What’s in the sandwich? New P–T constraints for the (U)HP nappe stack of southern Dora-Maira Massif (Western Alps)

The first discovery, more than 30 years ago, of coesite in continental crustal rocks (Chopin, 1984; Smith, 1984), demonstrated the possibility for continental crust to reach ultra-high pressure (UHP) conditions. However, the geodynamic processes responsible for the formation and exhumation of continental UHP units are still debated (e.g., Schenker et al., 2015; Reuber et al., 2016; Solarino et al., 2018). Conceptual and numerical models that try to explain how a continental crustal unit can reach (and can be exhumed from) UHP conditions are calibrated against geological and petrological data available from the tectonic nappe stack which includes the UHP unit itself (e.g., Li et al., 2010; Burov et al., 2014; Schmalholz et al., 2014; Gerya, 2015; Schenker et al., 2015). Precise knowledge of peak pressure and temperature (P–T) conditions experienced by both the UHP unit and its adjacent, often high-P (HP) units is therefore the crucial starting point to test the validity of the models (e.g., Manzotti et al., 2015, 2018).

The Brossasco-Isasca Unit (BIU) of the southern Dora-Maira Massif (DMM) in Western Alps, is one of the most studied UHP units worldwide and has been widely used in the past as a natural laboratory to discover new UHP minerals and to investigate metamorphic processes occurring during subduction at extreme pressures (e.g., Chopin & Ferraris, 2003; Schertl & O’Brien, 2013 and references therein). The models assuming subduction of continental crust to mantle depths (e.g., Chemenda et al., 1995; Chopin, 2003; Stöckhert & Gerya, 2005; Gerya & Stöckhert, 2006; Yamato et al., 2008; Butler et al., 2013) were proposed after the discovery of coesite in the BIU. Decades later, the alternative idea of tectonic overpressure explaining UHP tectonics arose again in the southern DMM (e.g., Ford et al., 2006; Schmalholz et al., 2014; Schmalholz & Duretz, 2015; Schenker et al., 2015). More than 30 years after its discovery, the interpretation of UHP metamorphism in the BIU still remains a challenge for many generations of geoscientists.

In the southern DMM, the UHP BIU is tectonically sandwiched between two quartz-eclogite facies units, the lower San Chiaffredo Unit (SCU) and the upper Rocca Solei Unit (RSU), which are in turn bounded by two blueschist-facies units, the Pinerolo Unit (PU) at the bottom and the Dronero-Sampeyre Unit (DSU) at the top of the tectonometamorphic package (Fig. 1). In contrast to the well-constrained P–T evolution of the BIU (e.g., Ferrando et al., 2017 and references therein), peak P–T conditions for its adjacent Units are poorly constrained, most studies dating back to over 20 years ago and mostly relying on conventional thermobarometric methods (e.g., Chopin et al., 1991; Michard et al., 1993; Compagnoni et al., 1995). Notably, many of these studies, which still represent milestones for all those interested in the interpretation of the tectonometamorphic evolution and architecture of the southern DMM, were carried out by Christian Chopin and co-workers. It is therefore a great pleasure for us to contribute to this Special Issue dedicated to Christian Chopin, with the aim of updating his legacy, taking advantage of the modern petrological methods and exploiting the great amount of data collected by our group over last decades.

With these premises in mind, this study constrains the prograde-to-peak evolution experienced by the HP units bounding the UHP BIU. For the first time, peak P–T conditions for all five units (PU, SCU, BIU, RSU, DSU) forming the southern DMM tectonic “sandwich” (Compagnoni & Rolfo, 2003) are estimated using the same, internally consistent and therefore comparable, modern thermobarometric approaches. The study focuses on metapelites (i.e., garnet-bearing phengitic micaschists) because they are widely exposed in all the investigated units and preserve the (U)HP assemblages well. Thermobarometric estimates are obtained combining multi-equilibrium thermobarometry (Average PT) and/or the P–T pseudosection approach applied to five samples (one for each unit). Concerning the HP units bounding the UHP BIU, our results confirm peak-T already known from the literature, but indicate systematically higher peak-P. The prograde path of the BIU is also significantly revised, with possible implications for the interpretation of burial mechanisms in the whole southern DMM tectonic “sandwich”.

In our opinion, these new data represent the inescapable starting point for any model aimed at a deeper understanding of the processes responsible for the attainment and preservation of UHP conditions in continental crust; however, further studies are needed in order to understand if (and how) the studied units shared part of their prograde evolution at HP to UHP conditions or if (and how) they were coupled during exhumation.

The Dora-Maira Massif (DMM) in the Western Alps, represents part of the subducted-exhumed European continental margin (e.g., Michard et al., 1996). It is now stacked in the Alpine orogenic wedge (Lardeaux et al., 2006), wherein it was overthrust by meta-ophiolite units of the Piedmont Zone (Balestro et al., 2018, and references therein). The southern portion of the DMM is a pile of imbricated thrust sheets, resulting from Alpine tectonic juxtaposition and metamorphic reworking of slices of Variscan continental crust and of Permian igneous bodies (e.g., Vialon, 1966; Henry, 1990; Chopin et al., 1991; Michard et al., 1993, 1995; Turello, 1993; Sandrone et al., 1993; Balestro et al., 1995; Compagnoni et al., 1995, 2012; Compagnoni & Rolfo, 1999, 2003; Groppo, 2002; Compagnoni et al., 2004, 2012; Botta, 2015; Nosenzo, 2018). Relics of both amphibolite-facies Variscan metamorphism and Permian contact metamorphism are locally still preserved (e.g., Biino & Compagnoni, 1992; Compagnoni et al., 1995; Compagnoni & Rolfo, 2003; Groppo et al., 2006, 2007b).

