Deciphering the metamorphic evolution of the Pulo do Lobo metasedimentary domain (SW Iberian Variscides)

. The Pulo do Lobo domain is one of the units ex-posed within the orogenic suture zone between the Ossa-Morena

Abstract. The Pulo do Lobo domain is one of the units exposed within the orogenic suture zone between the Ossa-Morena and the South Portuguese zones in the SW Iberian Variscides. This metasedimentary unit has been classically interpreted as a Rheic subduction-related accretionary prism formed during pre-Carboniferous convergence and eventual collision between the South Portuguese Zone (part of Avalonia) and the Ossa-Morena Zone (peri-Gondwanan terrane). Discrete mafic intrusions also occur within the dominant Pulo do Lobo metapelites, related to an intra-orogenic Mississippian transtensional and magmatic event that had a significant thermal input. Three different approaches have been applied to the Devonian-Carboniferous phyllites and slates of the Pulo do Lobo domain in order to study their poorly known low-grade metamorphic evolution. X-ray diffraction (XRD) was used to identify the mineralogy and measure crystallographic parameters (illite "crystallinity" and Kwhite mica b-cell dimension). Compositional maps of selected samples were obtained from electron probe microanalysis, which allowed for processing with XMapTools software, and chlorite semiempirical and thermodynamic geothermometry was performed. Thermometry based on Raman spectroscopy of carbonaceous material (RSCM) was used to obtain peak temperatures.
The microstructural study shows the existence of two phyllosilicate growth events in the chlorite zone, the main one (M 1 ) related to the development of a Devonian foliation S 1 and a minor one (M 2 ) associated with a crenulation cleavage (S 2 ) developed in middle-upper Carboniferous times. M 1 entered well into epizone (greenschist facies) conditions. M 2 conditions were at lower temperature, reaching the anchizone-epizone boundary. These data accord well with the angular unconformity that separates the Devonian and Carboniferous formations of the Pulo do Lobo domain. The varied results obtained by the different approaches followed, combined with microstructural analysis, provide different snapshots of the metamorphic history. Thus, RSCM temperatures are higher in comparison with the other methods applied, which is interpreted to reflect a faster re-equilibration during the short-lived thermal Mississippian event. Regarding the metamorphic pressure, the data are very homogeneous: very low celadonite content Published by Copernicus Publications on behalf of the European Geosciences Union. (0 %-10 %) in muscovite (and low values of K-white mica b-cell dimension; 8.995 Å mean value), indicating a low pressure-temperature gradient, which is unexpected in a subduction-related accretionary prism.
The metamorphism of the Iberian Variscides has been mostly studied on high-grade metamorphosed rocks in order to characterize and obtain the P -T -t (pressuretemperature-time) paths of suture-related units (e.g., Gil Ibarguchi et al., 1990;Abalos et al., 1991;Escuder Viruete et al., 1994;Barbero, 1995;Arenas et al., 1997;Fonseca et al., 1999;López-Carmona et al., 2013;Martínez Catalán et al., 2014). The low-to very-low-grade units have been also studied (e.g., Martínez Catalán, 1985;Bastida et al., 1986Bastida et al., , 2002López Munguira et al., 1991;Gutiérrez-Alonso and Nieto, 1996;Abad et al., 2001Abad et al., , 2002Abad et al., , 2003aMartínez Poyatos et al., 2001;Nieto et al., 2005;Vázquez et al., 2007), despite the scarcity of appropriate robust methodologies to apply in these kind of rocks. Obtaining new results from the low-grade rocks of the Pulo do Lobo domain, a suture-related low-grade unit in SW Iberia, is of prime importance in order to understand its significance and tectonometamorphic evolution, whose interpretations have been the cause of discrepancies, and to reconstruct the overall history of the SW Iberian Variscides.
In this work, three different methodologies are applied to a number of samples of the Pulo do Lobo domain ( Fig. 1): (i) X-ray diffraction (XRD) in order to identify minerals not easily recognizable with optical microscopy (fine-grained muscovite, paragonite, mixed-layer phyllosilicates, etc.) and obtain thermobarometric information via the measurement of crystallographic parameters (illite crystallinity and b-cell dimension); (ii) compositional maps derived from electron probe microanalysis (EPMA), which enables the recognition of different tectonometamorphic events by combining mineral composition and microtextural features (e.g., Airaghi et al., 2017), as well as the application of geothermobarometers based on chlorite and K-white mica compositions; and (iii) Raman spectroscopy of carbonaceous material (RSCM) to estimate peak temperatures using an adapted thermometric calibration. The results obtained enable a discussion of the tectonometamorphic evolution of the Pulo do Lobo domain. Moreover, the comparison of the different approaches tests their reliability and sensitivity to characterize different geological processes.
South of the Beja-Acebuches Amphibolites, low-to verylow-grade successions crop out: Devonian siliciclastics, earliest Carboniferous volcano-sedimentary rocks, and a southmigrating Carboniferous flysch (e.g., . These rocks are usually grouped into two geological domains: the Pulo do Lobo domain to the north and the SPZs to the south (Fig. 1a, b). The deformation in the SPZ consists of a south-to southwest-vergent fold-and-thrust belt with decreasing strain intensity and age southwards Simancas et al., 2004). The metamorphic grade also decreases southwards, from epizone to diagenetic, through the SPZ (Munhá, 1990;Abad et al., 2001). The Pulo do Lobo domain, which has been traditionally considered a suturerelated unit (see below), is the focus of this work.

