Zircon U-Pb dating and petrogenesis of the São José do Campestre Granite Complex, NE Brazil: an example of neoarchean mantle-derived post-collisional magmatism

Abstract This article discusses the geochemical and petrological evolution of the São José do Campestre granite complex (SJCgr), the last Neoarchean plutonic event so far described in the São José do Campestre massif in NE Brazil. We report field, petrographic, zircon U-Pb dating, and whole rock and mineral chemistry for representative SJCgr samples. Laser ablation zircon U-Pb data indicate that the granite emplacement took place at 2664 ± 13 Ma. The rocks comprising the SJCgr have relatively well-preserved primary textures and fabrics and compositions varying from gabbro to syenogranite. Major and trace element contents reveal a metaluminous, calc-alkaline through transitional to alkaline signature, and LILE- and LREE-enriched series analogous to late Archean sanukitoid and modern arc granitoid. The evolution of the SJCgr is envisaged as follows: (1st) partial melting of a metasomatized mantle (2.5–3.0 GPa, ~85–102 km, 1,000–1,200°C), generating a basaltic to basaltic andesitic magma; (2nd) fractional crystallization (FC) of olivine at mantle or lower crustal depth, leading to the parental magmas of the magmatic series; and (3rd) 40–15% FC of olivine gabbro-norite and olivine monzonite cumulates (400–600 MPa, 15–23 km). The SJCgr shares similarities with post-collisional granitoids and, thus, would represent the last Neoarchean episode of mantle-derived magma in Northeastern Brazil.

Recent analogs of Archean TTG, the so-called adakites, are formed in modern subduction zones and are distinguished by higher Mg, Ni, and Cr than Archean TGG (Martin 1999, Smithies 2000, Martin et al. 2005, Castillo 2006).Their generation involves a complex interaction system, in which the adakitic magma, once formed by partial melting of subducted oceanic crust, interacts with the overlying mantle and/or the lower arc crust (Defant and Drummond 1990, Drummond and Defant 1990, Sen and Dunn 1994, Maury et al. 1996, Stern and Kilian 1996, Rapp et al. 1999, Castillo 2012).
The gradual cooling of the Earth led to the progressive generation of the continental crust, of which two-thirds to three-quarters may have formed in the time range 2.5-2.0Ga Zircon U-Pb dating and petrogenesis of the São José do Campestre Granite Complex, NE Brazil: an example of neoarchean mantle-derived post-collisional magmatism (Taylor and McLennan 1985, Santosh et al. 2015, Sun et al. 2019, Ganade et al. 2021).Hence, the Archean-Proterozoic transition has long been considered a marker of relevant changes in the geothermal gradient that led to major formation of continental crust in the Archean and evolved to a regime of progressively lower geothermal gradient and dominance of crustal recycling in the Middle and Late Proterozoic (Taylor and McLennan 1985, Martin 1994, Laurent et al. 2014, Nebel et al. 2018).
In this article, we present new zircon U-Pb geochronological, whole rock, and microprobe data for the São José do Campestre granite complex (Fig. 1) and discuss the following topics: I. the evolution mechanism of a Late Neoarchean granite; II. the magma genesis and associated tectonics; III. the relative role of upper mantle and lower crustal materials as sources of magmas at that time.

GEOLOGICAL SETTING
The Borborema Province comprises several metasupracrustal sequences overlying Archean to Paleoproterozoic gneissic basement that have been subsequently intruded by several plutons during the Late Neoproterozoic (Figs. 1 and 2).This province resulted from the convergence of the West Africa -Amazonian and São Francisco -Congo cratons during the assembly of West Gondwana and extends from central and northeastern Brazil to the Trans-Saharan Orogen in northwestern Africa (Tuareg and Benino-Nigerian shields) and the Central African Orogen of Cameroon, Chad, and the Central African Republic (Caxito et al. 2020).All units forming this province are controlled and/or reworked by a complex system of continental-scale high-temperature shear zones (Caby et al. 1991, Vauchez et al. 1995, Neves et al. 2021).
Late Neoproterozoic high-temperature shear zones (Picuí-João Câmara Shear Zone) juxtapose the SJCM and the Rio Piranhas-Seridó Domain, which is composed of Neoproterozoic metasedimentary rocks (Seridó Group) overlying a Paleoproterozoic basement (Caicó Complex).Late Ediacaran structures are westward plunging synforms and antiforms in the western side of the Archean block and a southern plunging inverted antiform in the southern portion (Viegas 2007, Viegas andSouza 2007).