In the area between the Po Valley to the north and the Maira Valley to the south, the following tectonometamorphic units are distinguished from the lower to the upper structural position (Fig. 1): (i) the Pinerolo Unit (PU) is a monometamorphic unit that experienced Alpine epidote-blueschist facies metamorphism (e.g., Chopin et al., 1991; Avigad et al., 2003); (ii) the San Chiaffredo Unit (SCU) is a portion of pre-Alpine continental crust (Variscan amphibolite-facies basement intruded by Permian granitoids) that reached quartz-eclogite facies peak during Alpine metamorphism (Compagnoni & Rolfo, 2003); (iii) the Brossasco-Isasca Unit (BIU) is lithologically similar to the SCU but experienced Alpine peak P–T conditions in the coesite–eclogite facies (diamond stability field; e.g., Schertl et al., 1991; Nowlan et al., 2000; Rubatto & Hermann, 2001; Hermann, 2003; Ferraris et al., 2005; Di Vincenzo et al., 2006; Castelli et al., 2007; Groppo et al., 2007a; Ferrando et al., 2009; Gauthiez-Putallaz et al., 2016); (iv) the Rocca Solei Unit, lithologically similar to the BIU and the SCU, experienced Alpine peak P–T conditions at quartz-eclogite facies conditions (Chopin et al., 1991; Matsumoto & Hirajima, 2000); (v) the Dronero-Sampeyre Unit (DSU) includes polymetamorphic schists (Henry, 1990; Chopin et al., 1991) and a monometamorphic Permian volcano-detrital sequence (Vialon, 1966; Chopin et al., 1991; Michard et al., 1995), both of which experienced Alpine peak P–T conditions at epidote-blueschist facies (Chopin et al., 1991). The DSU is separated from the structurally underlying RSU by a shear zone, hundreds of metres thick, consisting of the tectonic juxtaposition of meta-ophiolite and metasediment slices (Valmala Shear Zone; i.e., the “ophiolitiferous band” of Henry et al., 1993). The term “tectonic sandwich” is hereafter used to indicate the whole package of juxtaposed tectonometamorphic units of southern DMM described above. All these units have been overprinted by a late-Alpine greenschist-facies recrystallization, which pervasively reworked and extensively obliterated the former (U)HP metamorphic mineral assemblages.

Sample locations are reported in Fig. 1 and additional details are given in the Supplementary material.

Qualitative major element X-ray maps of the entire thin sections were acquired using a micro-XRF Eagle III–XPL spectrometer equipped with an EDS Si(Li) detector and with an EdaxVision32 microanalytical system at the Department of Earth Sciences, University of Torino. Operating conditions were: 100 ms counting time, 40 kV accelerating voltage, and a probe current of 900 μA. A spatial resolution of about 65 μm in both x and y directions was used. Quantitative modal amounts of each mineral phase were obtained by processing the maps with the software Petromod (Cossio et al., 2002). For each sample, the processed X-ray maps are reported in Fig. 2. Additional microstructural details, at greater magnification, are given in Figs. 3–5. Mineral abbreviations are after Whitney & Evans (2010), except white mica (Wm) and phengite (Phe).

The rock-forming minerals were analyzed with a Jeol JSMIT300LV Scanning Electron Microscope at the Department of Earth Sciences, University of Torino. The instrument was equipped with an energy dispersive spectrometry (EDS) Energy 200 system and an SDD X-Act3 detector (Oxford Inca Energy). Operating conditions were: 50 s counting time, 15 kV accelerating voltage, spot size 2 μm. SEM-EDS quantitative data were acquired and processed using the Microanalysis Suite Issue 12, INCA Suite version 4.01; natural mineral standards were used to calibrate the raw data; the ϕρZ correction (Pouchou & Pichoir, 1988) was applied. Tables SM1 and SM2–SM6, freely available online as Supplementary Material linked to this article on the GSW website of the journal, https://pubs.geoscienceworld.org/eurjmin/, summarize the compositional variations (Table SM1) and the representative chemical compositions for the main minerals in each sample; Figs. 6 and 7 and SM1 show the composition of garnet and phengite. For practical purposes, the term phengite is used in the following to indicate phengitic white mica with Si > 3.20 apfu (on the basis of 11 oxygens), whereas muscovite is used for phengitic white mica with Si < 3.20 apfu.

The pseudosection modelling approach was applied to each metapelite sample, excluding DM1565. Bulk-rock composition of each sample (Table 1) was calculated by combining estimated mineral modes with mineral chemistry (see above). Determination of an effective bulk composition for sample DM1565 (DSU) was problematic because: (i) the fine-grained nature of Alpine Grt₂ hampered its correct quantification from micro-XRF maps; (ii) the presence of pre-Alpine garnet porphyroblasts (Grt₁), that should be subtracted from the bulk-composition effectively reacting during Alpine metamorphism, makes the conventional bulk-rock analysis (e.g., XRF or ICP-MS) not suitable.

For each considered sample, three P–T isochemical phase diagrams have been calculated (i.e., a total of 12 pseudosections) considering the fractionation effects on the bulk-composition due to the growth of zoned garnet porphyroblasts. Each pseudosection was used to model: (i) P–T conditions for garnet core growth (measured bulk-composition: MBC), (ii) P–T conditions for garnet mantle growth (MBC−garnet cores), and (iii) P–T conditions for garnet rim growth (MBC−garnet cores + mantles) (Table 1).

The P–T isochemical phase diagrams were calculated in the system MnNKCFMASTOH (MnO–Na₂O–K₂O–CaO–FeO–MgO–Al₂O₃–SiO₂–TiO₂–Fe₂O₃–H₂O), except for sample DM1281 (BIU), for which Fe³⁺ was neglected, because Fe³⁺-rich oxides are absent and the amount of Fe³⁺ in the analyzed minerals is very low. The pseudosections were calculated using Perplex 6.8.1 (version March 2018; Connolly, 1990, 2005, 2009), the internally consistent thermodynamic dataset of Holland & Powell (2011) (ds62) and the equation of state for H₂O of Holland & Powell (1998). The following solution models were used: garnet, chlorite, chloritoid, staurolite, white mica, biotite and ilmenite (White et al., 2014), omphacite (Green et al., 2007), amphibole (Green et al., 2016), carpholite (Smye et al., 2010), feldspar (Fuhrman & Lindsley, 1988) and epidote (Holland & Powell, 2011). Quartz, lawsonite, kyanite, rutile and titanite were considered as pure phases. The fluid was usually considered as pure H₂O (aH₂O = 1), but a reduced aH₂O = 0.4 was considered in the pseudosection that models the garnet rim growth in DM1281 (BIU) because experimental and fluid-inclusions studies (Sharp et al., 1993; Hermann, 2003; Ferrando et al., 2009) demonstrated that, at the UHP peak, the activity of H₂O was strongly reduced by the abundance of dissolved species.