Pulo do Lobo domain
The Pulo do Lobo domain constitutes a polydeformed structure affecting low-grade Devonian-Carboniferous sedimentary formations. These formations are the following, from bottom to top (Fig. 1b-c).
Solid Earth, 11, 469-488, 2020 www.solid-earth.net/11/469/2020/ i. The Pulo do Lobo Formation (sensu stricto) is constituted by a succession of satiny black to grey phyllites and fine-grained schists with minor intercalations of quartz sandstones (Fig. 2a). The presence of abundant segregated quartz veins (pre-to post-folding) is common. The palynological content suggests a middle Frasnian age .
ii. The Ribeira de Limas Formation is constituted by phyllites with thin beds of quartz sandstones and arkoses ( Fig. 2b). The presence of palynomorphs also suggests a middle Frasnian age for this formation . The contact with the underlying Pulo do Lobo Formation is gradual, with a progressive increase in sandstones and a decrease in phyllites upwards. For that reason, we will refer to the Pulo do Lobo and Ribeira de Limas formations as the lower formations of the Pulo do Lobo domain. Furthermore, these lower formations share the same structural fabrics consisting of three fold-related foliations ( Fig. 2a- , 2015). The first foliation of the lower formations (S 1 ) is preserved inside microlithons of the second foliation (S 2 ); usually, the angle between these two foliations is high. S 2 is the main foliation and consists of a crenulation-dissolution cleavage that commonly appears as a millimetric-to centimetric-spaced tectonic banding. This foliation is axial-planar to north-vergent folds. The third foliation (S 3 ) is a spaced crenulationdissolution cleavage that typically develops a characteristic decimetric-to metric-scale tectonic banding. S 3 is associated with upright to slightly south-vergent folds.
iii. The Santa Iría Formation is composed of alternating beds of slates and greywackes (Fig. 2c). The greywacke beds show normal grading and an erosive base. Paleontological and palynostratigraphic studies suggest an upper Famennian age for this formation (Pereira et al., 2008. However, an early Carboniferous age is also plausible, since more than 90 % of the palynomorphs correspond to reworked material (Lopes et al., 2014) and the younger detrital zircon population is early Carboniferous (Braid et al., 2011;Pérez-Cáceres et al., 2017;Pereira et al., 2019). The Santa Iría Formation only shows two foliations, correlative with the last two deformation phases in the lower formations. Therefore, an unconformity between them is inferred, which is consistent with the age and flysch character of the Santa Iría Formation (Pérez-Cáceres et al., 2015). S 2 is observed as a penetrative slaty cleavage, while S 3 is a disjunctive crenulation cleavage.
According to Silva et al. (1990) and Pérez-Cáceres et al. (2015), the two main foliations (S 2 and S 3 ) in the Pulo do Lobo domain resulted from the middle-upper Carboniferous collision between the OMZ and SPZ. The first foliation (S 1 ) in the Pulo do Lobo domain might have formed during the convergence of the Rheic Ocean subduction and/or the beginning of the Variscan collision, probably at Late Devonian times.
The Pulo do Lobo domain contains some decimetric-to metric-scale lenticular bodies of MORB-featured metamafic rocks. At some outcrops, the mafic rocks are embedded in a greenish detrital matrix, thus suggesting an olistostromic origin (the Peramora Olistostrome; Eden and Andrews, 1990). These rocks are tectonically imbricated with the phyllites of the Pulo do Lobo Formation and hence form a tectonic mélange (the so-called Peramora Mélange; Fig. 1b -c;Eden, 1991;Dahn et al., 2014). Based on this aspect and on the supposedly Rheic Ocean derived greenschists, the Pulo do Lobo domain has been classically interpreted as a pre-collisional subduction-related accretionary prism (Eden and Andrews, 1990;Silva et al., 1990;Eden, 1991;Braid et al., 2010;Ribeiro et al., 2010;Dahn et al., 2014;Quesada et al., 2019). However, the recently obtained Mississippian U-Pb zircon ages from the metamafic rocks (Dahn et al., 2014;Pérez-Cáceres et al., 2015) make it difficult to maintain such a hypothesis. More properly, they can be interpreted as mafic intrusions or extrusions in the frame of the intra-orogenic transtensional magmatic event that prevailed in SW Iberia during the Mississippian. The metamafic rocks display a foliation (equivalent to the S 2 of the enveloping metasediments) developed during loosely constrained greenschist facies metamorphism. These rocks would have been imbricated with the Pulo do Lobo metasedimentary rocks during the second deformation phase, which caused S 2 (Pérez- Cáceres et al., 2015). Our multidisciplinary metamorphic study of the Pulo do Lobo metasediments provides crucial data concerning the tectonic significance of this domain.

Samples and analytical methods
A total of 18 samples were collected from well-exposed outcrops of phyllosilicate-rich detrital rocks of the Pulo do Lobo domain along two north-south transects perpendicular to the structural trend. Five samples belong to the Santa Iría Formation (unconformable upper formation) and 13 to the lower formations (Pulo do Lobo and Ribeira de Limas formations) (the locations of the samples are shown in the map and cross sections in Fig. 1b-c, and the UTM coordinates are given in the Supplement). As a whole, the samples were selected in fresh outcrops, far from faults and joints, and were as homogeneous and representative as possible. The sampling strategy was to collect representative sites, both of the overall stratigraphic succession and along two transects. We also aimed to characterize the unconformity between the lower and upper formations from a metamorphic point of view, since the crystallinity aspect to the naked eye appears to be lower in the Santa Iría Formation. Some samples from the lowermost Pulo do Lobo Formation were collected ≈ 200 m from the metabasite lenses of the Peramora Mélange.
Samples were examined under an optical microscope and scanning electron microscope (SEM) for overall mineralogy, deformation textures, and mineral-foliation relationships using an environmental scanning electron microscope FEI model Quanta 400 operating at 15-20 keV (Centro de Instrumentación Científica, CIC, University of Granada, Spain).

