes in both isotopic systems and Sr concentrations are recorded at the mineral scale in response to episodic mixing of end-member fluids. In some cases, the transition from coarse barren barite to smaller ore-associated barite crystals (e.g., Galana on Milos) shows decreasing Sr from >1.5 to <1 wt % with increasing 87Sr/86Sr from 0.7092 to 0.7099 but constant δ34S. As suggested above, the low 87Sr/86Sr ratios could be explained by Sr from modern seawater or the Quaternary volcanic rocks, whereas the more radiogenic values imply mixing with a fluid that had equilibrated with Sr from the basement.

Fluid mixing is a fundamental process in many hydrothermal ore deposits, including the spectrum of mineral deposit types in the Cyclades mineral district (Table 1). However, it is challenging to draw a comparison of fluid mixing processes between the broad range of deposit types because of the variety of styles of mineralization, complicating the interpretations of their origins. In this study, we examined one mineral, hydrothermal barite, which occurs in all of the deposits and records mixing of different magmatic to surface-derived (seawater ± meteoric) fluids. In most cases, barite precipitation was close to the terminal stage of mineralization, where mixing was coincident with (or caused) the end of base metal deposition, whereas in others (e.g., Lavrion district) barite formation occurred over multiple stages and involved more diverse fluid sources (Bonsall et al., 2011; Scheffer et al., 2017, 2019), including magmatic, meteoric, and seawater-derived fluids. The distinct isotopic signatures of barite can be used to trace this complex interplay of different fluid and crustal sources. In particular, the geochemical and isotopic compositions reveal that seawater was a source of sulfate in many of the deposits, but magmatic and meteoric fluids also played a role.

Figure 14 is a summary of the inferred fluid and wall-rock sources for the different deposit types in the Cyclades mineral district. They include (1) a direct magmatic source (e.g., in carbonate-hosted deposits at Mega Livadi, Serifos), (2) magmatic fluids that have reacted with the carbonates (marble) of the Basal unit and/or Cycladic blueschist unit (Cape Kaparia, Serifos, stage I of Plaka, Lavrion), (3) magmatic fluids that have mixed with seawater and reacted with the carbonates (marble) of the Basal unit and/or metapelites of the Cycladic blueschist unit (stage II of Plaka and Villia, Lavrion), (4) magmatic (likely Pliocene) and seawater fluids that have reacted with the basement gneisses or schists of the Cycladic blueschist unit (Agios Georgios, Antiparos), (5) seawater-dominated fluids with magmatic ± meteoric contributions that reacted with the basement gneisses/metapelites and/or rock types in the hanging wall (Prassovounia, Antiparos; Cape Evros, Cape Charos, and Panormos Bay, Mykonos), (6) seawater-dominated fluids that have reacted with metapelites of the Cycladic blueschist unit and Quaternary volcanic rocks (Cape Vani, Kalogries, Galana-Triades, Profitis Ilias, Fourkovouni, and Kastana, Milos), and (7) seawater-dominated fluids with Quaternary magmatic contributions that have reacted with metapelites of the Cycladic blueschist unit (Katsimouti, Milos).

Strontium isotope ratios support the strong influence of the underlying basement in the source of labile elements (e.g., Ba and Sr) in barite and correlate with the regional trends in Pb isotope composition observed in Cycladic galena by Wind et al. (2020). We show a clear division between 87Sr/86Sr ≥ 0.711 values in barite from Antiparos and Mykonos in the central Cyclades and 87Sr/86Sr ≤ 0.711 in barite from Lavrion, Serifos, and Milos in the west Cyclades, separated by the Trans-Cycladic thrust (Fig. 1). This division tracks the exposures of the Lower Cycladic blueschist nappe and Upper Cycladic blueschist nappe.

Figure 2 schematically illustrates the inferred fluid and crustal sources of the different barite occurrences in the Cyclades from the late Miocene into the Quaternary. At Lavrion [1–2], Serifos [3], and Mykonos [4–5] regional detachment faults provided pathways for the deep circulation of the hydrothermal fluids. On Antiparos [6–7] and Milos [8–10] mixing of surface-derived fluids, predominately seawater, and magmatic fluids was facilitated by normal faults. In the following, the numbers in brackets correspond to the numbers in Figure 2.

Stage II barite mineralization at Plaka [1] and Villia [2] at Lavrion has a wide range of δ34S values, consistent with a range of S sources. Individual crystals of barite show strong zoning in δ34S, Sr concentration and 87Sr/86Sr ratios, with low δ34S close to magmatic values in the core and high δ34S values close to seawater sulfate at the rim. Barite with low δ34S values but high δ18O values, similar to the ore-hosting carbonates (Bonsall et al., 2011; Berger et al., 2013), indicates that the magmatic fluids likely exchanged with the carbonate wall rocks, whereas a shift to high δ34S values and intermediate δ18O values is permissive of a seawater-dominated fluid. Strontium isotope ratios are similar to the Miocene intrusions and interpreted to record dominantly magmatic hydrothermal fluids, where small variations are likely related to the interaction of the hydrothermal fluids with the wall rocks (e.g., marbles and/or metapelites of the Cycladic blueschist unit). Scheffer et al. (2017, 2019) observed similar variations in their study of carbonates from Lavrion, which were interpreted to have formed in a magmatic-dominated system that evolved to later-stage breccia and vein mineralization involving meteoric water and evaporated seawater.