The THERMOCALC “Average PT” (AvPT) method (i.e., “Optimal thermobarometry”: Powell & Holland, 1994) was applied to all the samples. THERMOCALC v3.40 and the Holland & Powell (2011) dataset (ds62) were used. Activity-composition relationships were calculated using the software AX. The method was applied to the peak assemblage for all the samples, except sample DM1281 (BIU), for which it was applied to the late prograde assemblage (see Section 4.3). Used sets of mineral compositions and of independent reactions are given in Tables SM2–SM7. AvPT results are discussed in terms of weighted means (i.e., values with smaller errors contribute more than values with larger errors) and are presented in Table 2.

This medium- to coarse-grained micaschist consists of white mica (mostly phengite; 42%), garnet (20%), chlorite (19%), quartz (13%) and minor chloritoid (2%) (Fig. 2). Retrograde chlorite, plagioclase (3%) and minor biotite, as well as accessory magnetite, rutile and ilmenite also occur. The main foliation (S_m) is defined by preferred orientation of phengite and chlorite, that form continuous pluri-mm -thick layers alternated with discontinuous pluri-mm thick quartzitic layers (Fig. 2). An earlier schistosity (S_m−1), defined by the preferred orientation of white mica and chlorite, is still preserved in microlithons.

Garnet porphyroblasts, up to several millimetres in size, are syn-kinematic with respect to S_m and show a snow-ball structure (Fig. 3a); the rotated internal foliation, continuous with the external S_m, is defined by inclusions of quartz, relict chloritoid, and minor white mica and chlorite (Fig. 3b). Quartz, phengite and chlorite occur in the pressure shadows of garnet. Garnet porphyroblasts are strongly zoned (Fig. 6a and Table SM1), with a reddish core and a pinkish rim.

Phengite defines the S_m and rarely occurs, in association with paragonite and relict chloritoid, in polymineralic inclusions within garnet core and mantle (Fig. 3b). A few retrograde muscovite flakes occur in matrix and in polymineralic inclusions within garnet. Paragonite is only observed in polymineralic inclusions in garnet, always associated with (or replacing) chloritoid; its prograde vs. retrograde nature is ambiguous. Bluish–greenish chloritoid is only preserved as inclusions in garnet porphyroblasts (Fig. 3a and b), but not in the matrix. Its composition changes according to the microstructural position within garnet (Table SM1). Chlorite is mostly in equilibrium with S_m, but a late chlorite generation replaces garnet along fractures. Chlorite rim is locally partially replaced by fine-grained biotite. Plagioclase (albite to oligoclase) occurs as large (pluri-mm) blasts overgrowing the S_m and it is thus interpreted as a retrograde phase.

Among the accessories, retrograde mm-sized magnetite exclusively occurs in the matrix (Fig. 2). Rutile and ilmenite within garnet core, mantle and rim appear in equilibrium, whereas in the matrix rutile is rimmed by ilmenite.

This garnet-bearing phengitic micaschist has a strongly heterogeneous grain size. It shows a mm-thick banded structure, with continuous medium- to coarse-grained phengite-rich layers alternating with fine- to medium-grained discontinuous quartz-rich layers (Fig. 2). It consists of white mica (mostly phengite: 42%), garnet (30%), quartz (18%) and minor chloritoid (1%). Retrograde chlorite (7%), epidote (1%), minor biotite, and accessory magnetite, rutile, ilmenite, tourmaline and apatite also occur. The main foliation (S_m) is defined by the preferred orientation of white mica.

Garnet porphyroblasts are centimetres to pluri-cm in size in the coarse-grained layers, and millimetres to pluri-mm in the finer-grained layers (Fig. 2). Garnet is pre- to syn-kinematic with respect to the S_m, being mostly enveloped by the main schistosity, but with the rim that appears in equilibrium with it. Garnet porphyroblasts are microstructurally and chemically zoned (Fig. 6b). Garnet core and mantle are large, often asymmetric, and crowded with inclusions, whereas rim is thin and with few inclusions. Fine-grained inclusions in garnet core consist of epidote (Fig. 3d), chlorite and chloritoid; inclusions in garnet mantle are medium-grained and mostly consist of chloritoid (Fig. 3c) and white mica. Garnet rim includes few phengite flakes and box-shaped pseudomorphs after former lawsonite, consisting of a fine-grained aggregate of epidote, muscovite, paragonite and chlorite (Fig. 3e). A thin (100–300 μm) and discontinuous outer rim slightly enriched in Mn has been interpreted as due to local retrograde re-equilibration.

Phengite defines the S_m and is locally included in garnet. Less abundant retrograde muscovite overgrows the main foliation and replaces chloritoid inclusions within garnet. Paragonite only occurs as inclusion in garnet, associated with (or replacing) chloritoid, and as pseudomorphs after probable lawsonite; therefore, it has been interpreted as both prograde and retrograde phase. Chloritoid is mostly included in garnet mantle (Fig. 3c), rarely in core and rim, and it is absent from the matrix. It is locally partially to completely replaced by an aggregate of white mica (muscovite ± phengite + paragonite) (Fig. 3c).

Epidote and chlorite occur in two different generations. Prograde epidote (often with an allanitic core) and chlorite are included in garnet core and mantle (Fig. 3d). Retrograde epidote occurs as few fine-grained idioblasts in matrix and in pseudomorphs after lawsonite included in garnet rim, whereas late chlorite mostly replaces garnet porphyroblasts along fractures and rims. Late biotite partially replaces white mica and chlorite at their rims.