X-ray diffraction
Sample preparation and analysis by XRD were done in the laboratories of the Department of Mineralogy and Petrology of the University of Granada (Spain). After washing and cleaning the patinas and oxides, samples were crushed to a < 2 mm fraction. The < 2 µm fractions were separated by repeated extraction of supernatant liquid after centrifugation, according to the Stokes law. Oriented aggregates were prepared by sedimentation on glass slides of whole-rock and < 2 µm fractions (the latter aims to minimize the content of detrital micas non-re-equilibrated during very-lowgrade metamorphism, which are generally larger than 2 µm; Solid Earth, 11, 469-488, 2020 www.solid-earth.net/11/469/2020/ Moore and Reynolds, 1997). Samples were also treated with ethylene glycol (EGC) to identify illite-smectite or chloritesmectite mixed layers on the basis of their expandability. Samples were analyzed using a PANalytical X'Pert Pro powder diffractometer equipped with an X'Celerator detector, CuKα radiation, operated at 45 kV and 40 mA, with an Ni filter and 0.25 • divergence slit. The resulting diffraction diagrams were examined to extract information on mineralogy based on their characteristic reflections and white mica crystal data. The illite crystallinity index (Kübler index, KI;Kübler, 1968) has been estimated from the measurement of the full peak width of K-white mica at half-maximum intensity (FWHM values), expressed as • 2θ of the Bragg angle. Preparation of samples and experimental conditions were carried out according to IGCP 294 IC Working Group recommendations (Kisch, 1991). A step increment of 0.008 • 2θ and a counting time of 52 s per step were used in the diffractometer. The KI was measured in all samples for both the 5 and 10 Å reflection peaks of K-white mica in order to identify possible effects of other overlapping phases (Nieto and Sánchez-Navas, 1994;Battaglia et al., 2004). Some XRD traces showing complex mixtures of mixedlayer minerals were decomposed with the MacDiff software (Petschick, 2004). The FWHM values obtained in the laboratory (x) were transformed to Crystallinity Index Standard (CIS) values (y) using the equation y = 0.972x+0.1096 (R 2 = 0.942), obtained from the measure in our laboratory of the international standards of Warr and Rice (1994).  (Warr and Ferreiro Mähnlmann, 2015). The thermal range for the anchizone is estimated as 200-300 • C, though the KI cannot be considered a true geothermometer (Frey, 1987;Kisch, 1987).
The b-cell parameter of white mica was obtained from the (060) reflection peak measured with quartz as an internal standard on polished rock slices cut normal to the sample main foliation S 2 (Sassi and Scolari, 1974). The b-cell dimension of K-white mica is typically proportional to the extent of phengitic substitution and is therefore considered a proxy for the pressure conditions during its crystallization. Thus, Guidotti and Sassi (1986) have shown that b values lower than 9.000 Å are typical of low-pressure facies conditions, while b values higher than 9.040 Å are related to relatively high-pressure facies metamorphism. Precise measurements of the basal spacing of white mica (d 001 ) have also been made using quartz from the sample itself as an internal standard; d 001 is related to the paragonitic Na/K substitution (Guidotti et al., 1992), thereby approximately reflecting the temperature of white mica formation (Guidotti et al., 1994).

EPMA-derived X-ray compositional maps and chlorite thermometry
From all of the collected samples, we selected those with the larger phyllosilicate grain size for electron probe microanalysis (EPMA). Thus, three carbon-coated polished thin sections were studied. The selected samples (PLB-84, PLB-88, and PLB-93) belong to the lower formations of the Pulo do Lobo domain (Fig. 2d-e). The Santa Iría samples could not be studied due to the tiny grain size of the minerals (commonly less than 3 µm).
Compositional maps and accurate spot analyses were performed on a JEOL JXA-8230 EPMA at the Institut des Sciences de la Terre (ISTerre) in Grenoble (France), according to the analytical procedure proposed by De Andrade et al. (2006) and Lanari et al. (2014a). The data acquisition was made in wavelength-dispersive spectrometry mode (WDS), 10 elements (Si, Ca, Al, K, Mn, Na, P, Ti, Fe, and Mg) were analyzed using five WD spectrometers: TAP crystal for Si and Al, PETL for Ti and P, TAPH for Na and Mg, PETH for K and Ca, and LIFH for Mn and Fe. The standardization was made by using certified natural minerals and synthetic oxides: wollastonite (Si, Ca), corundum (Al), orthoclase (K), rhodonite (Mn), albite (Na), apatite (P), rutile (Ti), hematite (Fe), and periclase (Mg). X-ray maps were obtained by adding successive adjacent profiles. A beam current of 100 nA and a beam size spot (focused) were used. The step (pixel) size was 1 µm and the dwell time was 200-300 ms per pixel. Spot analyses were conducted along the profiles within the mapping at 15 kV accelerating voltage, 12 nA beam current, and 2 µm beam size spot (focused). The on-peak counting time was 30 s for each element and 30 s for two background measurements at either side of each peak. A ZAF correction procedure was applied. The internal standards were orthoclase and/or chromium-augite (Jarosewich et al., 1980), which were run (three points on each standard) after each profile in order to monitor instrumental drift and estimate analytical accuracy. Drift correction was made, if necessary, using the corresponding regression equation.
The WDS X-ray maps were then processed with XMap-Tools (http://www.xmaptools.com, last access: 10 September 2009), a MATLAB-based graphical user interface program to process the chemical maps, link them to thermobarometric models, and estimate the pressure-temperature conditions of the crystallization of minerals in metamorphic rocks (Lanari et al., 2014a). The compositional maps were standardized with the spot analyses measured along the profiles and mineral compositions were plotted into binary and ternary diagrams using the interface modules Chem2D and Triplot3D. Chemical maps of the amount of tetrahedral aluminum (Al IV ) in chlorites were acquired because these amounts are used in many empirical chlorite thermometers (e.g., Cathelineau and Nieva, 1985;Cathelineau, 1988). The temperature conditions were estimated for each chlorite pixel of the maps using the chlorite thermometer of Lanari et www.solid-earth.net/11/469/2020/ Solid Earth, 11, 469-488, 2020 al. (2014b), as well as the approaches of Vidal et al. (2006) and Bourdelle et al. (2013), which are summarized in the Supplement.
In addition to the abovementioned compositional maps, white micas from seven carbon-coated thin sections of the lower formations of the Pulo do Lobo domain were analyzed before with a JEOL WDS four-spectrometer microprobe (JXA-8200 Superprobe) at the University of Huelva (Spain). A combination of silicates and oxides was used for calibration: standards used were wollastonite (Si and Ca), potassium feldspar (Al, K, and Na), forsterite (Mg), and fayalite (Fe). Single-point analyses were conducted with 20 nA probe current, 1-5 µm spot size, and 15 kV of acceleration voltage, with 5 s counting times.

