Metamorphic modeling and petrochronology of metapelitic rocks from the Luminárias Nappe, southern Brasília belt (SE Brazil)

The Luminarias Nappe was formed in the agglutination of West Gondwana. A high-pressure metamorphic gradient oblique to the geological contacts is recorded in metapelitic rocks from this Nappe. In the northern portion, the metamorphic peak conditions are at high-pressure lower-amphibolite facies at 580 ± 4°C and ca. 0.9 GPa (Chl + Ky + St + Ms + Qtz + Rt); in the central portion, they are at high-pressure amphibolite facies at 600 ± 15°C and 1.1 ± 0.3 GPa (St + Bt + Grt + Ms + Qtz + Rt); and in the southern portion, they reach the eclogite facies at 630 ± 13°C and 1.4 ± 0.6 GPa (St + Ky + Grt + Ms + Qtz + Rt). Clockwise metamorphic P-T-t paths are registered in the studied rocks, with temperature and pressure increase followed by a strong decompression with retrograde phases as chloritoid (northern portion), chlorite and ilmenite (central portion) and biotite, chlorite and ilmenite (southern portion). U-Th-PbT monazite ages range from 632 ± 4 Ma (southern portion) to 600 ± 8 Ma (northern portion included crystals in garnet and staurolite). The metamorphic age, the high-pressure conditions calculated in this paper and the clockwise metamorphic path indicate that the tectonic evolution of the Luminarias Nappe rocks is tightly associated with the subduction and collision processes of the southern Brasilia belt. The overprint of the younger Ribeira belt is interpreted to be responsible for rock pile tilting, thus producing the oblique metamorphic gradient.


INTRODUCTION
Pressure-temperature-time (P-T-t) paths of high-pressure rocks are key to understand the evolution of an orogen. High-pressure rocks are the main record of deep portions of continental crust formed in convergent plate boundaries. Furthermore, they provide an insight into geodynamic processes that transform the rocks from the crust and lithospheric mantle into metamorphic and igneous rocks . The root of Precambrian eroded orogens, such as the Brasília belt, Brazil, is crucial places to study high-pressure rocks. In eroded orogens, a variety of high-pressure metamorphic rocks is exposed for hundreds of kilometers, giving the opportunity to study processes that take place in convergent plate boundaries.
Thus, linking portions of the P-T path to ages, textures and equilibrium mineral assemblages are a main issue, since a well-defined P-T-t path may provide reliable information of metamorphic and tectonic processes. The concept of petrochronology has been introduced recently (Fraser et al. 1997) and deals with the fact that rocks face a complex history of heating, cooling, and exhumation and, therefore, do not record just a single age (Engi et al. 2017). This concept is especially useful in areas where orogenic processes are overprinted.
The Luminárias Nappe comprises a set of quartzite and metapelitic rocks that represent part of the passive margin sequence of the São Francisco Craton (e.g., Paciullo et al., 2000). This passive margin was metamorphosed during the Ediacaran-Cambrian orogenesis, which led to the agglutination of West Gondwana (Dardenne 2000, Fuck et al. 2017, Heilbron et al. 2017). There are controversial interpretations regarding the tectonic evolution of the Luminárias Nappe, that is, either in a single tectonic episode related to the formation of the southern Brasília belt (Campos Neto 2000, Westin et al. 2016 or as part of the interference zone between the southern Brasília and Central Ribeira belts (Peternel et al. 2005, 2013b, Heilbron et al. 2008, Coelho et al. 2017).
Previous regional studies in southern Brasília and central Ribeira belts describe a regional metamorphic gradient with conditions increasing from north to south and from east to west, from greenschist to granulite facies conditions (Trouw et al. 1980, Ribeiro & Heilbron 1982, Peternel et al. 2005, Reno et al. 2012, 2013b (with a regional Metamorphic modeling and petrochronology of metapelitic rocks from the Luminárias Nappe, southern Brasília belt (SE Brazil)  (ca. 500 Ma) showing the location of the study area (red rectangle). Extracted from Spencer et al. (2013). (B) Tectonic framework of the Ribeira and Southern Brasília belts extract from Heilbron et al. (2017). 1. Phanerozoic cover; 2. Upper Cretaceous alkaline plutons; 3 and 4 east verging units of the Brasília Belt, including the Guaxupé nappe and lower nappes; 5-7 Units of the São Francisco craton: 5. Paleoproterozoic Archean basement, 6. Neoproterozoic cratonic cover, Bambuí Group, 7. Mesoproterozoic to Neoproterozoic metasediments of autochthonous domains; 7-16 Terranes and structural domains of the Ribeira Belt: 8. Andrelândia and 9. Juiz de Fora domains of Occidental terrane, 10. Socorro Nappe; 11. Apiaí terrane; 12. Embú terrane; 13. Paraíba do Sul terrane; 14. Cambuci terrane; 15. Cryogenean to Ediacaran magmatic arc, 16. Neoproterozoic metasedimentary sucessions and 13. Tonian magmatic arc of the Oriental terrane; 12. Cabo Frio terrane. (C) Simplified geological map of the Luminárias Nappe. Modified after Trouw et al. (2013a); Quéméneur et al. (2002); Nunes et al. (2008); Paciullo and Ribeiro (2008). The geological map of Luminarias sheet is based on Almeida (1992) and mapping of undergraduate courses in the 80ths of Federal University of Rio de Janeiro. The circles (white and black) show the localization of the studied metapelite samples. The arrows in the legend of the biotite gneiss interlayered with schists and quartzite is due the different interpretations presented in the literature of the age of this unit, according Ribeiro et al. (1995) and Paciullo et al. (2000) this unit is Neoproterozoic (A1 and A2), however according Westin et al. (2016) this is a Paleoproterozoic unit (São Vicente Complex). Geographical coordinates of analyzed samples are present in Table 1. A C B is composed of quartzite, and in the Campestre Formation, which is comprised of metapelites and quartzites (Trouw et al. 1980(Trouw et al. , 1983Fig. 1). Almeida (1992) performed a detailed description of the geology, mapping the Luminárias region at 1:50,000 scale. Further papers (Paciullo et al. , 2003 interpret the Carrancas Group as two formations within the Andrelândia Mega-sequence, which is divided into six lithofacies (A1, A2, A3, A4, A5, and A6). The Campestre Formation from Luminárias Nappe corresponds ( Fig. 1) to the A4 lithofacies (Paciullo et al. , 2003. The Mega-sequence is interpreted as formed in a transition between a rift to passive margin succession, developed along the southern margin of the São Francisco paleocontinent during the Neoproterozoic (Ribeiro et al. 1995). An alternative interpretation for the provenance of the basal lithofacies (A1 and A2 -biotite gneiss interlayered with schist and quartzite; Fig. 1, from Andrelândia Megasequence) is that these rocks belong to the Paleoproterozoic and have affinity with fore arc basin and trench deposits (Westin et al. 2016(Westin et al. , 2019. According to Westin et al. (2016Westin et al. ( , 2019, the A1 and A2 lithofacies are called São Vicente Complex and are not related neither to the Carrancas Group nor Andrelândia Mega-sequence. The metasedimentary rocks from the southern Brasília belt are organized in a stack of syn-metamorphic thick-skinned nappes (Campos Neto et al. 2010). From the structurally highest levels, in the west, to the structurally lowest levels, in the east, the following sequence of nappes is recognized: • the Socorro-Guaxupé Nappe, which represents the middle and lower crust and is interpreted as the root of an intracontinental magmatic arc; • the Andrelândia Nappe System that is composed of the Três Pontas-Varginha Nappe and associated klippen, as well as the Liberdade and Andrelândia nappes (Campos Neto et al. 2010); • the Carrancas Nappe System and the Lima Duarte Nappe , Westin & Campos Neto 2013, Westin et al. 2016, also known as Lower Nappe System (Fig. 1).
The Carrancas Nappe system is the farthest east and the deepest in the pile, including deformed rocks of the former passive continental margin located at the margin of São Francisco paleocontinent (Campos Neto et al. 2010). This Nappe system is composed of rocks from Carrancas Group (Trouw et al. 1980(Trouw et al. , 1983, associated with basement slices and is divided from structural top to bottom in São Tomé das Letras Nappe, Luminárias Nappe, Carrancas-Itumirim Klippe, Serra da Bandeira allochthon, and Madre de Deus allochthon (Campos .
The Brasília belt is in contact with the Neoproterozoic Ribeira belt on its southeastern border (Hasui et al. 1975. Contrasting interpretations of this region were presented in literature, as it is either described as an interference zone between the Brasília and Ribeira belts, due to superposition of structures and metamorphism related to collision in both belts (Peternel et al. 2005, 2013b, Heilbron et al. 2008, Coelho et al. 2017, or it is considered as formed exclusively due to metamorphism and deformation associated with the Brasília Orogeny (Campos Neto 2000, Westin et al. 2016. Metamorphic conditions of Brasília belt rocks increase westwards from non-metamorphic and low-greenschist facies rocks, at the cratonic border in the east to high-temperature amphibolite to granulite facies rocks, up to ultra-high temperature conditions in the metamorphic core (Moraes et al. 2002(Moraes et al. , 2015, and decrease again westwards to amphibolite and greenschist facies towards the Goiás Magmatic Arc (Fuck et al. 2017). Previous metamorphic studies in the Carrancas Group rocks from the Luminárias Nappe show a metamorphic gradient, increasing from greenschist facies, in the north, to amphibolite facies, in the south (Trouw et al. 1980, Ribeiro & Heilbron 1982, Peternel et al. 2005, Reno et al. 2012, 2013b. According to Campos Neto & Caby (1999), the metamorphic conditions of Carrancas Group rocks (correlated with the study rocks) are high pressure and low temperature, with metamorphic assemblage garnet-kyanite-chloritoid. Silva (2010) described in the Carrancas Klippe, also composed of rocks from Carrancas Group, a metamorphic gradient that increases to the southeast, from upper greenschist facies to high pressure amphibolite facies in the transition to eclogite facies. A recent study in the metapelitic rocks of the Luminárias Nappe using the average PT mode of THERMOCALC indicates the presence of a metamorphic gradient with conditions increasing from greenschist/amphibolite facies, in the northern and center-northern portion, to amphibolite/eclogite facies in the southern portion .
Neoproterozoic ages of metamorphism, peak and retrograde, from whole southern Brasília belt vary, from 670 to 590 Ma (Vlach & Gualda 2000, Valeriano et al. 2004, Campos Neto et al. 2007, Campos Neto et al. 2010, Reno et al. 2010, Westin et al. 2016, Coelho et al. 2017, Rocha et al. 2017, Tedeschi et al. 2017. Monazite ages obtained with laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) from the Carrancas Klippe rocks (northeast of the Luminárias Nappe) are ca. 590-575 Ma (Valeriano et al. 2004, Campos Neto et al. 2010. According to Campos Neto et al. (2010, the younger ages of ca. 590 Ma determined for the Carrancas rocks represent the propagation and migration of the deformation and metamorphism through the pile of nappes, from the upper to the lower nappes, in which Luminárias Nappe is included. Alternatively, the younger ages can be interpreted as an effect of the Ribeira belt overprint.