FIELD GEOLOGY
The São José do Campestre granite (SJCgr) crops out in the southern limit of the SJCM near the city of São José do Campestre, with the northern and eastern limits nearby Santa Maria and Boa Saúde cities, respectively (Fig. 2), in the Rio Grande do Norte state.It is a holocrystalline, medium-to coarse-grained, equigranular to slightly inequigranular, gray to pink deformed granite (Figs. 3A and 3B).A prominent subhorizontal lineation (Lγ) with a WSW-ENE trend is marked by the orientation of hornblende + biotite + K-feldspar + magnetite.Late millimetric to centimetric size magnetite + K-feldspar ± biotite-rich veins crosscut (Figs.3A and 3C) or follow the linear fabric (Fig. 3B).
Hornblende-rich quartz diorite occurs as autoliths or in interdigitated contacts with the host granite (Figs. 3C and 3D).Incomplete mixing of granite and quartz diorite occurs in some places (Fig. 3E).Igneous layering (Sγ) is characterized by alternate layers of biotite-bearing pegmatite and hornblende-biotite-bearing tonalite crosscutting an earlier magmatic linear fabric (Lγ) of the hosting SJCgr, and both units are overprinted by a NW-SE tectonic planar fabric S 1 (Fig. 3F).

METHODS AND ANALYTICAL TECHNIQUES
Analytical methods included classical field and petrographic descriptions, as well as additional whole rock and mineral chemistry for selected SJCgr samples.These data were integrated with former results from Viegas (2007), Dantas et al. (2013), Souza et al. (2016), andRibeiro (2019).
Microprobe analyses of selected amphibole, biotite, plagioclase, microcline, and Fe-Ti oxide crystals were done on a Cameca SX50 electron microprobe hosted at the Geosciences Institute, Universidade de Brasília (UnB).Analytical conditions were 15 kV of accelerating voltage, a beam current of 25 nA, a 10 s counting time, and a spot size of 1 μm.In-house standards included synthetic and natural minerals.Amphibole structural formula, Fe 2+ /Fe 3+ ratio, and (OH) calculations were done with the spreadsheet of Locock (2014), according to the IMA nomenclature scheme (Hawthorne et al. 2012).Cation proportions for biotite and feldspars were computed based on 24 and 8 oxygen molecules, respectively (Deer et al. 2013).Fe-Ti oxides were computed for 4 oxygen molecules/3 cations and Fe 2 O 3 and FeO from charge balance after Carmichael (1967).For the feldspars and biotite, all iron was assumed as FeO.The analytical errors are ± 0.5-2% for SiO 2 , Al 2 O 3 , Fe 2 O 3 , MgO, MnO, CaO, and TiO 2 and 4.5-5.6%for Na 2 O and K 2 O.All microprobe results are reported in Suppl.Data Table 1.
A selected sample (CLZ40) was analyzed for major and trace elements at ALS Minerals Laboratories (www.alsglobal.com).Major elements were measured by inductively coupled plasma atomic emission spectroscopy (ICP-AES), whereas trace element abundances were computed by inductively coupled plasma mass spectroscopy (ICP-MS) after fusion with lithium tetraborate.Loss on ignition (LOI) was determined by the gravimetric method after 1 h of heating at 1,000°C.The analytical errors are < 7% for oxides and < 6% for trace elements.Other 14 whole rock analyses were compiled from Dantas et al. (2013).Supplementary Data Table 2 displays all wholerock chemical data.