Among accessories, mm-sized retrograde magnetite only occurs in the matrix. Rutile and ilmenite occur as inclusions within garnet porphyroblasts; in the matrix rutile is rimmed by ilmenite. Apatite and tourmaline are included in garnet.

This sample consists of white mica (mostly phengite, 41%), quartz (24%), garnet (21%), kyanite (10%), jadeite (3%) and minor chloritoid, staurolite and chlorite (Fig. 2). Albite and pyrophyllite are retrograde phases, whereas rutile, tourmaline and apatite are the main accessory minerals. The grain size is markedly heterogeneous, with pluri-mm garnet and kyanite porphyroblasts set in a fine-grained matrix mostly consisting of phengite and quartz (Fig. 2). The main schistosity (S_m) is defined by mm-thick, continuous phengitic layers, alternating with discontinuous quartzitic layers of similar thickness.

Garnet porphyroblasts are microstructurally and chemically zoned (Fig. 6c). Garnet core and mantle are large, often asymmetric, and crowded with inclusions, whereas rim is thin and with few inclusions (Fig. 4a). Garnet rim appears in equilibrium with S_m. Inclusions in garnet core are fine-grained and polymineralic; garnet mantle includes both fine-grained polymineralic inclusions and medium-grained monocrystalline inclusions; garnet rim contains few medium-grained monocrystalline inclusions. Polymineralic inclusions in garnet core mostly consist of chloritoid, paragonite, chlorite, staurolite and rare muscovite, whereas those observed in garnet mantle contain also kyanite and does not contain potassic white mica (Fig. 4d and e). Monocrystalline inclusions in garnet mantle are represented by pseudomorphs after jadeite and glaucophane (Fig. 4f and g), kyanite and quartz, whereas those in garnet rim consist of kyanite, phengite and quartz. Polycrystalline aggregates of quartz surrounded by radial cracks, interpreted as deriving from inversion of coesite, are located at the transition between garnet mantle and rim (Fig. 4a).

Phengite defines the S_m and is locally included in garnet rim. Retrograde muscovite is scarce and its occurrence is limited to the pseudomorphs after jadeite and glaucophane. A few prograde muscovite flakes are preserved in the polymineralic inclusions within garnet core. Paragonite is common in the polymineralic inclusions hosted within garnet core and mantle (Fig. 4e), where it probably has a prograde nature.

At least three different generations of kyanite have been recognized: (i) large (pluri-mm) porphyroblasts enveloped by the main foliation (Fig. 4b); (ii) smaller idioblasts in equilibrium with the S_m (Fig. 4b); (iii) local very fine-grained acicular crystals rimming the large kyanite porphyroblasts. Jadeite and glaucophane are not preserved, but are pseudomorphically replaced by retrograde phases. Pseudomorphs after jadeite, occurring both in the matrix (Fig. 4c) and as inclusion in garnet mantle (Fig. 4g), consist of a very fine-grained aggregate of albite + pyrophyllite + muscovite. Lozenge-shaped pseudomorphs after glaucophane have been observed only as inclusions in garnet mantle (Fig. 4f) and consist of a very fine-grained aggregate of paragonite + muscovite + minor biotite.

Chloritoid, chlorite and staurolite only occur in the fine-grained polymineralic inclusions within garnet core and mantle (Fig. 4d and e). Chlorite is a prograde phase and the X_Mg of the chlorite included in garnet mantle is higher than that of the chlorite included in garnet core (Table SM1). Chloritoid is generally associated with paragonite and chlorite. Staurolite is only present as inclusion in both garnet core and mantle, where it mostly occurs in association with kyanite, paragonite and chlorite, more rarely with chloritoid. Microstructural relationships clearly show that staurolite grew at the expense of kyanite, often preserved as rounded relics partially rimmed by staurolite (Fig. 4e).

Accessory rutile, tourmaline and apatite occur in the matrix and as inclusion in garnet.

This fine- to coarse-grained, garnet-bearing phengitic micaschist consists of white mica (mostly phengite, 49% and paragonite, 6%), quartz (20%), chloritoid (12%) and garnet (8%) (Fig. 2), with minor amounts of glaucophane and jadeite (almost completely replaced by retrograde phases). Retrograde chlorite (3%), epidote (1%) and biotite, and accessory rutile, ilmenite, magnetite and apatite also occur. The main foliation (S_m) is defined by mm- to pluri-mm-thick, continuous, micaceous layers alternated with mm-thick, discontinuous, quartzitic layers. Relics of an earlier S_m−1 schistosity are locally preserved in microlithons and are defined by the alignment of white mica.

Millimetre (mm)- to pluri-mm-sized garnet porphyroblasts are partially enveloped by the S_m, but their rim appears in equilibrium with S_m (i.e., garnet is pre- to syn-kinematic with respect to S_m). They include an internal rotated foliation mostly defined by quartz and chloritoid. Garnet core and mantle are large and rich of inclusions, whereas garnet rim is thin and with few inclusions (Fig. 5a–c). Inclusions in garnet are either polymineralic or monocrystalline. Polymineralic inclusions mostly consist of chloritoid + white micas (phengite ± paragonite ± muscovite) (Fig. 5b). Relics of glaucophane + jadeite, partially replaced by paragonite + albite ± muscovite, have been observed at the transition between garnet mantle and garnet rim (Fig. 5b). Garnet core and mantle locally include tabular-shaped aggregates of epidote + quartz + paragonite or phengite (Fig. 5c), possibly representing pseudomorphs after former lawsonite. Monocrystalline inclusions mostly consist of quartz, chloritoid and minor phengite.