LOCAL GEOLOGY
The Luminárias Nappe croups out in a ca. 80 km long N-S elongated nappe, near Luminárias town in Minas Gerais, southeastern Brazil. In the northern portion of the study area ( Fig. 1C), rocks from the Luminárias Nappe present an irregular exposure pattern due to low-dipping folded layers exposed on a steep topography. Observed outcrops and collected samples cover the whole extent of the Luminárias Nappe (Fig. 1, Tab. 1 and Suppl. Tab. 1).
The Luminárias Nappe was thrusted over the basement rocks, which are composed of metagranite and migmatitic orthogneiss, with lenses of mafic and ultramafic rocks. The basement is overthrusted by biotite gneiss, interlayered with schist and quartzite from the São Vicente Complex and/or the basal formation of Andrelândia Mega-sequence (A1, A2). A group of quartzite and muscovite quartzite occurs on the top of the biotite gneiss (São Tomé das Letras Formation). At its top, the Campestre Formation crops out that is composed of metapelitic rocks, which are the focus of this paper, and quartzite. The uppermost unit of the Luminárias Nappe is composed of biotite schist and gneiss from Santo Antônio Unit.
Deformational structures in the Luminárias Nappe region are grouped into three successive deformational phases (D 1 , D 2 and D 3 ). The deformational phases D 1 and D 2 are related and responsible for the main flat lying composite foliation S 1 // S 2 (Campos Neto & Caby 1999, Campos Neto 2000. Locally, these two phases can be distinguished using interference structures, such as fold and foliation overprint. Therefore, D 1 and D 2 are interpreted as representing a single progressive deformation related to later stages of subduction and continental collision . As proposed by Trouw et al. (2008), the main foliation observed in the rocks is a result of D 1 and D 2 deformational phases and is labeled here as S 1/2 . The classification of foliations and kinematic interpretations of porphyroblasts presented here are according Passchier & Trouw (2005). The S 1/2 foliation often occurs as a continuous foliation or as a slaty cleavage in the northern portion, and as schistosity in the southern portion of the Luminárias Nappe ( Fig. 2 A to I). Sometimes, S 1/2 occurs as a crenulation cleavage, where it is possible to identify the older S 1 foliation preserved in microlithons. Depending on the location of the rocks along the Luminárias Nappe, the main foliation S 1/2 is marked by different minerals, reflecting the metamorphic gradient. The relation between the porphyroblasts and the main foliation changes as well. For instance, in the northern portion, the chloritoid porphyroblasts do not present strain shadows and deflection of external foliation around the porphyroblasts (Fig 2B). This is a different situation from the garnet porphyroblasts in the southern portion, which present strain shadows and deflection of external foliation around them (Fig. 2G).

MATERIALS AND METHODS
The analytical work was performed at the laboratories of the Department of Petrology and Metallogeny of São Paulo State University, Brazil, except for the LA-ICP-MS analysis, carried out at the Micro-geochemistry Laboratory of the Department of Earth Sciences at the University of Gothenburg, Sweden.
Petrography was carried out in polished thin sections (see locations in Fig. 1 and Tab. 1) using optical and scanning electron microscopes (SEM). A JEOL JSM 6010LA SEM was used under a column, accelerating voltage of 15 kV and varied beam currents to obtain backscattered electron images (BSE).