Raman spectroscopy of carbonaceous material
Raman spectroscopy of carbonaceous material (RSCM) is based on the observation that sedimentary carbonaceous material (CM) is progressively transformed into graphite at increasing temperature. Because of the irreversible character of graphitization, CM structure is not sensitive to the retrograde path during the exhumation of rocks, but it depends only on the maximum temperature reached during metamorphism (Beyssac et al., 2002a). Temperature was determined in the range 330-650 • C with a calibration-attached accuracy of ±50 • C due to uncertainties in the petrologic data used for the calibration (Beyssac et al., 2002a). Relative uncertainties in temperature estimates were later reduced (around 10-15 • C; Beyssac et al., 2004). For temperature below 330 • C, Lahfid et al. (2010) performed a systematic study of the evolution of the Raman spectrum of CM in low-grade metamorphic rocks in the Glarus Alps (Switzerland). They showed that the Raman spectrum of CM is slightly different from the spectrum observed at higher temperature, and they established a quantitative correlation between the degree of structuration of CM and temperature.
In this work, 12 representative thin sections previously examined by optical microscopy were selected. From these, 10 samples were analyzed (according to their larger CM grain size and content): eight samples belong to the lower Pulo do Lobo and Ribeira de Limas formations, and the other two samples belong to the Santa Iría Formation. Polished thin sections cut perpendicularly to the S 2 foliation were analyzed at the Institut de Minéralogie, de Physique des Matériaux et de Cosmochimie at the Sorbonne University of Paris (France). We closely followed the analytical procedure described by Beyssac et al. (2002a, b;; see the Supplement). More than 15 Raman spectra (Fig. 3) were obtained for each sample using a Renishaw InVIA Reflex microspectrometer equipped with a 514.5 nm Modulaser argon laser under circular polarization. The laser was focused by a DMLM Leica microscope, and laser power was set below 1 mW at the sample surface. The Rayleigh diffusion was eliminated by edge filters, and the signal was dispersed using a 1800 g mm −1 grating and finally analyzed by a Peltier cooled RENCAM CCD detector. The recorded spectral window was large enough to precisely determine the background correction, from 700 to 2000 cm −1 in the case of low-temperature samples. Before each session, the spectrometer was calibrated with a silicon standard. CM was systematically analyzed behind a transparent adjacent mineral, generally quartz or white mica grains oriented along S 1 . For a full description of the temperature calculations see the Supplement.

Results
According to the petrographic study, all the samples correspond to slates or phyllites with phyllosilicates smaller than 500 µm, composed of variable quartz + Kwhite mica ± chlorite ± feldspar ± ore and accessory minerals ( Fig. 2d-f). Samples from the Santa Iría Formation have a much smaller grain size and apparently lower crystallinity (Fig. 2f). The first foliation S 1 is defined by the largest micas and chlorites (Fig. 2d-e), being folded by microscopic-to centimetric-scale tight folds of the second deformation phase ( Fig. 2a-b, d-e). The second foliation S 2 is the main foliation at outcrop (Fig. 2a-c), but the phyllosilicates (mostly white mica) are smaller than those in S 1 . The third foliation S 3 is a fracture cleavage that does not develop phyllosilicates (Fig. 2a-c).