Petrography
The SJCgr rocks vary from quartz diorites to syenogranites (Streckeisen 1976), the latter being the main litotype, with variable modal contents of felsic minerals.According to Dantas et al. (2013), the modal composition of the SJCgr follows the calc-alkaline intermediate K-enrichment trend of Lameyre and Bowden (1982).
Mafic phases (up to 65% in volume) are composed mainly of deep green to brown hornblende (poikilitic grains enclosing microcline, plagioclase, and zircon; up to 2-3 mm in size; Figs.4C and 4D), plus brown biotite (< 1-5%; usually related to alteration of hornblende; locally, it may be altered to green chlorite + titanite + magnetite) and magnetite (< 1-3%; irregular, skeletal to interstitial grains up to 4 mm in length).Hornblende predominates in the less evolved quartz dioritic facies, whereas biotite surpasses hornblende in the most evolved ones.
All analyzed samples show a marked fabric defined by continuous millimetric thick bands of feldspar plus quartz alternated with mafic-rich bands (cf. Figs. 3A to 3F).A new generation of hornblende postdates earlier poikilitic hornblende, and features of recrystallization of plagioclase and microcline point to the overprinting of magmatic minerals and fabrics.

Feldspar
Plagioclase is sodic and shows no significant chemical zoning between the core and rims, and the anorthite content of five single crystals ranges from 12.68 to 15.33% (average: 13.85 ± 0.92%).Microcline has the composition Or 92-71 Ab 8-29 ; perthitic exsolutions for one grain are still more sodic (Ab 99 An 1 ).

Amphibole
The two types of amphibole crystals are hastingsite and ferro-pargasite (Fig. 5A; cf.Hawthorne et al. 2012).They have Si contents between 6.206 and 6.325 cations per formula unit (cpfu), Al between 1.937 and 1.997 cpfu, and Na+K between 0.839 and 1.003 cpfu.They are relatively homogeneous from core to rim, have low Ti contents (0.163-0.248 cpfu), and are Fe-enriched, with the fe# number [Fe/(Fe + Mg)] ranging from 0.90 to 0.91.

Geochemical characterization
The analyzed samples were divided into three groups as follows (see Suppl.Data Table 2):    Figure 6 exhibits major element plots (Barker and Arth 1976, Miyashiro 1978, Cox et al. 1979, Maniar and Piccoli 1989, Frost et al. 2001).In the TAS diagram (Fig. 6A), the samples are subalkaline (groups 1 and 2) or transitional to alkaline (group 3).This is corroborated by the SiO 2 vs. MALI plot (Fig. 6B), in which the samples scatter from calcic (the least evolved) to transitional to alkali-calcic (the most evolved).In the molar A/CNK vs. A/NK plot (Fig. 6C), which measures aluminum saturation, all samples are metaluminous.Coherently, in the ternary cationic diagram K-Na-Ca, the least evolved (group 1 and part of group 2) plot close to the trondhjemitic field, while the most evolved ones (part of group 2 and the entire group 3) tend to follow the calc-alkaline (K-enriched) trend or toward the Na-K edge (Fig. 6D).
In the fe* vs. SiO 2 plot (Fig. 7A), less evolved samples show magnesian affinity, whereas the more evolved are ferroan (SiO 2 > 65 wt.%).They scatter within the I-to A-type granites (the  9/24 most evolved ones).In the fe* vs. Al 2 O 3 diagram (Fig. 7B), most samples have been classified from slightly oxidized to reduced A-type granite, while some less evolved samples are calc-alkaline.In the ternary plot (NaO 2 + K 2 O)-5Fe 2 O 3 t-5(CaO + MgO) of Fig. 7C, these less evolved samples are akin to subducted and active continental margin related granites, whereas the most evolved correspond to A 2 -type granites that are commonly associated with post-collisional geodynamic settings.In the bilogarithimic (Y + Nb) vs. Rb diagram (Fig. 7D), the studied rocks scatter within the post-collisional field of Pearce (1996).

Binary Harker-Type diagrams
Groups 1-3 as defined above show similar behavior for TiO 2 , CaO, Fe 2 O 3 t, and P 2 O 5 (Fig. 8) that decrease regularly with differentiation (i.e., increasing SiO 2 ).Al 2 O 3 shares similar behavior, decreasing toward high silica, although the curve is less pronounced.Alkalis (Na 2 O + K 2 O) have opposite patterns as their abundances increase with differentiation.Of note, MgO content decreases strongly for group 2 but is very low (< 0.4 wt.%) and constant for group 3. Considering all samples, they define concave upward (for MgO) and concave downward (for K 2 O) curves peaking at ~65 wt.% SiO 2 .This may imply that magmatic evolution has been achieved by at least two main stages of crystallization (Wilson 1989).

