Tectono-metamorphic evolution of shallow crustal levels within active volcanic arcs. Insights from the exhumed Basal Complex of Basse-Terre (Guadeloupe, French West Indies)

Elucidating the structural and mineralogical transformations driven by heat and fluid transfers within the crust of magmatic arcs is an important yet tricky problem. It controls the formation of metal ores, particularly porphyry and epithermal ore deposits, and as well is vital for enabling the exploitation of geothermal resources (Sillitoe and Perelló, 2005; Rowland and Simmons, 2012; Chambefort et al., 2017). Such tectono-metamorphic evolutions have been frequently deciphered within ancient, inactive, magmatic arcs, whereas similar data from present-day active oceanic island volcanic arcs remain scarce, mainly because of bad outcrop conditions (Coombs, 1960; Miyashiro, 1967; Oxburgh and Turcotte, 1970; Ernst, 1973; Brown, 2010; Ducea et al., 2015). In the Lesser Antilles active arc, the Guadeloupe archipelago provides a natural laboratory for studying arc-related tectonic and metamorphic processes, first because Basse-Terre, i.e. the volcanic part of Guadeloupe Island, is entirely volcanic showing a north-south decreasing age gradient of its volcanic phases (Samper et al., 2007, 2009) where the northern part, so-called Basal Complex, consists of the oldest – Pleistocene – eruptive rocks (lava, block & ash flow tuffites) exposed on the island. Second, located at tropical latitudes, Basse-Terre of Guadeloupe is prone to significant weathering and erosion processes (Dessert et al., 2015) where mechanical and chemical erosion rates of around 1–2 mm/yr are effective (Sak et al., 2010, Lloret et al., 2011; Rad et al., 2013; Ricci et al., 2015a, b). In this context, the Basal Complex is a key target for deciphering the evolution of shallow crustal levels of the Lesser Antilles arc as it exposes the most eroded part of the Basse-Terre volcanic chain, consisting of the basement of the volcanic chain. Within this frame, we investigate the tectono-metamorphic evolution of this exhumed piece of the upper crustal level of the Lesser Antilles volcanic arc. We mapped and characterized the lithologies (geochemistry, geochronology) and the structures (structural and kinematic analysis) of the Basal Complex and propose an updated geological map of this northern area of Basse-Terre. Furthermore, we accurately document its metamorphic evolution by the means of a detailed study of the metamorphic reactions preserved in the exposed lava flows and thermodynamic modeling of the mineralogical equilibrium. This dataset allows us to discuss the heat and fluid transfers in the Lesser Antilles magmatic arc, so as to propose a conceptual scheme of these transfers and their influence on the strain accommodation and metamorphic evolution, in the framework of the shallow crustal level of an active volcanic arc.

The Lesser Antilles volcanic arc is the consequence of the southwest-directed subduction of the North American plate beneath the Caribbean plate (Hawkesworth and Powell, 1980; Demets et al., 2000, Fig. 1A). Since Eo–Oligocene times, two main magmatic arcs were produced through time with a progressive westward migration from an external to an internal volcanic arc, which is still active today (Briden et al., 1979; Westercamp, 1979; Bouysse et al., 1990; Maury et al., 1990; Dixon et al., 1998; Feuillet et al., 2002; Germa et al., 2011).

The Guadeloupe archipelago is formed by the juxtaposition of two geological systems. To the east, Guadeloupe consists of Grande-Terre, La Désirade, Petite-Terre and Marie-Galante Islands, on which early Pliocene-middle Pleistocene limestones were deposited upon volcanic and/or sedimentary crust (Bouysse and Garrabé, 1984; Andreieff et al., 1987; Bouysse et al., 1993; Cornée et al., 2012; Lardeaux et al., 2013; Münch et al., 2013, 2014). To the west, Guadeloupe is composed of Basse-Terre and Les Saintes Islands, which correspond to the inner active arc (Andreieff et al., 1987; Westercamp, 1988; Carlut et al., 2000; Mathieu et al., 2013; Zami et al., 2014; Verati et al., 2016). Basse-Terre consists of a cluster of volcanic complexes, frequently called “volcanic chains”, with from north to south (Samper et al., 2007): the oldest Basal Complex characterized by a 3 to 2 Ma aged volcanism, the Septentrional Chain between 1.8 to 1.1 Ma, the Axial Chain active from 1 to 0.4 Ma, the Grande Découverte – Trois Rivières Complex (0.2 Ma-recent) including the present-day active La Soufrière volcano (Boudon et al., 1987, 2008), and the Monts Caraïbes Chain built at around 0.6–0.4 Ma (Blanc, 1983; Ricci et al., 2017).

Several authors have focused their petrographic and geochemical studies on fresh volcanic rocks of the southern active volcanic part of the archipelago, and most of the studies have concentrated further south or near La Soufrière volcano or Bouillante geothermal field providing a revision of the ages of the volcanic formations and the lithological contours (Brown et al., 1977; Dagain, 1981; Blanc, 1983; Boudon et al., 1989; Samper et al., 2007, 2009; Patrier et al., 2013; Verati et al., 2014; Ricci et al., 2017). A recent study has shown the evolution of petrophysical properties from fresh to hydrothermalized volcanic rocks in the volcanic island of Guadeloupe archipelago (Navelot et al., 2018).