X-ray diffraction
The mineralogy and crystal parameters of K-white mica obtained from the 18 samples of the Pulo do Lobo domain are summarized in Table 1. The results of KI values, the bcell parameter, and d 001 analyzed in K-white mica obtained from whole-rock and < 2 µm fractions are very similar, which suggests that detrital micas re-equilibrated during metamorphism.
The mineralogy of the samples is relatively simple: Qz + Ms + Fsp + Chl ± Pg ± C-S. The slates of the Santa Iría Formation have quartz, muscovite, and chlorite, with chlorite-smectite (C-S) interlayers in some samples. In the lower formations, besides quartz and muscovite, chlorite is present in almost all of the samples, paragonite appears in most of them, and chlorite-smectite interlayers are occasional.
KI values measured in the 10 Å peak of white mica from the < 2 µm fraction are shown in Table 1 and Fig. 1c with a relative color bar from orange (lower values) to green (higher values). Values of the Santa Iría samples (n = 5) range from 0.20 to 0.26 • 2θ , and the mean value is 0.23 (standard deviation 0.02). In the lower formations (n = 12), KI values range from 0.17 to 0.22, and the mean value is 0.19 (standard deviation 0.02). KI values measured in the 5 Å peak (not shown to avoid repetition) are very similar to those of the 10 Å peak. The measured b-cell parameter of white mica varies within a close range around 9 Å (8.991-9.002). The mean value is 8.995 Å (standard deviation 0.003) for the Santa Iría Formation samples and 8.997 Å (standard deviation 0.003) for the samples of the lower formations; d 001 values average 9.992 Å (standard deviation 0.004) and differ slightly between the upper and lower formations, with higher values in the upper formation.
The results obtained through X-ray diffraction reflect lowgrade metamorphic conditions due to the KI values between 0.17 and 0.26 • 2θ. In addition, b-cell parameters are lower than 9.000 Å, which show a low-pressure metamorphic gradient (low pressure-temperature metamorphic facies conditions; Guidotti and Sassi, 1986).

Compositional maps and chlorite thermometry
X-ray maps show the distribution of major elements and allow for the identification of white mica, chlorite, and some albite porphyroblasts, with ilmenite and rutile as accessory minerals ( Fig. 4a-b). Although quartz is abundant in all of the samples, the areas selected for detailed X-ray mapping (composed mostly of phyllosilicates) do not contain quartz ( Fig. 4a-b). White mica is abundant along both the S 1 and S 2 foliations (Figs. 2d-e and 4b). Chlorite is found mostly along S 1 but is very scarce and small-sized within S 2 (Figs. 2e and 4b), with the exception of sample PLB-93 wherein chlorite is similar in amount in both foliation domains (Fig. 4b).
Mapped compositions of end-members of white mica and chlorite have been plotted in the ternary diagrams in Fig. 5. The composition of white mica is similar in the three maps and is close to muscovite, with 25 % pyrophyllite and very minor celadonite content (0 %-10 %; Fig. 5a). The high content of pyrophyllite (high amount of interlayer vacancies) is typical of low-pressure illitic-mica compositions. Figure 6 shows white mica compositional ratios, which can be related to P -T conditions: they present a low degree of Na substitution and low phengitic component, so they are close to the muscovite end-member in composition. These results point to low-pressure conditions and agree well with XRD results: low b-cell parameter (< 9.000 Å) and high d 001 (> 9.985 Å; Table 1).
Chlorite compositions are variable, though all of them have in common ≈ 50 % clinochlore + daphnite and ≈ 50 % amesite + sudoite (Fig. 5b). Chlorites in sample PLB-88 are low in amesite, with a large variation of clinochlore + daphnite and sudoite contents. In sample PLB-84, chlorites vary in composition between amesite and sudoite, indicating a variation of Al IV , which implies an increase in temperature from rims to cores as shown in the chemical maps in Fig. 4c. Finally, PLB-93 chlorites are poor in sudoite and higher in Al IV content, thus suggesting higher average temperatures  than in PLB-84 and PLB-88. Altogether, the chlorite compositional data show the www.solid-earth.net/11/469/2020/ Solid Earth, 11, 469-488, 2020  . Point analyses were conducted on the microprobe at the University of Huelva (Spain). Qualitative information about temperature and pressure conditions is respectively according to Guidotti et al. (1994), Coggon and Holland (2002), Parra et al. (2002), Massonne and Schereyer (1987), and Massonne and Szpurka (1997).
presence of two end-members: sudoite-rich low temperature (PLB-88) and amesite-rich high temperature (PLB-93). Maps of Al IV in chlorites are displayed in Fig. 4c. Sample PLB-88 displays lower Al IV content (≈ 1.1-1.3 apfu) than sample PLB-93 (≈ 1.3-1.5 apfu). In sample PLB-84, some large chlorite grains oriented along S 1 are zoned, with higher Al IV content in the cores (≈ 1.4 apfu) than in the rims (≈ 1.0 apfu; see the white square in Fig. 4c). According to the empirical calibration of Cathelineau (1988), Al IV in chlorites increases with temperature. Thus, the Al IV content in chlorites manifests different temperatures in different samples and also varies from core to rim in singular grains.
Temperature maps have been obtained with the semiempirical thermometer of Lanari et al. (2014b), which assumes that Fe 2+ is the Fe total (Fig. 4d). Temperatures range between 100 and 200 • C in sample PLB-88, between 150 and 350 • C in sample PLB-84, and between 200 and 450 • C in sample PLB-93. Tiny chlorites developed along S 2 yield lower temperatures than larger and more abundant chlorites along S 1 , with the exception of sample PLB-93. Furthermore, some large chlorites oriented along S 1 are zoned, showing high-temperature relic cores (350-450 • C; see white insets in Fig. 4c-d) surrounded by low-temperature rims (150-250 • C).
To test the Vidal et al. (2005Vidal et al. ( , 2006 and Bourdelle et al. (2013) thermometers, an area of representative chlorites in an S 1 microlithon was selected from each map (see red insets www.solid-earth.net/11/469/2020/ Solid Earth, 11, 469-488, 2020  Fig. 4d for location); n represents the number of chlorites that could be used for each calibration. The number of analyses is lower in those with the Vidal et al. (2006) approach because of the assumption that the Si content of chlorite is lower than 3 apfu.
in Fig. 4d). Corresponding chlorite compositions were extracted and introduced in the chlorite-quartz-water equilibria ( Fig. 7a; Vidal et al., 2005Vidal et al., , 2006Fig. 7b, Bourdelle et al., 2013). The temperature estimates (Fig. 7) derived from both methods are fairly similar, averaging 120-230 • C in sample PLB-88 and 150-380 • C in sample PLB-84. These estimates are also in agreement with the estimates derived from temperature maps calculated with the Lanari et al. (2014a) model. Only sample PLB-93 shows a divergence on temperature averages: mostly 150-250 • C with the thermometer of Bourdelle et al. (2013) and 250-350 • C with that of Vidal et al. (2005. In both cases (Bourdelle et al., 2013and Vidal et al, 2005 methods), the higher-temperature analysis results are obtained from crystal cores in sample PLB-93.