Pressure and temperature
Pressure estimates were obtained from the analyzed amphibole of sample ES36 (from group 3, as defined in the geochemical section) with the Amp-TB2 (Ridolfi 2021), an updated model for the amphibole-only barometer of Ridolfi and Renzulli (2012; equation P1e), and the amphibole/plagioclase Al-Si partitioning barometer of Molina et al. (2015).The first barometer yielded pressures from 497 to 601 MPa (average = 558 ± 28 MPa), whereas the formulation of Molina et al. (2015) gave slightly higher pressures in the range of 550-718 MPa (average = 636 ± 36 MPa), calculated using temperatures derived from the Amp-TB2 thermometer (see below) and an An 14 Ab 84 Or 1 oligoclase (within the plagioclase compositional range used in Molina's calibration).Altogether, it is assumed that the SJCGR has an average confining pressure of 597 ± 46 MPa.However, a word of caution is necessary when Fe-rich amphiboles are used for pressure calculations.Anderson and Smith (1995) argued that amphiboles with fe# [Fe/(Fe+Mg)] > 0.65 are indicative of low oxygen fugacity and yield overestimated pressures.Yang (2017) proposed an empirical method using CIPW normative quartz (Qtz) and albite (Ab) plus orthoclase (Or) compositions to estimate emplacement pressure of granite intrusions, referred to as Qtz-geobarometer (Yang et al. 2019(Yang et al. , 2021)).Using the improved Qtz-geobarometer of Yang et al. (2021), which corrects pressure estimation for both normative anorthite (following Blundy & Cashman 2001) and redox estate, most of the SJCgr samples (with normative quartz < 35 wt.%) cluster around 335 ± 124 MPa.Therefore, considering all pressure estimates, we assume that the emplacement of the SJCgr took place at pressures from ~400 to 600 MPa that corresponds to emplacement depths between ~15 and 23 km (for an average crustal density of 2.7 g/cm 3 ).
The apatite (TsatAp) and zircon (TsatZr) saturation thermometric expressions after Watson and Harrison (1983) and Harrison and Watson (1984), along with the Amp-TB2 amphibole-only thermometer (Ridolfi 2021), were used to estimate magmatic temperatures for the SJCgr.The TsatAp thermometer yielded the highest estimates of 958-868°C (average = 904 ± 28°C) that should approach the liquidus temperature.The TsatZr was applied to a group of ten samples with SiO 2 ranging from 69.19 to 75.12 wt.%, in which Zr (648-212 ppm) behaves as a compatible element (see whole rock composition in Suppl.Data Table 2).Results are slightly lower than TsatAp, varying from 812 to 913 (average = 870 ± 32°C).Amphibole crystallization temperatures are between 803 and 835°C, with an average of 819 ± 7°C.Perthitic exsolution in K-feldspar point to temperature lower than 600°C and P H2O > 200 MPa (Bowen & Tuttle 1950) for the subsolidus stage.

Oxygen fugacity (fO 2 )
Redox conditions during magma crystallization can be qualitatively monitored by the fe# number [ = Fe/(Fe+Mg)] of amphibole and biotite (e.g., Anderson & Smith 1995, Anderson et al. 2008).The analyzed amphibole and biotite crystals present fe# values between 0.90-0.92and 0.89-0.9respectively, which are indicative of relatively reducing crystallization conditions.Quantitative fO 2 estimation was done with the Amp-TB2 oxybarometer (Ridolfi 2021) and the improved Qtz-geobarometer of Yang et al. (2021) using, respectively, amphibole and whole rock compositions.The Amp-TB2 oxybarometer yielded oxygen fugacity values between -14.7 and -14.1, whereas higher values were obtained with the Qtz-geobarometer (-9.9 to -13.7).These f O2 values are slightly under the NNO (Ni-NiO) buffer (-1.1 < Δ NNO < -0.5) and confirm the relatively moderate to reducing crystallization environments for the SJCgr magmas.Such a conclusion is also supported by the scattering of samples through the reduced and slightly oxidized A-type granite fields in Fig. 8B.
It is worth mentioning that magnetite is a common accessory phase in some SCJgr varieties, and this may pose a contradiction with the above-inferred redox conditions.Although magnetite-bearing granites are generally interpreted as being "oxidized" (cf.Ishihara 1981), many authors have argued that, in fact, the presence of magnetite in granites is not incompatible with a more reduced character (cf.Dall' Agnol and Oliveira 2007, Campos et al. 2016).