However, in the northern part of Basse-Terre, the Basal Complex is exposed and consists of strongly weathered volcanic formations (Mathieu et al., 2013). It has not been remapped since the late sixties (De Reynal de Saint-Michel, 1966), as mapping such area, where exposures are scarce and the vegetal coverage is dense, is challenging. Geophysical investigations provide a regional image of the Basal Complex appearing as the volcanic basement on which the younger volcanic systems have been emplaced (Gailler et al., 2013). Moreover, regional low-grade metamorphism and associated deformation has recently been evidenced in this complex (Verati et al., 2018). Although there is a general agreement to consider Basse-Terre as a cluster of volcanic complexes that become progressively younger southward, the identified boundary between the Basal Complex and the Northern Chain varies depending on previous studies and their interpretation of mapping (Westercamp and Tazieff, 1980; Blanc, 1983; Bouysse et al., 1990; Maury et al., 1990; Samper et al., 2007; Gailler et al., 2013). This limit is difficult to determine because exposure of outcrops is limited in northern Basse-Terre. Moreover, near the coastline, the attribution of the Deshaies eroded dome (so called “Gros Morne”) either to the Basal Complex or to the Northern Chain remains a topic of debate (Samper et al., 2007; Mathieu, 2010; Gailler et al., 2013; Ricci et al., 2015a; Barnoud et al., 2017; Verati et al., 2018; Fig. 1B). Thus, in order to overcome this ambiguity, we have investigated the age of this dome by the use of ⁴⁰Ar/³⁹Ar geochronology.

With this study, we improve the geological knowledge of the Basal Complex by mapping accurately this volcanic unit, we have distinguished the following formations (see Sect. 3.1 for a detailed description of the lithology):

• often very thick soils, which occupy a significant part (more than 50%) of the mapping area;

• superficial quaternary sedimentary formations;

• debris flows;

• lava flows.

This study presents a structural data set consisting of about 230 measurements of bedding, fractures, faults, schistosity, and kinematic criteria acquired over 12 sites across the North of Basse-Terre Island.

These data have allowed us to refine the map of this area, based on previous authors (De Reynal de Saint-Michel, 1966; Dagain et al., 1981).

We have inversed fault kinematics with the right dihedron method to obtain the paleo-stress tensor using WinTensor software (Angelier and Mechler, 1977; Angelier, 1979; Delvaux and Sperner, 2003). WinTensor allows rotational optimization of the obtained tensor by performing iterative tests of tensors in order to minimize a misfit function (Delvaux and Sperner, 2003). We have performed these inversions with a dataset consisting of fault strikes, dips, plunges of the striae and kinematic criteria. For consistency, a minimum of four structural measurements (faults direction and dip, pitch and kinematics) is needed to obtain a robust paleotensor. The quality of inversion is given by QRw (world stress map quality criteria) ranging from A (very good) to E (poor quality) depending of the amount of data use for the inversion (n), the ratio between the amount of data used for the inversion over the amount of input data (n/nt), deviation between observed and theoretical slip direction (α_w), the sense of slip confidence level (Cl_w) and the type of structure (fault, fracture, shear zone). The tensor quality is given by QRt, ranging from A (very good) to E (poor quality) and depending on QRw and the dispersion of the measured structures: more dispersed are the data, better is the quality Delvaux and Sperner (2003) (reported in the supplementary data in Appendix Tab. A1).

Grains of plagioclase were extracted from one fresh dacite lava flow sampled north of Deshaies. After magnetic separation (Frantz separator), approximately 200 grains aliquots of limpid plagioclase crystals (160–500 μm) were handpicked under a binocular microscope and were repeatedly ultrasonically cleaned in 2N HNO₃ and then in distilled water and methanol for a few minutes.

The samples were packed in aluminum foil for irradiation in the core of the Triga Mark II nuclear reactor of Pavia (Italy) with several aliquots of sanidine from Taylor Creek Rhyolite (TCR standard, 28.619 ± 0.034 Ma in Renne et al., 2010) as a flux monitor. The argon isotopic interferences on K and Ca were determined by the irradiation of KF and CaF₂ pure salts from which the following correction factors were obtained: (⁴⁰Ar/³⁹Ar)_K = 969 × 10⁻³ ± 3.6 × 10⁻⁵ at 1σ, (³⁸Ar/³⁹Ar) K = 1297 × 10⁻² ± 4.5 × 10⁻⁴ at 1σ, (³⁹Ar/³⁷Ar) Ca = 7.27 × 10⁻⁴ ± 4.0 × 10⁻⁵ at 1σ, and (³⁶Ar/³⁷Ar) Ca = 2.88 × 10⁻⁴ ± 1.6 × 10⁻⁵ at 1σ. ⁴⁰Ar/³⁹Ar step heating analyses were performed at Geosciences Montpellier (France). The gas extraction and purification lines consist of (a) an IRCO₂ laser of 100 kHz used at 3–20% power to heat the samples during 180 s, (b) a lens system for beam focusing, (c) a steel chamber, maintained at 10⁻⁸–10⁻⁹ bar, and (d) two Zr–Al getters for the purification of the gases. The argon isotopes were analyzed with ARGUS VI multicollector mass spectrometer. The mass discrimination was monitored daily using an automated air pipette and provided a mean value of 0.996817 ± 0.0013 (1σ) per dalton. Blank analyses were performed every three sample step. The raw data of each step and blank were processed, and the ages were calculated using the ArArCALC software (Koppers, 2002). Dating parameters (typical blank values, irradiation parameters and Ar–Ar analytical details) are reported in the supplementary data in Appendix (Tab. A2). The criteria for defining plateau ages are as follows: (1) Plateau steps should contain at least 50% of released ³⁹Ar, (2) there should be at least three successive steps in the plateau, and (3) the integrated age of the plateau should be in agreement with each apparent age of the plateau within a 2σ confidence interval.

In order to improve the definition of the lithotypes proposed on the geological map, the magmatic characters of the lavas flows were established. Whole rock major and trace element analyses were thus realized on several samples (analytical results are provided in the supplementary data in Appendix Tabs. A3 and A4). We carefully selected the rock exposure to ensure that the main types of lava flows were sampled. We analysed rock powders by Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES) and Inductively Coupled Mass Spectrometry (ICP-MS), for major and trace elements respectively, at the Geochemical and Petrographical Research Center in Nancy (SARM laboratory, CNRS-CRPG, with the procedure proposed by Carignan et al., 2001).