RSCM thermometry
The ratio parameters and corresponding maximum temperatures obtained from all the spectra analyzed are shown in the Supplement. The Raman spectra were decomposed into bands following the appropriate fitting procedure described in Beyssac et al. (2002a) for the lower formations (high-temperature Raman spectra; ratio parameter R2) and Lahfid et al. (2010) for the Santa Iría Formation (low-temperature Raman spectra; ratio parameters RA1 and RA2; Fig. 3). The average temperatures are shown in Table 1 and Fig. 1c with a relative color bar from red (higher temperature) to blue (lower temperature). The average temperatures for the lower formations range from 420 to 530 • C, with a mean value of 468 • C (standard deviation of 35). The highest temperatures are found in samples PLB-82 (530 • C) and PLB-93 (495 • C), and estimates of the remaining samples do not exceed 480 • C. As for the Santa Iría Formation, temperatures are lower (315-330 • C; Table 1) than in the underlying formations.

Deformation-metamorphism relationships
The obtained analytical results are interpreted below in the context of the Variscan evolution of the Pulo do Lobo domain. As described above, two regional deformational events, D 1 and D 2 , yielded two foliations (Devonian S 1 and Solid Earth, 11, 469-488, 2020 www.solid-earth.net/11/469/2020/ Carboniferous S 2 ) accompanied by metamorphic phyllosilicate growth (M 1 and M 2 ). In the cross sections in Fig. 1c, KI values derived from XRD and average temperature from RSCM suggest that the lowest metamorphic grade (green and blue) corresponds to the Santa Iría Formation. Textural observations suggest that in most samples of the lower formations M 1 was the main crystallization event, developing abundant and large-sized white mica and chlorite in S 1 microlithons, while M 2 generated small-sized white mica (e.g., Fig. 2e and map 1 in Fig. 4). On the other hand, polydeformed rocks commonly show previously grown minerals rotated towards a new foliation developed at lower-grade conditions, without new crystallization. This could be the case of the white micas that define S 2 in some samples (illustrated in Fig. 2d), which, in turn, is compatible with the similar chemical composition of S 1 and S 2 micas (Fig. 5a). As shown in these samples, S 1 is variably crenulated by D 2 , implying that M 1 minerals are variably rotated towards S 2 . Consequently, the metamorphic data obtained from the samples of the lower formations will be ascribed to D 1 -M 1 . Sample PLB-93 might represent an exception, since its slightly higher RSCM and chlorite-derived temperatures could be due to nearby intrusions ( Fig. 1b and 1c2; see below). In this respect, it is important to note the Mississippian transtensional event (basin development and abundant mafic magmatism) that took place between D 1 and D 2 (Pérez- Cáceres et al., 2015). The characterization of M 2 can be done by studying the samples from the Santa Iría Formation, which are only affected by S 2 accompanied by small-sized phyllosilicate growth (Fig. 2f). No crystallization has been observed related to the S 3 disjunctive crenulation cleavage.