Whole-rock chemistry and pseudosection
Whole-rock chemical compositions for 11 samples were obtained through X-ray fluorescence. Representative sample powders were mixed with lithium tetraborate to obtain fused disks, which were analyzed with a Philips PW 2400 equipment. Loss on ignition (LOI) was determined by the conventional gravimetric method.
Two P-T pseudosections were calculated for samples LR10C and LR44C, using the THERMOCALC (Powell & Holland 1988, Powell et al. 1998, version 3.40, and the internally consistent thermodynamic dataset tc-ds62 from Holland and Powell (2011), updated in February 2012. All calculations were based on the chemical model system K 2 O-FeO-MgO-Al 2 O 3 -SiO 2 -H 2 O-TiO 2 -O 2 (KFMASHTO), which represents a pelitic composition with the addition of TiO 2 and O 2 in order to evaluate rutile and ilmenite stability. As all samples lack plagioclase and present very low CaO (LR10C = 0.51 wt.%, LR44C = 0.15 wt.%) and Na 2 O (LR10C = 0.35 wt.%, LR44C = 0.28 wt.%) content, these oxides were not considered in the investigated model system. Oxygen (O 2 ) content was estimated as suggested by White et al. (2000) and, since the studied samples are ilmenite and rutile bearing free of hematite, a low O (0.11 mol.%) value was used.

Electron microprobe analysis
The chemical mineral analysis was carried out through wavelength dispersive spectrometry (WDS) using a JEOL JXA-8230 Electron Microprobe equipped with five WDS detectors. Matrix correction was carried out online by the software provided through JEOL, using the ZAF (Z as atomic number, A as absorption and F as fluorescence) method.
Garnet, chlorite, chloritoid, staurolite, biotite, and muscovite were analyzed under a 15 kV and 20 nA focused beam for the column accelerating voltage and beam current, respectively. Point analyses were carried out in the core and rim of minerals, as well as core-rim profiles, considering total counting times of 20 and 30 seconds for major and minor elements, respectively, which are equally distributed on peak and background positions. Minerals and synthetic oxides, as referenced on the laboratory, were used as standards. Mineral formulae were computed with the AX software .
Analyses of trace elements in rutile were made using a focused beam at 20 kV and 80 nA, following the method outlined by . Si, Al, Cr, Fe, Ta, Nb, and Zr were measured and Si concentrations were used as a quality control to detect and avoid zircon inclusions, as well as contaminations. Based on the recommendation of Zack et al. (2004), measurements with Si concentrations higher than 300 ppm were excluded from the dataset. R10 and Sy were applied as  Kretz (1983). HP: high-pressure.
A D E-I E-II F G-I G-II H I B-I B-II C secondary standards to ensure the quality and reproducibility of the analyses ). Zr-in-rutile temperatures were calculated with the calibration of Tomkins et al. (2007). Monazite dating (U-Th-Pb T ) has been performed in LR10E and LR44C samples and followed the recommendations of Williams et al. (2006). A full thin section x-ray map (15 kV, 200 nA, 20-50 s dwell time and 30 µm electron beam size and step) was done for Ce and P to identify all monazite crystals and examine their textural relationships with other mineral phases. On the selected monazite crystals, high resolution compositional x-ray maps were carried out for Y, Al, Th, U, Pb, Si, Ca, Fe, La, and Ce (15 kV, 100 nA, 100 s dwell time and 10 µm electron beam size and step). X-ray maps collected for crystals from the same sample were processed together (i.e., using the same color table for all crystals). Thus, concentration levels and zoning characteristics can be compared from crystal to crystal. The maps are then used to target distinctive domains for spot analyses and age calculations. Differently from the procedures highlighted by Williams et al. (2006), background measurements were performed for all analyses. Point analyses followed the method outlined by Vlach (2010), and the analytical conditions are presented in Table 2. Sample current values used for the analyses varied from 80 to 200 nA and were constantly monitored to evaluate and avoid beam damage. Every 10 to 20 analyses on the unknowns were bracketed by three analyses on the Moacir monazite secondary standard to evaluate quality of the analyses. Spectral interference corrections (Th on U Mb and Th + Y on Pb Ma) were done offline and considered matrix correction factors. Interference corrections and age calculations were performed using the Age_Cor program (Vlach 2010).

LA-ICP-MS
Trace elements analyses in quartz were conducted through a New Wave NWR213 laser ablation system coupled to an Agilent 8800 triple quadrupole ICP-MS. The carrier gas was a He-Ar mixture. Helium gas carries the laser ablated sample aerosol from the sample cell, the He gas is mixed with Ar carrier gas and N to enhance sensitivity that flows into the ICP-MS torch. It was flushed through the ablation cup at 1 mL/min. A laser beam with diameter of 10 µm at laser energy of 6.7 J/ cm 2 and a repetition rate of 4 Hz is used to ablate the sample. The signals were recorded over 60 seconds for each spot. The first 20 seconds were used to measure the background, while the next 30 ones were applied for acquiring the analysis  signal and the last 10 seconds were used for system wash out.
The following isotopes were analyzed: 7 Li, 27 Al, 29 Si, 48 Ti, 49 Ti, 57 Fe, and 72 Ge. Ti concentrations were calculated using 49 Ti to avoid the isobaric interference from 48 Ca (present in most well characterized reference glasses) on 48 Ti. The SRM 610 was used as the calibration standard. Ti-in-Quartz temperatures were calculated using the calibration of Thomas et al. (2010).