In order to constrain and quantify the petrological evolution of the Basal Complex, we investigated 80 thin sections. Mineral chemistry data were achieved by both electron microprobe analyses and diffractometry (system XPert-Pro MPD, Bragg Brentano geometry, Cu anticathode, KAlpha1 and KAlpha2 rays, PIXcel detector at the CEMEF in Sophia-Antipolis Scientific Center). Major elements analyses were performed at the “Service Commun de Microsonde” at Montpellier University on a Cameca-SX100 electron microprobe. The operating conditions were a 20 kV accelerating voltage and 10 nA beam current. The chemical analyses of the metamorphic phases are presented in the supplementary data in Appendix Table A5 (Schumacher, 1997).

We also used Raman spectrometry to complete mineral determination (in the supplementary data, Appendix A). The spectra were recorded with a LabRam HR800UV Jobin–Yvon^™ microspectrometer equipped with 1800 g/mm gratings and using 532 nm green laser excitation; acquisition time-span varied from 20 to 70 s during three accumulating cycles.

The Basal Complex is only found in the northern Basse-Terre, where the soils consist of vertisols, i.e. typical of old weathered volcanic rocks, with high kaolinite and montmorillonite contents (see Dessert et al., 2015 for a review). The superficial sedimentary formations consist of detrital deposits and alluviums. The debris flow deposits, interbedded with lava flows, contain blocks of contrasted sizes varying ranging between centimeter to meter of andesites and dacites as well as varicolored clay aggregates. In a restricted number of places, volcano-clastic deposits are interlayered within debris flows.

Lavas flows are easily observable along the coastline, and they extend inland up to 100 to 150 m above sea level, while eroded volcanic pitons and domes of 200 to 300 m high are still recognizable in the landscape. Along the coastal exposures, rocks show porphyric textures with different abundances (30–60%), of plagioclases, pyroxenes and/or oxydes phenocrysts. Columnar jointing of andesite and dacite lavas flows is a common volcanic structure that can be recognized at the regional scale. Lithological interfaces, stratigraphy or S₀, are easily observable within sequences of alternating lava flows and pyroclastic deposits. Lava flows fabrics, particularly well expressed in dacites and rhyodacites, precisely prismation (Fig. 2) and magmatic fluidality are also typical examples of primary volcanic structures.

In the Total Alkali versus Silica (TAS) diagram (Bas et al., 1986; Fig. 3A) SiO₂ contents range from 49 to 65 wt.% for lava flows and our sample’s compositions evolve from basalt to dacite. In the K₂O versus SiO₂ diagram (Peccerillo and Taylor, 1976 for arc volcanic lavas; Fig. 3B), the studied lavas fall within the fields of Median-K to Low-K series. In the SiO₂versus Nb/Y diagram (Winchester and Floyd, 1977; Fig. 3C), the studied samples belong to the calc-alkaline serie of basaltic-andesite to dacite. All together, these diagrams, classically used to define the chemical composition of volcanic rocks, show that the majority of the samples are andesites.

However, major and some trace elements, mainly Large Ion Lithophile Element (LILE) and Light Rare Earth Element (LREE), are severely mobile in weathered and/or hydrothermally altered lavas. Therefore, to improve the geochemical characterization of our samples, we selected diagrams based only on minor and trace elements considered as being the most immobile. Samples where LOI (loss on ignition) is below 1–2%, demonstrating that weathering is insignificant, are named “slightly altered” while samples with LOI above 3%, evidencing alteration, are named “altered” to “highly altered”. The Zr/Ti–Nb/Y diagram is classically employed for recognizing rock types even in altered rocks (Winchester and Floyd, 1977; Pearce, 1996), but in tropical regions due to extreme chemical weathering conditions, even Ti and Zr are mobile (Braun et al., 1993, 2005; Melfi et al., 1996; Nahon and Merino, 1996; Cornu et al., 1999; Ma et al., 2007). We therefore used the Th–Co diagram proposed by Hastie et al. (2007) for altered arc volcanic rocks, and in this diagram the studied rocks show typical characters of calc-alkaline volcanic series (Fig. 3D). Our samples, basaltic andesite to rhyodacite, thus followed a differentiated serie. All together, these geochemical data indicate that the Basal Complex lavas belong to the Medium K calc-alkaline series with the following more or less differentiated types: basaltic-andesites, andesites, dacites and rhyodacites. We used therefore this nomenclature in the proposed updated geological map (Fig. 4).

Moreover, we also used trace elements diagrams to compare our results to data from other volcanic rocks in Basse-Terre, as well as results obtained on the fresh rocks of the Basal Complex (Samper et al., 2007). The Rare-Earth-Element (REE) diagrams normalized to chondrite values (after Sun and McDonough, 1989; Fig. 5A) display relatively constant slopes, with small enrichment in Light REE (LREE) for most samples, typical for subduction related calc-alkaline REE chondrite-normalized patterns. The multi-element diagrams of incompatible elements normalized to primitive mantle (Sun and McDonough, 1989; Fig. 5E) display patterns typical of arc lavas, with Pb, U and Large Ion Lithophile Element (LILE) enrichments, and also Nb and Ta depletions (Labanieh, 2009).

In andesitic meta-basalts (Fig. 5B), samples display first significant HREE depleted patterns (with 80 to 110 times chondritic abundances for LREE and 20 times chondritic abundances for HREE), typically described after extensive weathering (Ma et al., 2007) and second slightly negative Eu anomaly which may be interpreted as plagioclase fractionation (Taylor and McLennan, 1995) or as hydrothermal alteration under mildly acidic low-temperature solutions. Finally, the multi-elements diagrams (Fig. 5F) show a strong Rb, Sr and K depletion, indicating the influence of surface alteration processes (Patino et al., 2003).

In meta-andesites (Fig. 5B), the observed patterns displays slightly LREE enrichment, and positives Eu, Ce and Gd anomalies attest for intense weathering (Ma et al., 2007). However, it has also been demonstrated that strong hydrothermal alteration may also lead to positive anomaly of Ce.