First tectonothermal event (Devonian M 1 )
The observed mineral assemblage (Qz + Ab + Ms + Chl ± Pg), together with the presence of C-S, is compatible with low-grade metamorphic conditions (Table 1). White mica crystallinity values (0.17-0.22 • 2θ; average 0.19) are consistently in the range of the epizone (low-grade or greenschists facies; > 300 • C; Frey, 1987;Kisch, 1987;Warr and Ferreiro Mähnlmann, 2015), in accordance with the values reported by Abad et al. (2001) in a more general study of the diagenetic-metamorphic evolution of the South Portuguese Zone metapelites. Nevertheless, both the values of KI, far from 0.14 • 2θ (high epizone conditions according to Abad et al., 2006), and their variability, suggest that temperature was not high enough to stabilize a highly crystalline white mica. This interpretation is in agreement with the low Na content of K-micas coexisting with paragonite (Fig. 6), implying a very low-temperature position in the muscovite-paragonite solvus for natural quasi-binary Pg-Ms pairs (Guidotti et al., 1994). By contrast, the maximum temperatures obtained from RSCM geothermometry are surprisingly high (420-530 • C; average 470 • C; corresponding to very high epizone or even medium-grade metamorphic conditions; Table 1).
The composition of paired chlorite and white mica is normally used to calculate pressure and temperature (e.g., Vidal et al., 2006), but the multi-equilibrium approach was not successful because the P -T equilibrium conditions did not converge. This result is indicative of chemical disequilibrium, precluding their use as a reliable geothermobarometer in this case. The temperatures calculated from chlorite compositions following various approaches (Vidal et al., 2006, Fig. 7a;Bourdelle et al., 2013, Fig. 7b;Lanari et al., 2014a, Fig. 4d) are as follows: 100-230 • C for sample PLB-88, 150-375 • C for sample PLB-84, and 150-450 • C for sample and Table 1). The slightly higher temperature of sample PLB-93 is inferred from its highest white mica crystallinity (0.17 • 2θ ), high RSCM temperature (495 • C), high-temperature (amesite-rich) chlorite, and higher chlorite thermometry (Table 1); this can be explained by its proximity to metric-scale mafic igneous bodies of the Peramora Mélange (located ≈ 200 m to the south; Pérez-Cáceres et al., 2015) and/or to a granite stock (located ≈ 5 km to the west) (Fig. 1b).
In our samples there is some evidence of chlorite retrogression: (i) the chemical disequilibrium shown by the white mica-chlorite geothermobarometer, (ii) the presence of C-S mixed layers that are not stable in the epizone (e.g., Potel et al., 2006), (iii) the difference between temperature estimates from crystal rims to cores and the higher temperature relic cores preserved in large chlorites defining S 1 (Fig. 4c-d), and (iv) the previously reported XRD and TEM data for chlorite retrograded to smectite and corrensite in the Pulo do Lobo domain (see Fig. 1 in Nieto et al., 2005). The existence of chlorites with different compositions crystallized at different temperatures is the typical scenario (e.g., Vidal et al., 2006Vidal et al., , 2016Lanari et al., 2012Lanari et al., , 2014aGrosch et al., 2012Grosch et al., , 2014Cantarero et al., 2014). In such situations, precise estimates of a single temperature and pressure attributable to peak conditions are really difficult to obtain. The maximum temperature shown by chlorite relic cores is 350-450 • C (Fig. 4d), which is consistent with the conditions estimated for M 1 by means of RSCM data.
An issue that deserves some discussion is the difference in temperature estimates between RSCM and other techniques. RSCM thermometry records the peak temperature and is not sensitive to the retrograde path. Alternatively, other methods based on phyllosilicate compositions are prone to record re-equilibration during the retrograde path; thus, they rarely record the peak conditions, except perhaps in the core of certain large crystals. Therefore, RSCM and phyllosilicatebased methods do not record the same information on temperature but are in fact complementary. The analyzed CM grains were carefully checked by microtextural observations and spectral geometry to ensure that these grains were actually derived from in situ organic matter graphitized during metamorphism.
www.solid-earth.net/11/469/2020/ Solid Earth, 11, 469-488, 2020 In our case study, at the high peak temperature given by the RSCM thermometry, minerals such as biotite and garnet are expected to crystallize in metasedimentary rocks, though they were not observed in our samples. Biotite has been said to exist in a few previous works (Apalategui et al., 1983;Braid et al., 2010;Rubio Pascual et al., 2013). However, in a few of our samples, biotite-looking crystals turned out to be oxichlorites under SEM analyses. The absence or exceptional presence of biotite can be due to whole-rock composition and explained by growth inhibition related to Na excess, as evidenced by the presence of albite and paragonite in our samples. Another possible explanation could be the higher sensitivity of CM graphitization to rapid re-equilibration during a short-duration thermal event. Thus, the Mississippian intrusions subsequent to M 1 in the Pulo do Lobo Formation (see the description in Sect. 2) could have exerted a fast and locally intense thermal imprint that influenced CM but not the crystal chemistry of silicates. Moreover, recrystallization processes are not only a function of temperature, but are also promoted by deformation and stress, time, and the fluid to rock ratio (Merriman and Frey, 1999). Observations of this kind (differing reaction kinetics between organic and inorganic material, e.g., illite) in a contact metamorphic setting can be found in Olsson (1999) and Abad et al. (2014). Regarding the timing of geological processes, Mori et al. (2017) investigated the importance of heating duration for RSCM thermometry by studying graphitization around dykes. They showed that small-scale intrusions generating short thermal events modify the structure of CM in the surrounding rocks and concluded that CM crystallinity is clearly related to contact metamorphism. The influence of low-pressure contact aureoles on RSCM temperature patterns is further supported by the results obtained by Hilchie and Jamieson (2014), who concluded that the variation of RSCM temperatures can be controlled by the subsurface geometry of a pluton. Finally, the long-distance thermal influence of plutonic intrusions on low-grade rocks located as far as 10 km has already been documented (e.g., Merriman and Frey, 1999;Martínez Poyatos et al., 2001) and was also recorded by RSCM thermometry in our samples.

Second tectonothermal event (middle-upper
Carboniferous M 2 ) The mineralogy of the Santa Iría samples (Qz + Fsp + Ms + Chl ± C-S) is compatible with verylow-to low-grade conditions. The K-white mica crystallinity values (0.20-0.26 • 2θ ; average 0.23) point to lower epizone conditions, very close to the boundary with the anchizone (≈ 300 • C; Frey, 1987;Kisch, 1987). The temperatures calculated by RSCM in two samples (315 and 330 • C) are compatible with the KI data from XRD analysis.
Our metamorphic data corroborate the existence of an unconformity between the lower and upper formations of the Pulo do Lobo domain (Pérez-Cáceres et al., 2015). Table 2 summarizes the relationship between the deformation and metamorphism of the Pulo do Lobo domain in the context of the Variscan tectonic evolution of SW Iberia (Pérez-Cáceres et al., 2015). The lower formations record a Devonian tectonothermal event that reached epizone or lower greenschist facies conditions (M 1 with generalized phyllosilicate growth at temperatures as high as 450 • C), whereas the overlying upper formation records a middle-upper Carboniferous tectonothermal event close to the anchizone-epizone boundary (M 2 with small-sized phyllosilicate growth at temperatures ≈ 300-330 • C; Table 1). Obviously, M 2 also affected somehow the lower formations, being, at least in part, the responsible for the observed retrogression of M 1 chlorite and/or crystallization of new chlorites at lower temperature.