Barite from carbonate-hosted, intrusion-related veins at Serifos [3] has a very narrow range of δ34S, δ18O, and 87Sr/86Sr values, suggesting that mixing of fluids was limited. As observed at Lavrion, the low δ34S and intermediate to high δ18O, as well as 87Sr/86Sr values similar to the granitoid intrusion, suggest a magmatic fluid with minor interaction with the carbonate wall rocks. These relationships are similar to the δ34S and δ18O systematics of barite in other well-known hightemperature carbonate-replacement and skarn deposits, such as in the Colorado mineral belt and in Peru (e.g., Stegen et al., 1990; Thompson and Beaty, 1990; Baumgartner et al., 2008).

On Mykonos, the isotopic signatures of barite from Cape Charos and Cape Evros [4] are distinct from those at Panormos Bay [5]. At Cape Evros, high δ34S values of barite are consistent with a fractionated Miocene seawater-dominated fluid source. To explain the range of O isotope values, especially low δ18O values of barite from Cape Evros close to the values from meteoric water, basinal brines are a possible source, but also interaction with the immediate host rock could cause the low δ18O values. Local variations in Sr isotope ratios suggest distinct sources of Sr, such as leaching of the basement gneisses or the Miocene conglomeratic sequence (Neogene basin sediments in Fig. 2), as observed at Panormos Bay. Differences in the metal sources at Panormos Bay and Cape Evros were previously reported by Wind et al. (2020) based on the Pb isotope ratios and trace element concentrations of galena. Also, Lahti and Govett (1981) observed local variations in the geochemistry of whole-rock samples, with an enrichment in Cu, potentially related to the metabasites of the Pelagonian zone in the hanging wall of the local detachments. These isotopic and geochemical differences between Cape Evros and Panormos Bay correlate with the position of an inferred N-S–trending normal fault in the center of the island (Lecomte et al., 2010; Denèle et al., 2011).

On Antiparos, specifically Agios Georgios [7], a mixture of magmatic and seawater-derived fluids is suggested to explain the δ34S values of barite, whereas at Prassovounia [6] seawater was dominant. Low δ18O values of barite from Prassovounia, similar to the values of barite from Cape Evros on Mykonos, might suggest a contribution of meteoric water or indicate interaction of the hydrothermal fluid with the host rock. The nearby Pliocene rhyolite intrusion (Fig. 2) is a probable source of the magmatic fluid components at Agios Georgios, although Sr and Ba probably were also derived by leaching of the radiogenic underlying basement.

On Milos, the isotopic signatures of barite among the different deposits are remarkably similar, despite the different styles of mineralization from sea-floor massive sulfide [9] (hybrid epithermal to VMS) to subaerial epithermal [10]. Barite occurrences on Milos formed mainly from a seawater-dominated fluid with small variations in the 87Sr/86Sr related to variable fluid-rock interaction or mixing of seawater and hydrothermal fluids that equilibrated with different source rocks (metapelites of the Cycladic blueschist unit and Quaternary volcanic rocks). The vein mineralization at Katsimouti [8] is the only location where the S isotopes in barite indicate a significant magmatic fluid component.

The abundance of barite in the different deposit types of the Cyclades mineral district highlights a range of enriched sources for Ba (and Sr) and the presence of chloride-rich seawater in many of the hydrothermal systems capable of leaching Ba from the basement. As in other continental back-arc settings, the basement rock types are accessible through large-scale detachments and high-angle normal faults that provide pathways for leaching of crustal source rocks. Mixing of fluids from magmatic to surface-derived (seawater ± meteoric), highlighted by the strong isotopic and compositional zoning, was responsible for the widespread formation of barite in the different mineralizing systems.

We thank two anonymous reviewers for their helpful comments that have greatly improved the manuscript, and Sarah Gleeson and Larry Meinert for efficient editorial management. Financial support for this work was provided by Natural Sciences and Engineering Research Council (NSERC) Discovery grants to MDH and DAS, a Society of Economic Geologists Canada Foundation Student Research Grant to SCW, and additional support from the Helmholtz Centre for Ocean Research Kiel (GEOMAR). This project was additionally supported by the NSERC Collaborative Research and Training Experience program (iMAGE-CREATE) on Marine Geodynamics and Georesources. We thank Hercules Katsaros and Nikos Leloudas for their assistance with underground sampling in Lavrion. Gabriela Marshy is thanked for her great help as field assistant and Joëlle Dufour for her SEM analysis and EMPA of barite as part of her honors thesis. Glenn Poirier is thanked for his assistance with SEM analysis and EMPA, and Paul Middlestead and the team from the Ján Veizer Stable Isotope Laboratory for IRMS analyses and data reduction. SCW thanks Jeffrey Hedenquist for discussions about the fractionation of S isotopes in hydrothermal systems.

Sandra C. Wind is a Ph.D. candidate at the University of Ottawa, Canada, and is about to start a position as a research associate at Kiel University, Germany. She is strongly interested in geochemical and isotopic studies and focuses on the metal and fluid sources of the polymetallic deposits in the Cyclades, Greece. She completed her M.Sc. and B.Sc. degrees at Kiel University and gained further research experience by participating in seagoing expeditions and undertaking international research internships. She is passionate about educating the next generation of geologists and teaches undergraduate classes at Kiel University and the University of Ottawa.

Gold Open Access: This paper is published under the terms of the CC-BY-NC license.

Supplementary data