In meta-dacites (Fig. 5C), the observed trends display a flat-type or a very slight LREE enrichment with negative Ce anomaly for a few samples. This feature is interpreted as a post-magmatic process due to surface alteration processes under tropical weathering conditions (Cotten et al., 1995; Samper et al., 2009; Ricci et al., 2017). A sample from Vieux-Fort creek (BT5_01) shows LREE depletion and negative Eu anomaly reflecting high-temperature alteration under mildly acidic conditions (Bau, 1991). In multi-elements diagrams (Fig. 5G), our patterns are comparable to those previously described for the Basal Complex. Several samples show positive Y anomaly, in relation with surface alteration processes, specifically under tropical climates (Cotten et al., 1995; Labanieh, 2009).

In meta-rhyodacites (Fig. 5D), two types of trends can be highlighted. A first one, similar to the patterns previously recognized for the same lavas within the Basal Complex patterns. A second pattern, observed for Tillet creek samples (BT9_01-02), displays a flat-type trend with negative Ce and large positive Eu anomalies. This likely result from rock interaction with acidic high-temperature hydrothermal fluids (Michard et al., 1983; Michard, 1989; Cotten et al., 1995). Multi-elements diagrams also show two different patterns (Fig. 5H), both with Zr–Hf positive anomalies. If the first one is similar to the trends depicted in the Basal Complex volcanic rocks, another one characterizes the samples with strong negative La–Ce anomalies, negative Sr–Nd anomalies and always a large Eu anomaly.

All together, these data demonstrate that the volcanic rocks of the Basal Complex were severely transformed after their crystallization in relation with hydrothermal fluid circulations and supergene alterations.

Fortunately, even in significantly altered volcanic rocks, the primary volcanic mineralogy is, at least partly preserved, allowing the mineralogy of the mapped formations to be characterized. From a mineralogical point of view, the basaltic andesites and andesites display a volcanic primary association with phenocrysts of mafic plagioclase, clinopyroxene and ilmenite and/or titanomagnetite, in a decreasing order of abundance. Groundmass is typically microlitic and oxide-rich. In the andesites, the primary volcanic mineralogy is an association of phenocrysts of plagioclase, clinopyroxene, orthopyroxene and ilmenite, with rare amphiboles. The groundmass is characterized by the great quantity of microlites of plagioclase and oxides. Dacites and rhyodacites are composed of phenocrysts of plagioclase, orthopyroxene, clinopyroxene, ilmenite and quartz, while the groundmass is rich in quartz, plagioclase, oxides and anorthoclase microlites.

At a regional scale, lithological interfaces and lava flows fabrics are oriented in two main directions N 160°–170° and N 20°–30°, moderately dipping between 4° and 44°, with a mean dip of 20° towards West (Legendre, 2018)(Fig. 6B). Primary magmatic structures are overprinted by secondary tectonic structures. We identify three types of tectonic structures that developed during the progressive cooling of this volcanic complex through time and classify these as ductile structures, structures developed at the brittle-ductile transition and brittle structures.

At a mesoscopic scale the tectonic fabric is defined by smooth parallel cleavage areas separated by centimeter-scale undeformed zones (i.e. microlithons). Because cleavages are frequently interrupted at the scale of the outcrop (Fig. 7A), this tectonic fabric corresponds to pervasive disjunctive spaced cleavages (Powell, 1979; Passchier and Trouw, 2005). Spaced cleavage stripes (in the sense of Beach, 1982) are common and appear yellow-greenish to pink, with respect to the volcanic matrix, because of the crystallization of secondary minerals (Fig. 7). In some cases, the cleavage stripes turn into stylolithic joints (Fig. 7B). Moreover, seams are underlined by insoluble residues as well as volcanic enclaves (Fig. 7C) truncated by disjunctive schistosity planes (Fig. 7D), representing evidences for a pressure-solution processes driven strain (Fig. 7E; Rutter, 1976; Beach, 1979). Networks of quartz, albite, pyrite or calcite bearing veins frequently interconnect the schistosity planes. This geometry is comparable to the one described when dilatant micro-faults occur in jointed cohesive rocks (Kettermann et al., 2016). Thin sections also show that dissolution of phenocrysts is underlined by the development of dark and pink seams of insoluble minerals (Fig. 7F), mainly fine-grained Fe-oxides, at the grain boundaries. Regional finite strain pattern displays localized schistose domains surrounding lenses of undeformed country rocks (Fig. 2), preserving primary magmatic structures. Structural mapping evidences moderately dipping schistosity planes with a N 80°–100° main direction (Fig. 6A).

In a specific area, north of La Perle creek, we have identified “hydrothermal tectonic breccias” (Sibson, 1977; Jébrak, 1997; Woodcock and Mort, 2008). Along steeply dipping fault planes we observed textures of fragmentation containing large fragments of volcanic rocks. More than 60% of the fragments are larger than 2 cm in diameter giving rise to “mosaic breccias” according to the classifications of Jébrak (1997) or Woodcock and Mort (2008). These volcanic rocks fragments are cemented in a fine-grained matrix entirely composed of hydrothermal and strongly altered minerals. Thus these breccias are cohesive following Sibson’s (1977) textural classification. Most of the fragments are angular (Fig. 8B), a common feature for fluid assisted brecciation. However, some fragments, generally small-sized, are more rounded due to local dissolution processes (Fig. 8A). In a few cases, the spaced volcanic fragments are oriented parallel to the observed fault planes. All together, these features are compatible with deformation processes developed at the brittle to ductile transition. These hydrothermal tectonic breccias occur under two aspects that are easy to identify in the field by their colour: (1) red, within which some volcanic clasts are partly preserved in the core of hydrothermal secondary minerals (Figs. 8A, 8C and 8D), (2) yellow-green within which volcanic clasts are only phantoms replaced by clay minerals indicating an almost complete supergene alteration of the whole hydrothermal mineralogy (Figs. 8A–8C). Besides this difference in the degree of supergene alteration, these breccias show differences of orientation and two major directions were highlighted (Fig. 8E): N110-N140 and N160-N10.