Phengite defines the S_m and is locally included in garnet and chloritoid. Retrograde paragonite and muscovite occur both in the matrix and as inclusions in garnet. In the matrix, paragonite forms large flakes statically overgrowing the S_m (Fig. 2). When included in garnet, paragonite and muscovite generally replace other phases, such as glaucophane, jadeite or chloritoid. Chloritoid occurs both in the matrix and as inclusion in garnet (Fig. 5a). In the matrix, it forms medium-grained bluish–greenish nematoblasts aligned with the S_m and it is slightly zoned. Rare glaucophane and jadeite relics are included in garnet mantle/rim (Fig. 5b).

Epidote, chlorite, albite and biotite are retrograde phases. Epidote within garnet occurs in tabular-shaped aggregates, in association with quartz and white micas, interpreted as deriving from former lawsonite (Fig. 5c). In the matrix, it forms few fine-grained idioblasts, often with an allanitic core. Aggregates of chlorite statically overgrow the S_m and replace garnet along fractures and at the rim. Its rim is locally replaced by biotite. Albite occurs as replacement of jadeite and glaucophane.

Rutile mostly occurs as inclusion in garnet, whereas it is pervasively replaced by ilmenite in the matrix. Retrograde magnetite is fine-grained and widespread in the matrix, and it is strongly oxidized. Apatite occurs as relatively large grains both in the matrix and included in garnet.

This garnet-bearing phengitic micaschist consists of white mica (mostly phengite, 54% and paragonite, 8%), quartz (26%), garnet (5%) and chloritoid (3%) (Fig. 2). Retrograde chlorite (3%) and biotite and accessory rutile, ilmenite and tourmaline also occur. The main schistosity (S_m) is defined by pluri-mm thick, continuous, micaceous layers alternating with mm-thick quartzitic layers. Large (pluri-mm) garnet porphyroblasts are preferentially set in the fine-grained micaceous layers (Fig. 2).

In contrast to the other samples, this metapelite preserves mineralogical evidence of a polymetamorphic evolution. Two different garnet generations can be, in fact, recognized: (i) large porphyroblasts (Fig. 5d) enveloped by the S_m, interpreted as amphibolite-facies Variscan garnets (Grt₁); (ii) small (ca. 100 μm) idioblasts in equilibrium with S_m and a thin and discontinuous corona around porphyroblastic Grt₁ (Fig. 5e), both interpreted as Alpine garnets (Grt₂). The strongly-fractured Grt₁ is chemically zoned (Table SM1). Locally it is poikiloblastic and includes quartz, white mica, ilmenite and minor staurolite, now replaced by fine-grained pseudomorphic aggregates of white mica + chloritoid. The small Grt₂ idioblasts are also zoned and include very fine-grained quartz (Fig. 6e and Table SM1).

Phengite defines the S_m and it also occurs at the rim of the chloritoid-bearing pseudomorphs after former staurolite. Minor retrograde muscovite and paragonite occur in the matrix. Chloritoid mostly constitutes large (pluri-mm) fine-grained aggregates enveloped by the S_m and interpreted as pseudomorphs after pre-Alpine porphyroblastic staurolite (Fig. 5f) (see also Henry, 1990; Chopin et al., 1991). These pseudomorphs also contain white mica (phengite + muscovite + paragonite), generally concentrated toward the rim. Chloritoid also forms minor idioblasts aligned with the S_m or partially overgrowing it.

Retrograde chlorite occurs as large flakes statically overgrowing the S_m and partially replacing garnet at the rim and along fractures. Biotite is limited to late and discontinuous shear bands crosscutting the main foliation. Accessory ilmenite is included in pre-Alpine garnet (Grt₁) and is rimmed by rutile, if it communicates with the external matrix through fractures. In the matrix, rutile is rimmed by ilmenite.

The prograde-to-peak P–T evolution of the PU, SCU, BIU and RSU samples was constrained using the pseudosection approach and on the basis of the predicted stability fields of the observed mineral assemblages, combined with garnet, chloritoid and phengite compositional isopleths. The results of thermodynamic modelling show that the fractionation effects on the bulk-composition due to the growth of zoned garnet porphyroblasts are generally minor during the initial growth of garnet (i.e., after the fractionation of garnet core). However, they become significant after the fractionation of both garnet core and mantle. The general topology of the calculated pseudosections for all samples is similar as far as the relationships between the main phases of interest is concerned, and is in line with previous studies related on phase equilibria modelling in the same system (e.g., Guiraud et al., 1990; Proyer, 2003; Wei & Powell, 2004, 2006; Smye et al., 2010; Manzotti et al., 2015): (i) garnet is always stable over a large range of P–T conditions, whereas chlorite is limited to P < 18–25 kbar; (ii) chloritoid is stable at T < 550–600 °C, and kyanite is instead stable at increasing temperature (see Smye et al., 2010); (iii) paragonite, glaucophane and jadeite represent the Na-rich phases stable at lower, intermediate and higher pressure, respectively.

The Average PT approach of THERMOCALC was further applied to peak assemblages for the PU, SCU, RSU and DSU samples and to the late prograde assemblage for the BIU sample. In the following, the results of both approaches are summarized.

The observed prograde and peak assemblage (Qz + Phe + Grt + Chl + Cld + Rt + Ilm) is not modelled in the P–T region of interest because chloritoid is always predicted to be stable with a Na-rich phase (either paragonite or glaucophane), which was not observed in the studied sample. However, the occurrence of large oligoclase porphyroblasts in the matrix statically overgrowing the main foliation suggests that one or more Na-rich phases (e.g. glaucophane, paragonite) was likely stable in the prograde and peak assemblages and it is not preserved. The glaucophane/paragonite-bearing assemblage (Qz + Phe + Grt + Chl + Cld + Gln/Pg + Rt + Ilm) is modelled in a wide range of P–T conditions (450–570 °C, 10–25 kbar) (Fig. 8).