SAMPLE DESCRIPTIONS: MICROSTRUCTURES AND MINERAL ASSEMBLAGES
In the following descriptions and throughout the text, the metapelitic rocks from Campestre Formation are grouped according to their geographical distribution and mineral assemblages. In all results and discussion, quartz and muscovite are omitted from mineral assemblages, as they are excessive phases. A fluid rich in H 2 O was also in excess and present during the metamorphism. A summary of mineral chemistry of the samples is further presented. A detailed description of the mineral chemistry of the studied rocks can be found in Fumes et al. (2017).
High-pressure lower amphibolite facies: Northern portion of the Luminárias Nappe (LR04, LR50 and LR05 samples) Staurolite and kyanite-bearing chlorite-chloritoid schist; and chloritoid and kyanite-bearing muscovite schist are the metapelitic rocks from the northern portion (Fumes 2017). In both rocks, the S 1/2 foliation is a slaty cleavage/fine-grained schistosity and its texture is lepidoblastic.
In the chloritoid and kyanite-bearing muscovite schist (LR04 and LR50 samples), the rock matrix is composed of muscovite, quartz, kyanite, ilmenite, rutile, zircon, and chlorite (Figs. 2B and 2C). The foliation is slightly crenulated according to F 3 microfolds. Chloritoid occurs as porphyroblasts, with abundant inclusions of quartz, ilmenite and rutile, which are orientated parallel to main foliation S 1/2 . Based on their microstructural relationship, main cleavage overgrowing, and absence of strain shadows, chloritoid porphyroblasts are interpreted as post-tectonic to the main foliation S 1/2 (Fig. 2B) 16 , and X Fe ranges between 0.51 and 0.59.
The peak metamorphic assemblage in the northern portion is represented by Chl + Ky + St + Rt, under high-pressure lower amphibolite facies conditions. As chloritoid in LR04 sample is post-tectonic to S 1/2 (Fig. 2B), it is interpreted as a retrograde phase, which is not present in the peak metamorphic assemblage.
High pressure amphibolite facies: Central portion of the Luminárias Nappe (LR10C and LR10E samples) In the central portion, the main rock type is a staurolite-biotite-garnet muscovite schist (LR10C sample, Fig. 2D), in which muscovite, biotite and quartz define the main and homogeneous schistosity (S 1/2 ). Ilmenite, zircon, monazite, and rutile occur as accessory minerals and follow the main foliation fabric. Garnet porphyroblasts range from 1 to 5 mm in diameter, whereas staurolite varies from 1 to 2.5 mm in length. Both porphyroblasts contain abundant inclusions of quartz, monazite, ilmenite, and tourmaline. Biotite crystals occur oriented along the S 1/2 foliation, in contact with garnet as well as included in staurolite. This texture indicates that biotite is in equilibrium with garnet and staurolite (Figs. 2D and 2E). Furthermore, garnet and staurolite are interpreted as syntectonic associated with D 1/2 , because their inclusion patterns are rotated and curved (Fig. 2D). Therefore, the peak metamorphic assemblage in the central portion is Bt + St + Grt + Rt, which is stable under amphibolite facies conditions. Chlorite occurs as a retrograde phase at garnet and staurolite rims, as well as replacing biotite. Chlorite does not follow the main foliation (Fig. 2D).
LR10E sample is from the same outcrop as the LR10C sample, but this one is banded with distinct mineral assemblages in each band. One band is composed of garnet, muscovite, staurolite, and quartz (Fig. 2E), whereas the other comprises chloritoid, staurolite, quartz, muscovite and biotite (Fig. 2F). In both bands, ilmenite, rutile, monazite and zircon are accessory minerals and chlorite occurs as a retrograde phase (Fig. 2E). Garnet, chloritoid and staurolite are porphyroblasts, with average diameter of 2.0, 1.5 and 1.0 mm, respectively. The foliation is defined by the preferred orientation of muscovite, biotite, staurolite and quartz and is classified as a fine-grained schistosity (S 1/2 ). Garnet and staurolite porphyroblasts have inclusions of ilmenite, quartz, and tourmaline. Both staurolite and garnet porphyroblasts are interpreted as syntectonic to S 1/2 based on the inclusion pattern that indicates rotation and on the continuity of the internal foliation defined by inclusions (Si) and external foliation (Se // S 1/2 ). Chloritoid porphyroblasts with corroded rims are associated with euhedral staurolite (Fig. 2F) and are interpreted as pre-tectonic to S 1/2 . The compositional banding and the mineral assemblages described above indicate that the equilibrium was only attained within the same compositional band. Although mineral phases from both compositional bands may have coexisted in different domains in the rock, they are not necessarily in equilibrium with each other. Since LR10C sample has a relatively homogeneous composition (i.e., it is not banded), it is taken as a representative sample of the central portion of Luminárias Nappe.
In the central portion of Luminárias Nappe, the garnet porphyroblasts are intensely zoned with higher content of grossular and spessartine in the core, whereas almandine and pyrope increase towards the rims. Garnet formula is Fe 2+ 2,2-2,6 Mn 0-0,2 Mg 0,2-0,4 Ca 0,2-0,5 Ti 0-0,3 Fe 3+ 0-0,2 Al 1,9-2 Si 2,8-3 O 12 , and the medium X Fe has almost no variation from core to rim, from 0.86 to 0.87. The average formula of the staurolite in this sample is Fe +2 Monazite is observed in all samples from the central portion. This mineral occurs in the matrix and included in garnet, staurolite, and rutile. In LR10A sample, like LR10C, rounded crystals of allanite occur associated with anhedral monazite crystals (Fig. 3).
Eclogite facies: southern portion of the Luminárias Nappe (LR44C and LR44A Samples) Southern portion rocks are represented by sample LR44C, a garnet-staurolite schist, in which the matrix is composed of staurolite, muscovite, quartz, kyanite, rutile, ilmenite, zircon, monazite, and garnet. The latter occurs as porphyroblasts with inclusions of quartz, staurolite, ilmenite, monazite, and rutile (Fig. 2G). Oblique internal inclusions and strain shadows indicate that the garnet porphyroblasts are syntectonic to S 1/2 . The continuous schistosity is defined by preferred orientation of staurolite, quartz, and kyanite. Kyanite crystals are often observed associated with staurolite ( Fig. 2H), which indicate equilibrium between the two minerals. Therefore, textural relationships show that the metamorphic peak assemblage is St + Ky + Grt + Rt, a high pressure, amphibolite to eclogite facies assemblage. Biotite and chlorite are interpreted as retrograde phases, since both minerals occur with random orientation in the matrix or at the garnet rims (Fig. 2I).
Garnet porphyroblasts from the southern portion are also zoned with representative formula of Fe 2+ 1,6-2,5 Mn 0-0,6 Mg 0,1-1,2 Ca 0-0,5 Ti 0-0,5 Fe 3+ 0-0,8 Al 1,8-2 Si 2,6-3 O 12 ; in the cores, X Fe is 0.90, and in rims, it is 0.89. Garnet cores present higher content of spessartite and lower almandine percentages, whereas pyrope and grossular maintain constant concentrations along sections through all the garnet crystals. Staurolite average composition is Fe    Figure 4. The projection is in the system KFMASHTO, and the minerals involved in the calculations are quartz, muscovite, garnet, chloritoid, staurolite, ilmenite, rutile, chlorite, kyanite, sillimanite, K-feldspar, and water. Quartz and muscovite are ubiquitous in the studied rocks and metamorphism took place under sub-solidus P-T conditions. Therefore, quartz, muscovite, and water were set to be in excess. The two P-T pseudosections (LR10C and LR44C samples) have similar topology and were calculated between 0.4-1.8 GPa and 450-700°C (Figs. 4A and 4B). In both diagrams, the chloritoid stability is exclusively controlled by temperature, and the mineral is stable at temperatures below 590°C. The occurrence of chlorite is also controlled by temperature. In the pseudosection calculated for LR10C sample, chlorite happens along all the P range and in T lower than 615°C. In the pseudosection calculated for LR44C sample, chlorite occurs under P conditions lower than 1.4 GPa and T in lower than 580°C. For both samples, staurolite occurs in T between 505 and 690°C and P in lower than 1.7 GPa. Garnet is stable in P higher than 0.8 GPa and T higher than 490°C in the pseudosection of the LR10C sample and 555°C in the pseudosection of LR44C sample. Biotite occurs at P lower than 1.1 GPa and at T higher than 480°C in the pseudosection of LR10C sample and 530°C in the pseudosection of LR44C sample. Kyanite and sillimanite are the aluminum silicates that appear in pseudosections within the investigated P-T range. Kyanite is stable at T higher than 580°C and P higher than 0.8 GPa, whereas sillimanite is only stable at P lower than 0.8 GPa and T higher than 580°C. Ilmenite and rutile are the Ti-bearing phases, the former is stable at P lower than 1.0 GPa and T lower than 500°C and the latter is only stable in conditions of P higher than 1.0 GPa and T higher than 500°C. Muscovite, quartz, and water are stable and appear in almost all the fields, apart from a small P-T field at T higher than 575°C and P lower than 0.5 GPa, where the K-feldspar is stable, instead of the muscovite.
According to the calculated pseudosections (Fig. 4), the mineral peak assemblage of the central portion of Luminárias Nappe (St + Bt + Grt + Rt, abbreviations according to Kretz 1983) is stable from 590 to 685°C and from 0.9 to 1.2 GPa. The mineral peak assemblage that represents the southern portion of the Luminárias Nappe (St + Ky + Grt + Rt) occurs in a tight diagonal field and is stable from 585 to 690°C and from 0.9 to 1.65 GPa, which corresponds to high pressure amphibolite -eclogite facies transition.