The Basal Complex is also affected by brittle structures. The four families of regional directions are represented on the mapped area (Fig. 6A), with a main dominant fault family of N140-N150 (Fig. 6C) and with centimetrical offsets. Tectoglyphs along fault planes show normal dip-slip to oblique slip kinematics at all measurements sites, except at one locality where the offset of an andesitic pebble shows an apparent inverse kinematics, although the fault plane does not display any kinematic indications. At each site and although kinematics indicators were scarce, giving weak inversion data at individual sites, paleostress inversion shows a σ₃ trending NE-SW and NW-SE along N 50° and N 150° trending faults, respectively. Inversion of the whole dataset provides a pure-extensional stress regime with σ₃ trending NE-SW (Fig. 6D).

The trend of the southern boundary of the Basal Complex is a lineament at N 50° defined by a series of N 150° and N 50° faults, showing the same relationship with N 150° crosscutting N 50° faults.

The two plagioclase aliquots from the sampled dacite (aliquots A and B, Fig. 9, Tab. 1, Tab. A2) yield two concordant plateau ages of 4.10 ± 1.03 Ma (91.36% of ³⁹Ar released, MSWD = 0.49) and 4.42 ± 0.91 Ma (86.67% of ³⁹Ar released, MSWD = 0.51) respectively. The inverse isochron calculations yield ages at 4.05 ± 1.39 Ma (MSWD = 0.61; initial ⁴⁰Ar/³⁶Ar ratio at 301.09 ± 53.47) and 3.46 ± 1.92 Ma (MSWD = 0.20; initial ⁴⁰Ar/³⁶Ar ratio at 360.2 ± 111.0) respectively. The combination of aliquot analyses yields a plateau age at 4.28 ± 0.68 Ma (87.89% of ³⁹Ar released, MSWD = 0.48) and an inverse isochron age of 3.97 ± 1.03 Ma (MSWD = 0.46) with an initial ⁴⁰Ar/³⁶Ar ratio at 316.4 ± 48.4 indicating that the trapped ⁴⁰Ar/³⁶Ar is indistinguishable from atmospheric ⁴⁰Ar/³⁶Ar (Lee et al., 2006). However, the uncertainty of the plateau age is relatively high because of important atmospheric ⁴⁰Ar contribution with respect to radiogenic ⁴⁰Ar (16 to 81%, see in supplementary data). All single ages are concordant at the 2σ level and we retain the combined plateau age of 4.28 ± 0.68 Ma as the best estimate of the age of the dacite lava flow. This new ⁴⁰Ar–³⁹Ar dating demonstrates that the “Gros Morne” dacite dome belongs to the Basal Complex. This Ar–Ar age is older than recent previous obtained K–Ar ages for the Basal Complex (Samper et al., 2007). We sampled the Gros Morne volcanic dome located further west than the sampled lava flows dated in Samper et al. (2007) (Fig. 2). Similarly to these authors, we have selected only the rocks that seemed fresh enough in order to minimize the potential impacts of hydrothermalism and alteration on geochronological results in relation with K losses and Ar enrichments. Given that we considered previous K–Ar ages as significant, two different volcanic events may have occurred. Taking into account this new chronological constraint, south of the Deshaies dome, the Basal Complex is therefore clearly limited by a series of faults with N 150° and N 50° main directions. Its northeastern limit is also tectonic marked by mainly N 50° oriented faults. Far for the coastline, in areas where outcrops are of bad quality and thus in absence of field observations, a morphological limit, with Basal Complex formations covered by lava flows from the Northern Chain, is not excluded.

Mineralogical evidences for low-grade metamorphism have been recently discovered in the Basal Complex (Verati et al., 2018), however detailed investigations of the diversity of the mineralogical evolutions observable in the various volcanic lithologies still lack.

Under low-temperature conditions, metamorphic recrystallizations are incomplete and metamorphic phases coexist with metastable magmatic minerals (Frey and Robinson, 1999). It is the case in the studied rocks, which present systematic textural evidences of such chemical disequilibrium, particularly partial replacement of volcanic phases.

This peculiar metamorphism affects the whole Basal Complex and is identified by the systematic occurrence of post-magmatic reaction phases (Fig. 10).

In andesitic meta-basalts (Fig. 11A), secondary minerals are: quartz, albite, chlorite, actinolite, titanite, scarce white micas, prehnite, epidote and calcite. Chlorite occurs as millimeter patches within the groundmass or relicts of magmatic pyroxenes. Actinolite is present as acicular crystals around volcanic clinopyroxenes or in association with chlorites in the groundmass. Titanite is recognized as fine-grained brownish aggregates within relicts of plagioclase phenocrysts, while white mica occurs, in association with epidote and quartz, as replacement patches in volcanic phenocrysts. Prehnite occurs as colourless pseudomorphic patches or radially developed aggregates of low relief. These rocks are severely altered resulting in an argilic overprint (significant development of clay minerals) typical of alteration under tropical conditions.

In meta-andesites, chlorite, actinolite, albite, quartz, white mica, titanite are the ubiquitous metamorphic assemblage accompanied by variably developed epidote and oxides minerals. Chlorite colors range from pale green to various shades of yellow and brown. Actinolite commonly occurs as colorless crystals, sometimes in association with chlorite, at the rims of volcanic clinopyroxenes. Magmatic amphiboles are progressively transformed into a fringe of epitaxial acicular actinolites. Small sized titanite grains are associated with actinolite within reaction rims around pyroxenes. In some samples, coronas of metamorphic albite occur surrounding volcanic plagioclases, while in other rocks plagioclase phenocrysts are partially replaced by albite, white micas and calcite. Millimeter-sized grains of epidote are observable in the volcanic groundmass. Numerous cracks and veinlets are filled by calcite and tridymite (Fig. 12A) ± quartz ± albite ± chlorite. This assemblage must be interpreted as lower temperature assemblage developed during progressing cooling of the meta-andesite. These meta-andesites are also deeply altered giving rise to a typical assemblage of quartz, pyrite, calcite, magnesite, chalcedony (Fig. 12B), gypsum and clays minerals (kaolinite, illite, montmorillonite, smectite).