Prograde and peak P–T conditions are more tightly constrained by the intersection of compositional isopleths. Specifically: (i) early prograde P–T conditions are constrained at 480–495 °C, 19–21 kbar using compositions of garnet core (Sps₅Prp₆Grs₈Alm₈₀) and of its inclusions of chloritoid (X_Mg = 0.11) and phengite (Si = 3.48 apfu) (Fig. 8a); (ii) late prograde P–T conditions are constrained at 490–505 °C, 20–22 kbar using the compositions of garnet mantle (Sps_2.5Prp₈Grs_7.5Alm₈₁) and of its inclusions of chloritoid (X_Mg = 0.15) and phengite (Si = 3.44–3.45 apfu) (Fig. 8b); (iii) peak P–T conditions are estimated at 500–515 °C, 20–23 kbar using the compositions of garnet rim (Sps_0.5Prp₁₁Grs₁₁Alm₇₈), of its chloritoid inclusions (X_Mg = 0.18) and of matrix phengite (Si = 3.38–3.51 apfu) (Fig. 8c).

The AvPT approach applied to four different sets of mineral compositions gives peak P–T conditions of 496 ± 7 °C, 19.0 ± 0.7 kbar (Table 2), consistent with the results of pseudosection modelling (Fig. 8c).

Although glaucophane was not observed in prograde and peak assemblages, the modelled pseudosections predict its stability over a large P–T interval. However, the predicted modal amount of glaucophane is so low (<5 vol%) that its (former) occurrence could likely have been overlooked. Ignoring glaucophane, the observed early prograde (Qz + Phe + Grt_C + Chl + Cld + Ep + Rt + Ilm) and late prograde (Qz + Phe + Grt_M + Cld ± Chl ± Ep/Lws + Rt + Ilm) assemblages are modelled by large tri- and quadri-variant fields at 420–500 °C, 12–17 kbar and 400–500 °C, 14–25 kbar, respectively (Fig. 9). Garnet, chloritoid and phengite compositional isopleths allow to further constrain the prograde P–T conditions: (i) early prograde conditions are constrained at 460–470 °C, 17–19 kbar using the compositions of garnet core (Sps₅Prp_3.5Grs₂₆Alm₆₅) and its inclusions of phengite (Si = 3.34–3.46 apfu) (Fig. 9a); (ii) late prograde P–T conditions are constrained at 480–490 °C, 19–21 kbar using the compositions of garnet mantle (Sps_2.5Prp₅Grs₂₄Alm₆₈) and of its inclusions of chloritoid (X_Mg = 0.20) and phengite (Si = 3.40–3.49 apfu) (Fig. 9b). Although garnet core includes epidote, its growth is predicted to occur within the lawsonite stability field, but close to the Ep-out boundary.

The observed peak assemblage (Qz + Phe + Grt_R + Cld + Lws + Rt + Ilm) is modelled by a relatively small quadri-variant field at 450–520°, 22–27 kbar. The intersection of compositional isopleths for garnet rim (Sps_0.8Prp₁₀Grs₁₈Alm₇₄) and its inclusions of chloritoid (X_Mg = 0.20) and phengite (Si = 3.36–3.43 apfu) defines peak P–T conditions at 500–520 °C, 21–24 kbar (Fig. 9c). The AvPT approach applied to four different sets of mineral compositions gives peak P–T conditions of 512 ± 8 °C, 21.9 ± 0.9 kbar (Table 2), consistent with the results of pseudosection modelling (Fig. 9c).

Garnet core and its inclusions define the early prograde assemblage (Qz + Phe + Grt_C + Cld + Pg + Chl + Rt). Staurolite is not considered as part of this assemblage due to its supposed retrograde nature (see microstructural relationships presented in Section 4.3). This assemblage is modelled by a quadri-variant field at 450–570 °C, 10–18 kbar. The best fit between observed and modelled garnet (Sps_6.5Prp₁₂Grs₆Alm₇₄), chloritoid (X_Mg = 0.25) and chlorite (X_Mg = 0.62) compositions constrains P–T conditions for the growth of garnet core at 520–540 °C, 16–23 kbar (Fig. 10a), in the Qz + Phe + Grt + Cld + Pg ± Chl ± Gln ± Rt fields. These P–T conditions are at significantly lower T and higher P than the predicted stability field of staurolite, further confirming that staurolite is not compatible with the early prograde assemblage (see Section 4.3).

The late prograde assemblage consists of garnet mantle and its inclusions (Qz/Coe + Phe + Grt_M + Cld + Pg + Ky + Jd + Gln + Rt); once again staurolite is not considered as part of this assemblage. Pseudomorphs after coesite are hosted in the outermost mantle domain, at the transition between Grt_M and Grt_R, suggesting that garnet mantle already grew at UHP conditions. The coexistence of chloritoid + kyanite constrains the temperature to a very narrow range (540–560 °C), whereas the coexistence of jadeite, glaucophane and paragonite in the same garnet domain limits the pressures to the interval 24–28 kbar, i.e., close to the transition between paragonite-bearing and jadeite-bearing fields (Fig. 10b). The modelled garnet (Sps_1.5Prp₁₉Grs₂Alm₇₆) and chloritoid (X_Mg = 0.30) compositional isopleths intersect within these fields and tightly constrain the late prograde P–T conditions at 540–560 °C, 25–30 kbar, i.e., at the transition between the quartz and coesite stability fields (Fig. 10b). This is consistent with the observed occurrence of polycrystalline aggregates of quartz after coesite at the transition between garnet mantle and rim. The AvPT approach applied to four different sets of mineral compositions gives late prograde P–T conditions of 557 ± 5 °C, 27.4 ± 0.4 kbar (Table 2), consistent with the pseudosection results (Fig. 10b).

The peak assemblage (Coe + Phe + Grt_R + Ky + Jd + Rt) is modelled by a large penta-variant field at T > 530 °C and P > 25 kbar. Although garnet (Sps_1.7Prp₂₆Grs_3.5Alm₆₈) compositional isopleths are far apart, their intersection with phengite (Si = 3.45–3.49 apfu) isopleths allows constraining peak P–T conditions at 660–730 °C and 38–43 kbar (Fig. 10c), consistent with the peak P–T conditions proposed in previous papers and based on other lithologies (i.e., pyrope-bearing whiteschists, eclogites, marbles; see review in Ferrando et al., 2017). The AvPT approach applied on the peak assemblage did not converge to a result, because its variance is too high to define enough reactions.