SINGLE ELEMENT THERMOBAROMETRY
Single element thermobarometry is applied as an independent technique to constrain the metamorphic peak. The geothermometers applied in this study are Zr-in-rutile and Ti-inquartz, and both are suitable to be applied here since all rocks contain quartz, zircon and rutile in excess.

Trace elements in rutile and quartz
Rutile crystals from five samples, LR04, LR05, LR10C, LR10E and LR44C, were analyzed. Data from the trace elements in rutile obtained for all the studied samples are presented in Table 3    DL: 2 sigma minimum detection limit (in ppm); RSE: Relative standard errors from counting statistics (2 sigma); T (Temperature) is calculated using equation of Tomkins et al. (2007); P (Pressure) estimate based on the pseudosection; *: data plotted in Figure 5 (grains with more than 1 spot); T (°C) + and T (°C) -: errors in temperature calculated based on the RSE, BD: bellow detection limit.
In LR04 and LR05 samples, from the northern portion of Luminárias Nappe (high pressure lower amphibolite facies mineral assemblage, since the mineral peak assemblage is chlorite + kyanite + staurolite + muscovite + quartz + rutile), rutile is abundant and occurs as subhedral crystals in the matrix. Most rutile crystals contain ilmenite needles and zircon inclusions (Figs. 5A, 5B and 5C).
In LR10C and LR10E samples, from the central portion (high-pressure amphibolite facies mineral assemblage), rutile is rare, has anhedral shape and is frequently associated with ilmenite in a texture indicating that the latter replaces rutile (Fig. 5E). Rutile does not occur as inclusion in garnet, and several crystals have zircon inclusions. Due to the characteristics described above, only 6 out of 23 analyses of LR10C and LR10E samples passed adopted quality control, that is, Si content below 300 ppm.
In LR44C sample, from the southern portion (eclogite facies mineral assemblage), rutile crystals are mostly free of inclusions, although sometimes a few ilmenite needles and zircon inclusions are observed. Rutile crystals have subhedral shape and occur either in the matrix or as inclusion in garnet (Fig. 5D). Rutile crystals from the northern portion samples (LR04 and LR05) have similar composition of trace elements and present the highest content of Nb (several rutile crystals from sample LR44C also have high Nb content), Fe, Al and Cr among all the studied samples (Tab. 3 and Fig. 6). High values of Nb are not followed by Ta. The average Zr concentration is the lowest among the studied samples, although maximum (148 ppm, LR04) and minimum (52 ppm, LR05) values are like those of LR10C and LR10E samples. It is noteworthy that Zr was below the EPMA detection limit (45 ppm) in one analysis of LR04 sample and six analyses of LR05 sample.
In the central portion samples (LR10C and LR10E), rutile crystals have the lowest content of Nb, Fe, Cr and Ta among the studied samples (Tab. 3 and Fig. 6). Al content is similar to LR44C sample. As presented above, maximum (149 ppm, LR10E) and minimum (74 ppm, LR10C) Zr content in rutile from LR10C and LR10E samples are like those of LR04 and LR05 samples, although the average is higher.
Differently from the other samples, rutile occurs as an inclusion in garnet and in the matrix in LR44C sample (southern portion). Crystals from the two textural contexts have distinct composition of trace elements. Those crystals included Figure 6. Boxplots showing concentration (in ppm) of the trace elements (Nb, Cr, Fe, Ta, Al and Zr) in rutile crystals from studied samples. In LR44C sample data from matrix rutile (Mtx) and included in garnet rutile crystals (Incl) are individualized. Whiskers represent the 5th and 95th percentile. Boxes represent the second (bottom-25%) and third quartile (top-75%). The square and bar inside the boxes represent the average and the media, respectively. The numbers on top of each box represent the number of analyses that are showed in the graph. When the numbers of analyses above detection limits are equal or lower than three only the value of the analyzed spots are plotted (solid squares). Only analyses above the minimum detection limit are presented. in garnet have the highest content of all trace elements when compared to those from the matrix. All analyses of Ta and Zr are below the EPMA detection limit (85 and 45 ppm, respectively) for matrix rutile. Zr content in included rutile crystals from LR44C sample is the highest among all analyzed samples, with average of 159 ppm and maximum concentration of 192 ppm.
Quartz trace element concentrations were measured by LA-ICP-MS from LR44C sample, southern portion (Tab. 4). Only the quartz from this sample was analyzed because Zr-in-rutile content shows a small spread and the sample is from the higher metamorphic conditions of the Luminárias Nappe. Ti content ranges from 6.6 to 11.3 ppm.

Temperature calculations
Isopleths for the Zr-in-rutile (Tomkins et al. 2007) and the Ti-in-quartz (calibration of Thomas et al. 2010) geothermometers are presented in Figure 7. Since both geothermometers are pressure-dependent, it is crucial to have a pressure estimation to calculate temperature. However, if both Zr-inrutile and Ti-in-quartz geothermometers are applied for the same sample, equilibrium pressure and temperature conditions are given by the crossing of the isopleths in P-T space. According to Tomkins et al. (2007), a conservative estimation of temperature is given by the upper end of the box-plot box (third quartile). Temperatures presented herein are calculated according to this recommendation.   Taking into account the stability field for the peak assemblage (Bt + St + Grt + Rt) in LR10C and LR10E samples (Fig. 4) and the Zr-in-rutile isopleths (Fig. 7B), the results indicate only the highest concentration of Zr plot in the stability field, defining a pressure range from 1.0 to 1.1 GPa. For this pressure range, temperatures calculated for the third quartile content (149 ppm) are 597 ± 3 and 601 ± 15°C, respectively.

Li Al Ti Fe Ge T (°C) T (°C) + T (°C) -RSE-Al RSE-Ti RSE-Fe
For LR44C sample, the Zr-in-rutile and Ti-in-quartz isopleths intercept each other at ca. 630 ± 13°C and 1.4 GPa (third quartile values for both elements). These results are in compliance with the stability field for the peak assemblage of the rock (Grt + St + Ky + Rt - Fig. 7).
For the northern portion, high pressure lower amphibolite facies conditions are constrained by the peak mineral assemblage Chl + Ky + St + Rt. This occurs along a wide pressure window (see KFMASH petrogenic grid in White et al. 2014). However, since results for LR10C/E and LR44C confirm that the metamorphic conditions decrease to the north, an estimated pressure of 0.9 GPa was used to calculate Zr-in-rutile temperatures for the northern portion. For LR04 sample, the calculated temperature is 580 ± 4°C (125 ppm -third quartile).