Comparable to what has been recently proposed by Verati et al. (2018), for meta-dacites (Fig. 11B) and meta-rhyodacites (Fig. 11C), the secondary metamorphic association is: albite, muscovite, chlorite, titanite, quartz, clinozoisite/epidote and ilmenite. White mica appears as irregular pseudomorphic patches, mainly rectangular, and close to quartz veins. We also observed sulphate, like barite, typical for low- temperature conditions and supergene alteration products with clays minerals (smectite and montmorillonite).

In some debris flow, metamorphic assemblages can also be identified (Fig. 11D): chlorite, white micas, quartz, actinolite, epidote, biotite and ilmenite. The meta-debris flows are also hydrothermally altered giving rise to a typical assemblage of pyrite, calcite, SiO₂ polymorphs (tridymite, chalcedony) and sulphate (jarosite) (Fig. 12C). In the southern part of Grande Anse beach, veins and fractures that cut the debris flows are filled by calcite and heulandite (Fig. 12D).

Together with their geochemical evolution (see trace elements evolutions presented in previous paragraphs), the observed mineralogical evolution reveals that the Basal Complex offers a spectacular example of allochemical metamorphism in relation with hydrothermal fluid circulation during cooling of the magmatic arc, ending with supergene alterations.

The mineral assemblages recognized within the meta-volcanites from the Basal Complex are a diagnostic for sub-greenschist facies conditions, close to the onset of the greenschist facies (Liou et al., 1985; Schiffman and Day, 1999; Robinson et al., 2005).

A first approach for a qualitative estimation of P–T conditions is the use of available petrogenetic grids. Such grids have been determined for metabasites and can thus be used only for the studied andesitic meta-basalts (Liou et al., 1985; Frey and Robinson, 1999). At the transition between sub-greenschist and greenschist facies, in the simplified basaltic system (i.e. NCMASH, Na₂O–CaO–MgO–Al₂O₃) system, an invariant point (so-called CHEPPAQ) is identified, which separates a higher-pressure mineral association (actinolite + chlorite + epidote + pumpellyite + quartz) from a lower-pressure assemblage (prehnite + epidote + chlorite + quartz). Depending on the selected activity models for thermodynamic modeling, the P–T position of the invariant point between prehnite bearing assemblages, pumpellyite bearing assemblages and greenschist facies assemblages varies between 3 and 4 kbar and between 300 and 380 °C (Liou et al., 1987; Frey et al., 1991; Powell et al., 1993; Beiersdorfer and Day, 1995; Day and Springer, 2005). Whatever the selected model, the metamorphic temperature for the mineralogical assemblages depicted in the studied samples is well constrained between 220 and 380 °C. For such a temperature range, the absence of clear pumpellyite + actinolite + chlorite + epidote assemblage suggests that metamorphic pressures did not exceed 2.5–3 kbar.

Thermobarometry is a second approach available and chlorite thermometry is particularly well adapted for low-grade metamorphic rocks, particularly using recent thermodynamic calibrations (Lanari et al., 2014; Bourdelle and Cathelineau, 2015). The application of these geothermometer to all the analyzed chlorites in both andesitic meta-basalts and meta-rhyodacites, with the assumption that Total Fe = Fe²⁺, indicates a temperature range of 230–320 °C, thus consistent with the previous temperature estimation. Pressure values can be roughly estimated by the application of the geobarometer of Brown (1977), to amphibole composition within andesitic meta-basalts, when this mineral is associated with albite, chlorite and iron oxide. With Na M4 values lower than 0.05 and Al^iv below 0.5 the estimated pressure is less than 2 kbar. Furthermore, the values of the K_D ^Mg–Fe for coexisting actinolite and chlorite range between 1.1 and 1.3, typical of metamorphic amphiboles under low-grade conditions (Robinson et al., 2004).

Furthermore, thermodynamic calculations were performed with the free energy minimization programs THERIAK and DOMINO (De Capitani and Petrakakis, 2010; updated software v. 4 February 2017) and the internally consistent thermodynamic “tcdb55c2d” database (Holland and Powell, 1998; updated Nov. 2003). However, thermodynamic modeling requires a strict definition of the chemical system to ensure it corresponds to the effective bulk chemistry equilibrated during the metamorphic peak event. The previously presented geochemical data demonstrate that a significant number of the studied samples were subject to severe chemical modifications during cooling and supergene alterations. REE patterns normalized to chondrite values (Fig. 5) allowed the identification of the samples that were not affected by late alterations and thus that best preserved the chemical system effective during metamorphic peak conditions.

Theses two samples (BT9_01-02) are meta-rhyodacites from Tillet creek (Fig. 5D). We calculated P–T pseudosections, at the whole rock scale with the metamorphic assemblage: albite + muscovite + chlorite + titanite + quartz + clinozoisite + ilmenite.

As a consequence of weathering under tropical conditions, the estimation of Fe³⁺ content in the studied samples is always questionable. We therefore used the values proposed for the Fe-oxidation state of sub-arc magmatism, proposed by Lee et al. (2010), and recently discussed and applied to meta-volcanites by Verati et al. (2018). Following these previous studies, a maximum of 5% of Fe³⁺ for the composition of the metamorphic equilibrium has to be envisaged and thus two chemical systems, with 0 and 5% of Fe³⁺ were simulated.

The two pseudosections were calculated in the system SiO₂–Al₂O₃–TiO₂–FeO–MgO–Na₂O–CaO–K₂O–H₂O, and assuming H₂O in excess (Fig. 13). In the system calculated with O (Fe₂O₃) = 0.29 corresponding to a 5% conversion of all Fe as FeO to Fe₂O₃ (Fig. 13A), minerals accommodating Fe³⁺, like epidote or riebeckite are largely stabilized. Whatever the Fe³⁺ content, with H₂O in excess in the system zeolites are particularly well developed, even under temperatures higher than 300 °C and pressures higher than 1.5 kbar (Figs. 13A and 13B). Moreover, there is no stability field that corresponds exactly to the observed mineralogical composition, as paragonite is systematically associated with muscovite. It is in the temperature range of 300–320 °C for pressure higher than 1.4 kbar for 5% of Fe₂O₃ and in the temperature range of 280–320 °C for pressure higher than 1.7 kbar for 0% of Fe₂O₃, that we can find the mineralogical association which is the closest to the one really observed, although two white micas are always present in the modeled pseudosections.