The observed prograde and peak assemblage (Qz + Phe + Grt + Cld + Gln + Jd + Lws + Rt) is modelled by a large quadri-variant field at 450–540 °C, 19–25 kbar. The modelled garnet, chloritoid and phengite compositional isopleths intersect within this field and constrain prograde and peak P–T conditions as follows: (i) early prograde conditions are constrained at 490–510 °C, 20–23 kbar using the compositions of garnet core (Sps₉Prp_8.5Grs₁₆Alm₆₇), of its chloritoid inclusions (X_Mg = 0.20) and of phengite (Si = 3.31–3.43 apfu) (Fig. 11a); (ii) late prograde conditions are constrained at 500–520 °C, 20–24 kbar using the compositions of garnet mantle (Sps₆Prp_9.5Grs₂₀Alm₆₄), of its chloritoid inclusions (X_Mg = 0.22) and of phengite (Si = 3.31–3.43 apfu) (Fig. 11b); (iii) peak conditions are estimated at 510–525 °C, 20–23 kbar using the compositions of garnet rim (Sps_4.5Prp₉Grs_18.5Alm₆₈), its chloritoid inclusions (X_Mg = 0.24) and phengite (Si = 3.31–3.43 apfu) (Fig. 11c). The AvPT approach applied to four different sets of mineral compositions gives peak conditions of 495 ± 8 °C, 20.3 ± 0.6 kbar (Table 2), consistent with pseudosection results (Fig. 11c).

(Alpine) peak P–T conditions for this sample were estimated using the AvPT approach, applied to three different sets of mineral compositions. Although not observed in the sample, lawsonite was included in the peak assemblage, allowing to consider the reactions involving grossular, which are relevant because garnet is relatively enriched in this component. This strategy is supported by the fact that other metapelites from the same unit show unambiguous evidence for the occurrence of lawsonite in the peak assemblage (Nosenzo, 2018). The AvPT results point to peak P–T conditions of 452 ± 8 °C, 17.2 ± 0.4 kbar (Table 2).

The prograde P–T evolution inferred for the four units bounding the UHP BIU is summarized below:

• The PU experienced a prograde evolution starting in the lawsonite–blueschist facies and reaching peak conditions in the eclogite-facies (500–515 °C, 20–23 kbar; Fig. 12a). Peak-T conditions are similar to those estimated by Avigad et al. (2003) (i.e., 530 °C) using multi-equilibrium thermobarometry applied on garnet-bearing metapelites from the same structural position. However, peak-P conditions are significantly higher than the 14–16 kbar estimated by the same authors (Fig. 12a).

• The P–T prograde path inferred for the SCU and RSU is remarkably similar to that constrained for the PU, with eclogite-facies peak P–T conditions at 500–520 °C, 20–24 kbar (Fig. 12b and d). The estimated peak-P conditions are significantly higher than those (500–550 °C, 15 kbar) constrained by Chopin et al. (1991) and Compagnoni & Rolfo (2003), whereas peak-T are in agreement with literature data (Fig. 12b and d).

• In contrast to the other units, the DSU did not experience eclogitic metamorphism. Estimated peak conditions (450–470 °C, 17–18 kbar) point to lawsonite–blueschist facies conditions (Fig. 12e); peak-T are similar to those reported by Chopin et al. (1991) (500 °C, 10–12 kbar), but pressures are significantly higher (Fig. 12e).

Overall, our new data for the HP units bounding the UHP one substantially confirm peak-T already estimated more than 25 years ago, but point to a significant increase of peak-P estimates (ΔP = 5–10 kbar). This is the same trend already followed in the definition of peak P–T conditions for the UHP BIU, that were initially estimated to be at about 30 kbar (e.g., Chopin et al., 1991; Compagnoni et al., 1995) and progressively increased (35 kbar: e.g., Compagnoni & Rolfo, 2003) up to 40–43 kbar (e.g., Hermann, 2003; Castelli et al., 2007; Groppo et al., 2007a; Ferrando et al., 2009, 2017), in parallel with the progressive improvement of thermodynamic modelling approaches that extended the limits of conventional thermobarometry.

Peak P–T conditions (730 °C, 40–43 kbar) and retrograde evolution of the UHP BIU are now very well defined and were constrained in the last decades through detailed petrological studies on a great variety of lithologies, among them pyrope-bearing whiteschists (Schertl et al., 1991; Hermann, 2003; Ferrando et al., 2009; Gauthiez-Putallaz et al., 2016), eclogites (Nowlan et al., 2000; Di Vincenzo et al., 2006; Groppo et al., 2007a), marbles and calc-silicate rocks (Rubatto & Hermann, 2001; Ferraris et al., 2005; Di Vincenzo et al., 2006; Castelli et al., 2007). Garnet-bearing metapelites have never been used to constrain the BIU peak P–T conditions and have only been marginally used to constrain its retrograde evolution (Groppo et al., 2006). This is due to the fact that the highly variant assemblage (i.e., Coe + Phe + Grt + Ky + Jd) stable at UHP conditions is not suitable for application of conventional thermobarometry. Conversely, metapelites have been already used to constrain the early prograde evolution of the BIU, thanks to the preservation of abundant prograde inclusions in the core of porphyroblastic garnets. Chopin et al. (1991) and Michard et al. (1993) described a garnet-bearing phengitic micaschist (sample DM8412) very similar to our sample DM1281. Both their and our garnet porphyroblasts are, in turn, remarkably similar to the cores of the “superzoned” garnets described by Compagnoni & Hirajima (2001) (sample DM880). Chopin et al. (1991) and Compagnoni & Hirajima (2001) interpreted the coexistence of chloritoid, kyanite, staurolite and chlorite in the polymineralic inclusions hosted within garnet cores as evidence that the BIU prograde trajectory passed close to the invariant point involving Cld + Ky + St + Chl, located around 600 °C, 15 kbar. This was considered as a pinning point for the BIU prograde path for more than 25 years, resulting in a steep prograde trajectory located at temperatures significantly higher than the peak-T conditions estimated at that time for the adjacent units.