Pressure conditions
The measured b-cell parameters of K-white mica (in a short range between 8.991 and 9.002 Å; average 8.996; standard deviation 0.003) are very similar in the lower and upper formations of the Pulo do Lobo domain. Thus, the b parameter shows little variation and reflects very low phengite substitution in mica, as expected in low-pressure settings (Potel et al., 2006(Potel et al., , 2016 near the intermediate pressure gradient boundary (Guidotti and Sassi, 1986).
In agreement with the low b-cell parameters, the composition of K-white mica is close to muscovite with very low celadonite and higher pyrophyllite content (Fig. 5a), as ex-Solid Earth, 11, 469-488, 2020 www.solid-earth.net/11/469/2020/ pected for illite-rich mica formed at low-pressure gradients.
In the case of high-or medium-pressure conditions, a continuous trend in mica compositions would reflect the decompression path after the peak pressure, while the b-cell parameter would represent an average value of the range of mica compositions found in the sample (Abad et al., 2003b). In contrast, in low-pressure settings, the overall range of recorded pressure is very restricted, and micas present similar compositions and b-cell parameters among the various samples, as in the case of the Pulo do Lobo samples (Figs. 5a and 6, Table 1). The Pulo do Lobo domain has been classically interpreted as a pre-collisional subduction-related accretionary prism based on the MORB geochemistry of mafic rocks (see Sect. 2.1). According to this classical interpretation, features typical of modern subduction systems should be expected, such as high-pressure metamorphic gradient remnants of partial subduction-exhumation in an accretionary wedge (e.g., Platt, 1986;Ernst, 2005) or slices of oceanic slab-derived lithologies (varied mid-ocean ridge meta-igneous lithologies and also deep ocean bottom metasediments). Thus, recent works on the Makran accretionary prism (Omrani et al., 2017) and the subduction system of Japan (Endo and Wallis, 2017) describe an accretionary mélange complex composed of pelagic sedimentary rocks, ophiolites, greenschists, amphibolites, and blueschists with high-pressure minerals such as lawsonite and glaucophane. However, most of the geological data concerning the Pulo do Lobo domain do not support such an interpretation (see Sect. 2.1), and our new estimates of pressure conditions are also in disagreement. The only evidence supporting a high-pressure gradient in the Pulo do Lobo domain is the interpretation of some rhomboidal aggregates of epidote porphyroblasts as the remnants of supposed lawsonite grown prior to S 2 in some samples in the Pulo do Lobo mafic schists (Rubio Pascual et al., 2013). However, no analytical data have been presented to support the presence of lawsonite pseudomorphs.

Conclusions
A total of 18 samples of metapelites from the Pulo do Lobo domain have been studied to determine their Variscan lowgrade metamorphic conditions. The microstructural analysis of the samples of the lower formations (Devonian Pulo do Lobo and Ribeira de Limas) shows the existence of two superposed low-grade tectonothermal events with associated foliation and phyllosilicate growth (S 1 -M 1 and S 2 -M 2 ; Table 2). M 2 was less intense and is the only event that affected the overlying Carboniferous Santa Iría Formation. The regional geology also shows that a Mississippian thermal (magmatic) event occurred between M 1 and M 2 .
The temperature estimates obtained from RSCM are higher than the estimates obtained from chlorite geothermometry and white mica data. This discrepancy can be explained by the fact that RSCM records the maximum temperature because it is not affected by retrogression, in contrast to the other methods. In addition, this difference could be the consequence of the high sensitivity of CM to quickly re-equilibrating to maximum temperatures during short-duration thermal events such as the magmatic intrusions emplaced during the Mississippian thermal event.
Thermodynamic disequilibrium between white mica and chlorite has precluded their use for geothermobarometry, and a variety of data (including the existence of relic hightemperature chlorite cores, the presence of chlorite-smectite mixed layers, or the very low temperature calculated with chlorite geothermometers) indicate chlorite retrogression after the M 1 metamorphic climax and crystallization of new chlorite grains at lower temperature.
The low-pressure conditions derived from white mica indicators (very low celadonite content and b-cell values) are incompatible with the high-pressure metamorphic gradient expected in a subduction-related accretionary wedge, which has been the classical interpretation of the Pulo do Lobo domain. Instead, we interpret the Pulo do Lobo rocks as having been deposited in a platform setting located near the northern border of Avalonia, but they were never involved in the subduction-related processes of the SPZ-OMZ suture.
Data availability. The software and models used are available in the bibliography. The RSCM data set is given in Supplement.
Author contributions. IPC and DJMP collected the samples. IPC, OV, OB and FN did the analysis. All the authors contributed to the interpretation of the results and writing of the paper: IPC, DJMP, JFS and AZ focused on the regional geology, and OV, OB, FN and FB on the methodologies used.
Competing interests. The authors declare that they have no conflict of interest.
Special issue statement. This article is part of the special issue "The Iberian Massif in the frame of the European Variscan Belt". It is not associated with a conference.
Acknowledgements. We thank Valérie Magnin for her assistance with the microprobe analysis in Grenoble and Pierre Lanari for his support with thermodynamic software. Detailed revisions by Cecilio Quesada, J. B. Murphy, and an anonymous reviewer contributed to improving this paper.
Financial support. This research has been supported by the Spanish Ministry of Science and Innovation (grant no. CGL2011-24101), the Spanish Ministry of Economy and Competitiveness (grant nos. CGL2015-71692-P, CGL2016-75679-P), the Andalusian Government (grant nos. , and the doctoral scholarship to Irene Pérez-Cáceres from the Spanish Ministry of Science and Innovation (grant no. BES-2012-055754). The Raman facility in Paris has been funded by the city of Paris (Emergence program).
Review statement. This paper was edited by Emilio Clavijo and reviewed by Cecilio Quesada and one anonymous referee.