Therefore, it appears critical to test the H₂O impact on the stability field of the mineral assemblages. Using the Theriak-domino program, it is possible to generate a P-X(H₂O) equilibrium phase diagram at a fixed temperature. We choose a temperature of 280 °C in agreement with chlorite thermometry results and two equilibrium diagrams were calculated respectively with 5 and 0% Fe₂O₃ (Figs. 13C and 13D). With 5% of Fe₂O₃ in the system, riebeckite is still well developed and two white micas are always stable together (Fig. 13A). It is only when we consider a phase diagram with all Fe as FeO (0% Fe₂O₃) (Fig. 13D), that we obtain the expected mineralogy. The association albite + ilmenite + chlorite + muscovite + clinozoisite + titanite + quartz is indeed for 0.60 < X(H₂O) < 0.75 in the pressure range of 0.6–2.0 kbar. As demonstrated in Verati et al. (2018), in a totally different chemical system, the amount of H₂O available within the metamorphic fluids appears to be a key parameter for the stability of the observed metamorphic assemblage of Basse-Terre.

All together, available petrogenetic grids, thermobarometry and pseudosection modeling provide coherent estimation of the metamorphic P–T conditions suffered by the volcanic rocks of the Basal Complex in the range of 250–300 °C and 0.6–2.0 kbar. This metamorphic evolution is the result of a high-temperature and low-pressure regime typically developed within magmatic arcs setting.

Structural analysis allows the determination of the different directions of deformation through time during progressive cooling:

• N80-N100 for tectonic ductile structures;

• N110-N140 and N160-N10 for tectonic structures at the brittle-ductile transition;

• mainly N140-N150, and in a few cases N30-N50, N80-N100 and N160-N10, for tectonic brittle structures.

Dissolution-schistosity planes highlight ductile deformation. Pressure solution is now widely regarded as a mechanism accomplishing fluid-assisted ductile deformation under low-temperature conditions within the upper crust (Rutter, 1976; Gundersen et al., 2002; Gratier et al., 2013). However, strain is heterogeneous at all scales in the Basal Complex. Indeed, the regional finite strain pattern displays localized schistose domains surrounding lenses of undeformed country rocks, preserving primary magmatic structures. Structural mapping evidenced moderately dipping schistosity planes with a N 80°–100° main and exclusive direction (Fig. 6A). This structural pattern with sub-horizontal undeformed lenses strongly suggests a regional flattening stress regime, characterized by a component of vertical shortening.

During progressive cooling of the Basal Complex formations, brittle-ductile (hydrothermal tectonic breccias) and brittle structures (fractures, faults) develop with several observable directions. The change in stress regime remains unknown.

We did not observe systematic relations between the orientation of the breccias and their colours. Furthermore, we did not decipher relative chronological criteria of crosscutting relationships between both directions of breccias, as the stepping criteria are repeatedly reversed at the outcrop scale. If we take into account these various observations, it seems preferable to interpret these two structural directions as being synchronous rather than the result of two superimposed extensive events in time.

It is also difficult to define a strict chronology between the different brittle structures. This is explained by the limited observation of outcrops displaying kinematics or intersections at the scale of the Basal Complex. For this reason, we propose two hypotheses. Although we cannot rule out that the activation of theses faults may be synchronous, similarly to what we observed along semi-brittle structures, scarce outcrop-scale evidences suggest that the N140-N150 trending faults are crosscutting the N30-N50 trending faults. At the regional scale this observation is corroborated by the N 50° trending paleovalleys in which the St Rose Piton lavas have flowed and by the sharp N 150° boundary west of La Perle lava flows that may result from faulting along the directions N140-N150 (see Fig. 6A). The N140-N150 faulting event seems to postdate the N30-N50 one. However, many studies (Bouysse, 1979; Bouysse and Westercamp, 1990; Feuillet et al., 2002, 2004; Corsini et al., 2011; Lardeaux et al., 2013; De Min, 2014; Legendre, 2018) have shown that the deformed areas of the Guadeloupe archipelago have a long history and that active deformation is largely conditioned by the reactivation of structures, emphasising structural heritage. Following this hypothesis, structures with different orientations can be reactivated at the same time. Moreover, our paleostress inversion of the whole dataset of kinematics measurements gives a pure-extensional stress regime σ₃ trending NE-SW (Fig. 6D).

On the scale of the Guadeloupe archipelago with onshore and offshore investigations (Feuillet et al., 2002, 2010, 2011; Thinon et al., 2010; Mathieu et al., 2011; Corsini et al., 2011; Laigle et al., 2013; Lardeaux et al., 2013; Münch et al., 2013; Leclerc et al., 2014), statistically, the N150-N10 structures are presented as the most recent ones, while the N40-N50 structures are the oldest in the proposed relative chronology. According to many authors, the N120-N140 structures have been synchronously activated by transtension.

Finally, similar trends, observed in brittle, semi-brittle and ductile tectonic structures suggest strong inheritances within the crust of the volcanic arc of the Lesser Antilles and corroborates observations from previous workers along the Lesser Antilles volcanic arc (Corsini et al., 2011; Lardeaux et al., 2013; De Min et al., 2015; Legendre et al., 2018). In such a framework, the boundaries between volcanic complexes are possibly controlled by brittle normal faults, developed under recent to active extensional tectonic regimes, as it is the case for the geometric relation between the Basal Complex and the Northern Chain (Fig. 14).