Our study unambiguously demonstrates that staurolite in the polymineralic inclusions hosted within garnet core and mantle has a retrograde nature, growing at the interface between garnet and kyanite, the latter mineral preserving a rounded/corroded shape (Fig. 4e). We suggest that staurolite grew during the retrograde evolution through a reaction that involved garnet, kyanite and chlorite as reactants. The aqueous fluid needed to trigger the reaction would not have been necessarily introduced from outside through fractures, but it could derive from the breakdown of chlorite associated with kyanite in the polymineralic inclusion (i.e., internally-derived fluid). Reaction modelling by the least square method (freeware application available on demand; Godard, 2009) applied to the composition of garnet core/mantle, chlorite, kyanite and staurolite yielded balanced reactions of this type: Chl + Ky + Grt = St (details are given in Table SM8), accounting for the retrograde formation of staurolite in a closed system, i.e. the polymineralic inclusion. The above-mentioned reaction is predicted to occur in the KFMASH system at about 600–650 °C, 13–15 kbar (White et al., 2014), and it would have been therefore intersected during the retrograde evolution of the BIU (Fig. 12c). It is worth noting that the growth of retrograde staurolite was already observed in the matrix of a garnet-bearing metapelite from the BIU, and was constrained exactly at these P–T conditions (Groppo et al., 2006).

The new interpretation of the nature of staurolite occurring in the polymineralic inclusions hosted in the BIU garnet porphyroblasts has important consequences for the definition of the BIU prograde P–T path. The results of thermodynamic modelling and AvPT tightly constrain the growth of garnet core at 520–540 °C, 16–23 kbar, i.e., at significantly lower T and higher P than previously estimated, and at P–T conditions close to the peak P–T conditions of the adjacent units. Moreover, the growth of garnet mantle is modelled at UHP conditions, consistent with the occurrence of pseudomorphs after coesite included in the mantle domain. The whole prograde evolution of the BIU is thus significantly different from that previously assumed; it does not follow a regularly steep P/T gradient, but it is rather characterized by: (i) an early prograde evolution along a moderately steep P/T gradient, similar to that followed by the adjacent HP units; (ii) an intermediate steep, almost isothermal, pressure increase, and (iii) a late prograde temperature increase along a moderately steep P/T gradient, still at increasing pressure.

An in-depth discussion of the geodynamic processes responsible for the architecture of the southern DMM is beyond the aim of this paper. However, the following conclusions are worth noting and could represent new constraints for future investigations:

1. The PU experienced eclogite-facies metamorphism under peak P–T conditions very similar to those registered by the SCU and RSU (500–520 °C, 20–24 kbar). Moreover, the PU, SCU and RSU units followed the same prograde path, along a medium T/P gradient of about 12 °C kbar⁻¹, suggesting similar burial mechanisms. The similarity of the peak P–T conditions experienced by the PU and SCU, combined with the fact that the PU sample was collected from an outcrop located very close to the contact with the BIU (Fig. 1), could potentially raise the doubt that the two samples actually belong to the same unit (i.e., SCU). The contact between the PU and overlying units (either SCU or BIU) has been located thanks to the abrupt occurrence, moving downward in the DMM nappe stack, of abundant graphitic micaschists with quartzite intercalations (see Fig. 1). These micaschists have never been described within the other units (either SCU, BIU or RSU) and, in turn, have been considered diagnostic of the former PU (e.g., Vialon, 1966; Chopin et al., 1991; Avigad et al., 2003). Following this criterion, sample DM1485 has been attributed to the PU. Our results thus highlight that, in the investigated area, the PU does not correspond to a blueschist unit, as previously described (Avigad et al., 2003), and therefore the actual occurrence of a blueschist unit stacked below eclogitic ones has to be restricted to northernmost sectors of the DMM (i.e., the Brianconnais-like tectonic window in the Pinerolo area; e.g., Manzotti et al., 2016).

2. The UHP BIU seems to have followed an early prograde evolution similar to that of the other eclogitic units of the southern DMM tectonic “sandwich”. The attainment of UHP peak conditions occurred through an earlier almost isothermal increase in pressure (ΔP = 8–10 kbar) along a significantly lower T/P gradient (ca. 3 °C kbar⁻¹) and a later increase in temperature (ΔT = 150–170 °C) along a medium T/P gradient (ca. 15 °C kbar⁻¹). Our results suggest that UHP metamorphism was not the result of a single, homogeneous, process, but rather the product of a two-step process. Further geochronological studies are needed in order to test if the UHP BIU shared a synchronous prograde evolution with the adjacent eclogitic units.

3. The DSU is the only unit of the southern DMM tectonic “sandwich” that, according to our results, did not experience eclogite-facies metamorphism. It is worth noting that, on the field, the DSU is separated from the eclogitic units by a hundreds of metres thick shear zone (tectonic mélange sensuFesta et al., 2019), which includes slices of meta-ophiolite (Valmala Shear Zone: Nosenzo, 2018). The detailed description and interpretation of this shear zone is beyond the aim of this paper and further studies are necessary to understand its role in decoupling the southern DMM during subduction and/or in driving exhumation of related (U)HP units.

Roberto Compagnoni introduced most of us to the fascinating southern Dora-Maira Massif and to the beauty and challenges of metamorphic petrology. R. Compagnoni, R. Turello and T. Hirajima collected sample DM1281 studied in this paper. Two anonymous reviewers provided constructive comments which significantly improved the manuscript; the Guest Editor is warmly thanked for its several constructive remarks and its meticulous editorial handling. This research was funded by University of Torino, Ricerca Locale, ex-60% 2016, 2017 funds: FERS_RILO_17_01 (SF), FESA_RILO_17_01 (AF), GROC_RILO_17_01 (CG), ROLF_RILO_16_01 (FR), and by the Italian Ministry of University and Research, PRIN 2015: 2015EC9PJ5 (GB).