MONAZITE GEOCHRONOLOGY
U-Th-Pb T monazite dating was carried out in samples of the central (LR10E) and southern (LR44C) portions and comprised crystals from the rock matrix, as well as included in garnet, staurolite, kyanite, and rutile. Major and trace elements analyses, as well as compositional maps of 32 crystals were carried out ( Fig. 8 and Tab. 5).
A compositional zoning of Y, Th, Pb and U is observed in several monazite crystals (Fig. 8), but it has no effect on the calculated ages. Systematic variations on the mean ages are linked to textural monazite settings in LR10E sample, in which monazite crystals included in garnet and staurolite present older ages (615 ± 6 Ma) when compared to those in the matrix (600 ± 8 Ma) ( Fig. 9 and Tab. 6). In LR44C sample, results for matrix and included monazites crystals (in garnet, staurolite, kyanite and rutile) are indistinguishable, and the mean age is 632 ± 4 Ma. Monazite crystals from LR44C sample have the highest content of Y among all the analyzed samples (Fig. 10).
Allanite occurs associated with the monazite in LR10A sample (Fig. 3). All monazite analyses in this sample were discarded due to the low total (sum of all analyzed elements) and no coherent U-Th-Pb T ages. Although the analyses are not used in further discussion, the presence of allanite replacing monazite is useful to establish the metamorphic P-T path that is further discussed.

Metamorphic conditions and P-T paths
In the present study, the metamorphic gradient that has been previously described for the Luminárias Nappe region (Trouw et al. 1980, Ribeiro & Heilbron 1982, Peternel et al. 2005, Silva 2010) is characterized based on peak metamorphic mineral assemblage, pseudosection modeling and geothermometry of trace elements. It is the first attempt to quantify, by a multi-method approach, the peak metamorphic conditions of the studied rocks. Furthermore, P-T paths are presented based on pre-peak, peak and post-peak mineral assemblages (Fig. 11). Peak mineral assemblages indicate that the metamorphic grade increases from high pressure lower amphibolite facies conditions in the north (Chl + Ky + St + Rt) to high pressure amphibolite facies in the central portion (St + Bt + Grt + Rt), and to eclogite facies in the southern portion (St + Ky + Grt + Rt). The high-pressure Al-silicate, kyanite, is stable in the studied rocks, consistent with the metamorphic high-pressure conditions which the samples were submitted.    of plagioclase, quantification of pressure conditions is problematic. Even when mineral composition isopleths are calculated and presented in the respective fields of the pseudosections, they form sub-parallel lines, and their crossing results in large uncertainties when pressure is calculated. Post-tectonic chloritoid (to S 1/2 ) in LR04 sample (Fig. 2B) is interpreted as a retrograde phase in respect to peak condition attained under high pressure lower amphibolite facies (northern portion). In the high pressure amphibolite facies rocks from the central portion (Fig. 2D), chlorite overgrows the main foliation and is interpreted as retrograde. Chloritoid porphyroblasts associated with staurolite are present in a specific compositional layer in LR10E sample (Fig. 2F). Two possible interpretations for this association are presented here. The first hypothesis is that chloritoid occurrence is controlled by the chemical composition of the layer. Another one is that the chloritoid is pre-tectonic and was preserved only in this band. Since staurolite has a subhedral shape, we interpret that staurolite is replacing chloritoid. We acknowledge that the metamorphic texture is complex and an opposite interpretation, in which chloritoid replaces staurolite, cannot be ruled out.

Sample
The metamorphic peak assemblage observed in samples from the southern portion of Luminárias Nappe is represented by St + Ky + Grt + Rt, which indicate the highest temperature and pressure condition observed in the region. According to the calculated pseudosection, this metamorphic peak assemblage   is formed under eclogite facies conditions (Fig. 11). Staurolite occurs as inclusions in garnet porphyroblasts. Kyanite also occurs as inclusions in garnet, but only in the rim. Both staurolite and kyanite happen as oriented crystals in the matrix defining the foliation (Fig. 2G). Biotite and chlorite with no preferred orientation are observed in several samples from the southern portion and are interpreted as retrorade phases. Taking into consideration mineral assemblages, the textural relationships and the stability fields obtained in the pseudosection, clock-wise P-T paths can be defined for the studied samples (Fig. 11). No pseudosection has been calculated for northern portion samples (LR04 and LR05). Peak T condition (580 ± 4°C at ca. 0.9 GPa) is given by the Zr content in rutile from LR04 sample and is compatible with the peak mineral assemblage Chl + Ky + St + Rt. The post-peak path is characterized by the crystallization of post-tectonic chloritoid (Fig. 2B). For the central portion (LR10C and LR10E samples), the P-T path starts in the stability field of the Chl + Cld + Grt + Rt assemblage (Fig. 11B). The replacement of chloritoid by staurolite (Fig. 2F) and the biotite presence (Fig. 2D) indicate that the P-T path crosses the Chl + St + Grt + Rt field, reaching the Bt + St + Grt + Rt field in peak metamorphic conditions. Post-peak metamorphism is evidenced by chlorite and ilmenite crystallization (Fig. 2E) in the Chl + St + Grt + Ilm field (Fig. 11B). For the southern portion (LR44C sample), the P-T path starts in the stability field of the St + Grt + Rt assemblage, reaches the metamorphic peak at the St + Ky + Grt + Rt assemblage field, crossing through Bt + St + Grt + Rt and Bt + St + Grt + Ilm fields, finishing in the retrograde assemblage Chl + St + Grt + Ilm field (Figs. 4B and 10C).
Metamorphism temperature in the Luminárias Nappe rocks is close to the temperature of resetting of Zr in rutile, at ca. of 550-650°C (Cherniak 2000, Cherniak et al. 2007, Treibold et al. 2007, Triebold et al. 2012, Kooijam et al. 2010, Kohn et al. 2016. Although the Zr content in rutile from LR04 sample provides consistent results, the registered temperature is lower than the expected re-equilibration rutile temperature. It is noteworthy that the temperature of ca. 580°C obtained for this sample agrees with the peak mineral assemblage (KFMASH petrogenic grid in White et al. 2014). According to Cruz-Uribe et al. (2018), single element thermometers in temperatures lower than 600°C can only be applied if trace element equilibrium can be demonstrated. This is the case of the studied samples, in which quartz, rutile and zircon are present, buffering the equilibrium of the SiO 2 -TiO 2 -ZrO 2 system.
In LR44C sample, only rutile crystals included in garnet have Zr content above the EPMA detection limit, indicating that post-peak metamorphism may have changed the rutile composition in the matrix. Higher content of Zr in rutile is often observed in crystals included in garnet and kyanite (Zack et al. 2004, Hart et al. 2018. With the trace element data, a correlation between Nb content in rutile and the occurrence of biotite is interpreted. Higher Nb content in rutile is correlated with the absence of biotite (rutile in LR04 and LR05 samples and rutile included in garnet in LR44C sample, where peak assemblage is biotite free). As pointed out before by , high Nb content in rutile can be linked to rutile growth associated with biotite breakdown during prograde metamorphism, since rutile favors Nb over Ti, when compared to biotite. In LR10C and LR10E samples, rutile with lower Nb content occurs in the matrix where rutile coexists with biotite in the peak metamorphic assemblage. In LR44C sample, biotite is retro-metamorphic and only occurs in the matrix where the rutile also presents lower Nb content. According to our data (Fig. 4B), biotite retro-metamorphic crystallization in the presence of rutile in LR44C sample occurs in T conditions between 590°C (lowest T stability in Fig. 4B) and 630°C (peak T condition) and P conditions lower than 1.05 GPa (highest P stability in Fig. 4B).
Pressure conditions of the metamorphic peak are higher than the Barrovian metamorphism. High-pressure conditions were also obtained by Silva (2010) in the Carrancas Klippe, northeast of the Luminárias Nappe. The author calculated peak P-T conditions (1.0 ± 0.17 GPa and 577 ± 8°C in the north and 1.29 ± 0.1 GPa and 608.5 ± 19.5°C in the south), which are similar to those presented here, and somewhat higher in pressure than previously calculated for these units (Campos Neto & Caby 1999).
The basal unit gneiss from the São Vicente Complex and/ or A1/A2 (Fig. 1) has plagioclase. Lenses of amphibolite are common in this unit. Mineral assemblages in both rocks are incompatible with eclogite facies metamorphic conditions. It may suggest that either the high-pressure conditions calculated here are overestimated or that the Carrancas Group was thrusted over the São Vicente Complex after metamorphic peak conditions, and that they are, therefore, unrelated as suggested by recent data (Westin et al. 2016(Westin et al. , 2019.