The proposed new geological map shows a diversity of volcanic lithologies consistent with the magmatic differentiation of calc-alkaline series. Geochemical investigations allow the identification of a complete serie from basaltic-andesites to rhyodacites that has never been identified before. Even in the most altered parts of the Basal Complex, REE and multi-elements diagrams (Fig. 5) provide evidence of a triple geochemical signature: calc-alkaline magmatic differentiation, high-temperature transformation (metamorphism) and low-temperature sub-surface alteration.

These post-magmatic mineralogical transformations are the record of a progressive cooling of the Basal Complex through time and three main stages can be recognized:

• a high-temperature stage with albite, chlorite, white-micas, actinolite, biotite, prehnite, epidote, quartz, titanite, typical for sub-greenschist to greenschist facies conditions;

• a low-temperature re-equilibration stage under zeolite facies conditions with tridymite, chalcedony, calcite, quartz, chlorite, jarosite, heulandite and barite;

• and a sub-surface stage with development of clays minerals and gypsum typical for supergene alteration.

If we compare the calculated P–T conditions with the values of 70.0 to 98.2 ± 8.8 °C/km (conductive geotherm, Manga et al., 2012) measured for the present day geotherm near Basse-Terre, our data supports the idea that the development of this metamorphic pattern and the building of the active magmatic arc are synchronous. The temperatures reached during the Greenschist to sub-Greenschist facies metamorphic stage are high enough to lead significant chemical exchanges between contrasted rock types. In such a case potential K and Ar losses/enrichments, are still possible in dacite lavas. Consequently, the new age of 4.3 Ma obtained on the Deshaies eroded dome must be considered as a “maximal” age for the Basal Complex.

The mineralogical evolution, which reflects cooling (or exhumation), is consistent with the structural evolution, showing an evolution from ductile structures (controlled by dissolution-recrystallization process), to brittle-ductile (hydrothermal breccias) and brittle structures (fractures, faults). The Basal Complex is thus a key area to understand the thermo-mechanical evolution of the shallow crustal levels of active volcanic arcs. As a working hypothesis, we suggest that similar “arc-related metamorphism”, although recognized on a small area of northern Basse-Terre of Guadeloupe, is a common feature at the scale of the whole Lesser Antilles arc, and could now be recognized in other eroded zones of the recent or ancient volcanic arcs as in St. Kitts or Martinique for example.

Relationships between metamorphism and tectonics enable a conceptual model (Fig. 15) to be proposed including the depth of the brittle-ductile transition. If the conductive geothermal gradient is uniform (Manga et al., 2012; Verati et al., 2018), our observations can be extended to the whole island, and thus temperatures in the range of 250–300 °C possibly occur at a depth of 3–4 km. Consequently, the faults observed at the surface, more often strongly dipping, must be rooted on the brittle-ductile transition, and potentially tend to be more horizontal at that transition depth. This rheological interface is then potentially a preferential pathway for regional heat and fluid transfers, and in connection to the already known fault network and/or to volcanic structures (interfaces between lava flows, volcanic fabrics,…), offering a system of connected physical vertical and more horizontal discontinuities able to drive the hydrothermal fluids circulation at the whole scale of Basse-Terre arc crust. This result opens new exciting perspectives first for the exploration of geothermal resources, only known and already exploited in the Bouillante bay (Bouchot et al., 2010; Calcagno et al., 2012; Verati et al., 2014), and second in a more general way for the understanding of genesis of metallic ore deposits in volcanic arcs, following the idea that epithermal systems are comparable to geothermal systems (White & Hedenquist, 1995).

In the northern part of the island of Basse-Terre of Guadeloupe, the Basal Complex, including the Deshaies eroded dome, is composed of a well-differentiated calc-alkaline volcanic serie, with basaltic-andesites, andesites, dacites and rhyodacites. Its southern boundary corresponds to a series of faults of N 150° and N 50° main directions.

New ⁴⁰Ar–³⁹Ar dating on magmatic plagioclases demonstrates a magmatic activity with a significant time span, in the range of 4.3 to 2 Ma.

The Basal Complex offers a spectacular example of arc-related metamorphism in relation with a hydrothermal fluid circulation during the building and the cooling of the Lesser Antilles magmatic arc. Three main stages of post-magmatic mineralogical transformations record this evolution with: first a high-temperature stage under Greenschist to sub-Greenschist facies conditions (0.6–2 kbar for 250–300 °C), a re-equilibration stage under Zeolite facies conditions and finally a sub-surface alteration stage. This metamorphism is consistent with the present day geothermal gradient measured near the northwestern coast of Basse-Terre Island (Manga et al., 2012).

In parallel with this metamorphic evolution, a significant deformation history has been characterized. The finite strain pattern consists in the development and superposition through time of N80-N100 oriented schistose domains, driven by pressure-solution processes, of hydrothermal breccias associated to N110-N140 and N160-N10 faults, and finally of brittle normal faults, with dominant trends N140-N150.

The Basal Complex is an exhumed and eroded piece of the shallow crustal levels of the Lesser Antilles volcanic arc and thus represents a key-target for understanding its whole thermo-mechanical evolution.

The depicted tectono-metamorphic evolution is compatible with brittle-ductile transition located at around 3–4 km deep at the scale of the Basse-Terre Island. In such a framework, steeply dipping normal faults are probably rooted on this rheological interface. The connection of brittle faults to schistose domains provides connected vertical and horizontal physical discontinuities allowing heat and hydrothermal fluid transfers at shallow depth within the Guadeloupe archipelago.

This paper is a contribution to the GEOTREF program, funded by the French government with the program of “Investissements d’Avenir” and tutored by ADEME. We thank the researchers from the GEOTREF consortium for greatful discussions during the fieldwork campaigns in Guadeloupe and during scientific meetings. We sincerely would like to thank the reviews, the comments and the suggestions of Agnès Samper and Christoph von Hagke that significantly improved the manuscript. The authors thank Vivien Navelot for his donation of some samples and geochemical data, Matthieu Zanti for sampling preparation, Arthur Iemmolo for his technical support during ⁴⁰Ar/³⁹Ar analysis and Gabriel Monge for his welcome to CEMEF for the use of the X-ray Diffraction device.