Age of metamorphism
Three distinct U-Th-Pb T monazite ages ( Fig. 9) have been obtained for the Luminárias Nappe rocks: 600 ± 8 Ma for matrix A B C monazite in LR10E sample (central portion), 615 ± 6 Ma for monazite included in peak metamorphic minerals in LR10E sample, and 632 ± 4 Ma for matrix and included monazite in LR44C sample (southern portion). The textural monazite setting in LR44C sample has no influence on the results and the age (average of 632 ± 4 Ma) is older than those obtained for LR10E sample. The results indicate, therefore, that a single monazite growth episode is registered in rocks from the southern portion. There are two possibilities to explain the older age. It may represent the age of the metamorphic peak associated with the collision of the southern Brasília belt. According to Campos Neto et al. (2004, rocks from the Carrancas Nappe System are the youngest among all related nappes from the southern Brasília belt, in a continuous migration of the orogen model. Rocks from the Carrancas Nappe System can be lithologically and stratigraphically correlated with the Luminárias Nappe rocks. The age (632 ± 4 Ma) determined in this paper for LR44C sample is, however, older when compared with those from Carrancas Nappe System (590-575 Ma, U-Pb monazite ages, Valeriano et al. 2004, Campos Neto et al. 2011. Alternatively, the monazite crystals in LR44C sample may be interpreted as detrital, deriving from metamorphic units within the southern Brasília belt, for example, Socorro-Guaxupé Nappes (Rocha et al. 2017), Passos Nappe (Valeriano et al. 2004), Carmo da Cachoeira Nappe and Três Pontas-Varginha Nappe (Reno et al. 2010(Reno et al. , 2012. Reno et al. (2012) have presented a comprehensive set of monazite ages for the southern Brasília belt. Oldest ages (662-665 Ma) were obtained for high-pressure granulite from Três Pontas-Varginha Nappe and were interpreted to represent near-peak conditions. Ages of 640-631 Ma are obtained for high yttrium monazite from the same unit and are interpreted as monazite growth from local garnet breakdown. These ages are like those obtained for LR44C sample (632 ± 4 Ma). In addition, monazite crystals from LR44C sample have the highest yttrium content among all analyzed grains (Fig. 10). Younger ages (616-613 Ma) are presented by Reno et al. (2012) for the Carmo da Cachoeira Nappe and the Carvalhos Klippe. For these units, ages younger than 605 Ma are also obtained and are interpreted to be related to the orogenic loading associated with terrane accretion in the Ribeira Belt.
The age difference between included and matrix monazite in LR10E sample is quite significant (ca. 15 Ma) and may indicate changes in the tectonometamorphic conditions along the P-T-t path. Records of prograde metamorphism steps in garnet porphyroblasts are confirmed by their chemical zoning, as documented by Fumes et al. (2017). The older age presented by the included monazite represents the time of garnet and staurolite porphyroblasts crystallization at metamorphic peak conditions. Younger ages are interpreted to represent post-peak monazite crystallization in the rock matrix. A possible interpretation is that monazite grew during a single metamorphic event, during the retrograde part of the P-T-t path. An alternative interpretation is that the area faced two metamorphic events, an older one at ca. 632-615 Ma, related to the collision of the southern Brasília belt, and a younger one at about 600 Ma related to collision of the Ribeira belt, as proposed in literature (Trouw et al. 2013b, Coelho et al. 2017, Reno et al. 2012, Peternel et al. 2005, Vinagre et al. 2016. Replacement of allanite by monazite is observed in LR10A sample (Fig. 3). LR10A sample is related to the central portion of Luminárias Nappe; it is similar to LR10C sample, a staurolite-biotite-garnet schist. The metamorphic peak assemblage is St + Bt + Grt + Rt and corresponds to high pressure amphibolite facies. According to Gowami-Banerjee & Robyr (2015), the monazite-forming reaction occurs at temperatures higher than ~600°C, in Barrovian terranes. Although the pressure conditions in Luminárias Nappe rocks are higher than Barrovian terranes, the temperature of 600°C is near to the peak temperature observed P-T-t path of rocks in the northern portion from Luminárias Nappe. According to the calculated pseudosection (Fig. 4A), an invariant curve occurs at ca. 600°C, in which chlorite becomes instable and biotite is crystalized. The allanite also probably became instable and the monazite was crystalized close to this reaction.

Tectonic implications of P-T-t path and ages
Pressure conditions of peak metamorphism registered in the Luminárias Nappe rocks indicate that equilibrium took place under conditions that are higher than those of typical Barrovian metamorphism. Therefore, it points to a thick-skinned tectonic setting that is coherent with the continental collision, which may involve subduction of continental crust, proposed for the southern Brasília belt (Dardenne 2000, Fuck et al. 2017, Heilbron et al. 2017, Tedeschi et al. 2017. The ages presented here (Fig. 9) for the Luminárias Nappe rocks (615 ± 6 Maincluded crystals, 600 ± 8 Ma -matrix crystals for LR10E sample and 632 ± 4 Ma -matrix and included crystals for LR44C sample), are within the time interval of the events that led to the formation of both the southern Brasília belt (630-607 Ma is the age of metamorphic peak according to Rocha et al. 2017, Mora et al. 2014, Coelho et al. 2017, Tedeschi et al. 2017 and the Ribeira belt (620-520 Ma according to Machado et al. 1996, Bento dos Santos et al. 2007, Heilbron et al. 2017. The oldest ages obtained in the present paper are, therefore, in agreement with processes associated with the Brasília orogeny, while younger ages may be related to retrometamorphism or to processes associated with the Ribeira orogeny, extending the interference zone to the study area. Based on the orogen migration model proposed by Campos Neto et al. ( , 2010Neto et al. ( , 2011 for the southern Brasília belt, the Carrancas nappe system, which encompasses the studied rocks, corresponds to the last stage of the collision pile and has metamorphic monazite ages of ca. 590-575 Ma. The ones presented in this paper, especially those from the southern portion and from the included monazite crystals in the central portion, are considerably older than the ages presented by the authors and, therefore, are inconsistent with the model proposed by Campos Neto et al. (2011). However, a word of caution with respect to monazite ages seems appropriate, considering that many reports on such ages show a wide range of results casting some doubt as to the precise meaning of an average monazite age (e.g., Reno et al. 2012, Campos Neto et al. 2011 and this paper).
In addition, recently published data (Tedeschi et al. 2017) on high pressure amphibolite facies mafic rocks from Pouso Alegre (SW of the Luminárias Nappe) indicate that these rocks were metamorphosed at peak P-T conditions of 690 ± 35°C and 1.35 ± 0.30 GPa. The authors interpret that these rocks represent the deep root of the Ediacaran (630 Ma, U-Pb in zircon and monazite) continent-continent collision zone associated with the Brasília belt. The age and the metamorphic conditions presented here for the southern portion of the Luminárias Nappe agree with the data presented by Tedeschi et al. (2017) and may reflect the continental collision associated with the Brasília Belt.
Based on metamorphic and geochronologic data of metamorphic and magmatic rocks from Andrelândia Nappe System (SE of the Luminárias Nappe), Coelho et al. (2017) described an orogen-scale interference model for the area, with continental subduction at 625 ± 6 Ma and nappe emplacement at 618 ± 5 Ma, which are both associated with the Southern Brasília Orogen and a second heat pulse associated with Central Ribeira Orogen at 586 ± 9 Ma. Although more detailed geochronological studies need to be carried out, the ages obtained for the southern portion of the Luminárias Nappe (632 ± 4 Ma), for the included monazite crystals (615 ± 6 Ma) and for the matrix monazite crystals from the central portion (600 ± 8 Ma), agree, respectively, with the steps of subduction, nappe emplacement, and heat pulse associated with the Central Ribeira Orogen proposed by Coelho et al. (2017).
Sillimanite replacement over kyanite is described in rocks from the southern portion of Luminárias Nappe (Ribeiro & Heilbron 1982, Peternel et al. 2005, Reno et al. 2012, Trouw et al. 2013b and is interpreted as a post-peak superposition of deformation and metamorphism associated with the Ribeira belt. Caputo Neto et al. (2018) have recently presented data on metasedimentary rocks formed between the late collisional stage of the southern Brasília belt and the main collision in the central Ribeira belt (max. deposition at ca. 611 Ma, evolution range between 611 and 580 Ma). The data presented by the authors suggest that an important uplift through erosion took place in this interval. The decompression associated with the uplift, followed by heating associated with the main collision in the central Ribeira Belt may be responsible for the sillimanite replacement over kyanite. The fact that sillimanite does not occur in any of the studied samples may be related to the whole rock composition. As seen in Figure 4, the lower limit of the peak stability fields for rocks from the central (Bt + St + Grt + Rt) and southern portions (Grt + St + Ky) is strongly controlled by P. A decompression to conditions of about 1.0 to 0.9 GPa would lead to the ilmenite crystallization, in the St + Bt + Grt + Ilm stability field, in which neither kyanite nor sillimanite would be present. Ilmenite is indeed present in the studied rocks. An even greater decompression would destabilize garnet in the St + Bt + Ilm field but, still, no aluminum silicate would be present. According to our modeling, a further decompression to conditions of approximately 0.7 GPa, at temperatures between 630 and 680°C, would be required for the sillimanite crystallization in the studied samples.
The regional metamorphic gradient that is described for the Southern Brasília belt rocks, at the interference zone with the Ribeira belt, is oblique to the main geological contacts (Trouw et al. 1980, Ribeiro & Heilbron 1982, Peternel et al. 2005, Reno et al. 2012. This observation is corroborated by our results, since P-T-t conditions increase from north to south along the Luminárias Nappe rocks. One possible scenario to explain the high pressure signature of the studied rocks, as well as the oblique position of the metamorphic gradient in respect to the geological contacts, would be a progressive event that starts with the deformation and metamorphism associated with the Brasília belt and evolves to a collisional, or near collisional, stage (older monazite ages). Our data does not show clearly if the overprint of the younger Ribeira belt occurred before the rocks were exhumed to shallower settings or after uplift and erosion of the Brasília Orogen. In any case, the overprint would be responsible for the oblique metamorphic gradient (tilting of the rock pile in respect to the isotherm and isobar), as well as the younger ages (monazite in the matrix).
The overprint of two distinct tectonic events is not undoubtedly confirmed by the data presented here. It is still possible that the evolution of Luminárias Nappe rocks is associated with one single event entirely related to the Brasília belt event.

CONCLUSIONS
The metamorphic gradient described in literature for the Luminárias Nappe rock is confirmed through the present paper. In the northern portion, the peak metamorphic condition is ca. 580 ± 4°C and ca. 0.9 GPa (high pressure lower amphibolite facies); in the central portion, it is ca. 600 ± 15°C and 1.1 ± 0.3 GPa (high pressure amphibolite facies); and in the southern portion, it is ca. 630 ± 13°C and 1.4 ± 0.6 GPa (eclogite facies).
Petrographic data, combined with metamorphic modeling, thermobarometry calculations and U-Th-Pb T monazite dating indicate that the Luminárias Nappe rocks followed a continuous, single step, clockwise P-T-t path.
The peak metamorphic age from Luminárias Nappe is 615 ± 6 Ma for the central portion of Luminárias (LR10E sample) and 632 ± 4 Ma for the southern portion (LR44C sample) according to the included monazite. The age of monazite crystallization in the matrix in the central portion is ca. 15 Ma younger, and there is no difference between the included monazite and matrix monazite in the southern portion.
The integrated approach used in this paper, combining pseudosection modeling and single element thermometers (Zr-in-Rt and Ti-in-Qtz), is a useful tool to constrain the metamorphic conditions.
According to the petrological textures, deformation features, metamorphic conditions and ages presented in this paper, the tectonic evolution of Luminárias Nappe rocks is tightly related to the east-verging orogenic processes of the southern Brasília belt. The effect of the superposition of the younger, northwest-verging, Ribeira belt is interpreted to have caused the tilting of the rock pile in respect to the isotherm and isobars, leading to the formation of an oblique metamorphic gradient. Since kyanite is the only stable Al-silicate, the studied rocks were deformed and metamorphosed under high pressure conditions, which is in compliance with the calculated values of 630°C and 1.4 GPa for the highest-grade rocks.