U-Pb zircon geochronology
Sample ES471 (Latitude 6.236 o S/Longitude 35.710 o W; see location in Fig. 3 and whole composition in Suppl.Data Table 2) of the geochemically defined group 3 was selected for in-situ zircon U-Pb dating.The sample corresponds to a slightly deformed magnetite-apatite-amphibole-bearing medium-to coarse-grained granite (Figs.4A and 4B) with euhedral zircon grains (length up to 0.9 mm), often with triangular terminations, crosscutting amphibole (Fig. 4B).
Slightly younger ages at 2605 ± 23 and 2582 ± 25 Ma (e.g., spots z26c and z26r in Fig. 11B) may represent a new generation of igneous zircon or post-magmatic lead loss.The results here obtained agree within the analytical errors with previous ages obtained by TIMS U-Pb zircon of 2683 ± 7 Ma (sample CE116B) and 2655 ± 4 Ma (sample EC80) by Dantas (1996).
Compatible (Ni, Zr) vs. incompatible (Rb) trace element correlation (Fig. 12) suggests fractional crystallization mechanism as the main process of evolution for group 3 samples (cf.Cocherie 1986, Martin 1987).Accordingly, major element least-square mass-balance modeling was done with the spreadsheet PetroMode (Christiansen 2022).Mineral compositions are those reported herein (feldspars, amphiboles, biotite, magnetite, and ilmenite), besides apatite and pyroxenes from Deer et al. (2013).Further trace element modeling followed the classical fractional crystallization equation (Rayleigh 1896): C L = C 0 F (D-1) , in which C L = concentration of the trace element in the most evolved sample, C 0 = concentration of the trace element in the least evolved sample, F = (1-FC) [FC is the degree of fractional crystallization, with FC < 1], and D = bulk partition coefficient.The partition coefficients are those provided by the database from Nielsen (2022), as well as values reported for zircon by Martin (1987), Thomas et al. (2002), Rubatto and Hermann (2007), Burnham andBerry (2012), andGudelius et al. (2020) and apatite by Prowatke and Klemme (2006).The precision of the modeling is monitored by the parameter Σr 2 .The results for both groups 2 and 3 are shown in Table 1 and Figs.13A and 13B.

Crustal contamination
It has long been demonstrated that crustal assimilation coupled with fractional crystallization (AFC) is a relevant process during evolution of mantle-derived magmas through their ascent and emplacement into the lower and upper continental crust (DePaolo 1981, Huppert et al. 1985, Wilson 1989).For geochemical modeling, we have applied mixing equations of DePaolo (1981) and trace element composition of the lower continental crust (LCC; Wedepohl 1995).Calculations for different fractions of crustal material reveal the possibility of contamination of the dioritic (group 2) magma by ~5% of LCC, followed by 30-40% of fractional crystallization (Fig. 13C).Some amount of crustal contamination seems to account for group 3 set as well (Fig. 13D), which is corroborated by negative ε Nd (2.7 Ga) of -4.02 to -6.18 (Suppl.Data Table 2).

Magma genesis and evolution
The geochemical characteristics described above suggest an upper mantle peridotite as the most probable source for the SCJgr complex.Any earlier crustal material would be impressed as a slab component (low-and high-silica adakites) hybridized with the upper mantle, leading to late Archean sanukitoids (or the ~2.6 to 2.5 Ga Closepet-type) and modern arc granitoids (Stern and Hanson 1991, Rapp et al. 1999, Smithies and Champion 2000, Moyen et al. 2001, 2003, Lobach-Zhuchenko et al. 2005, Martin et al. 2005, Nebel et al. 2018).This hybridized source should be LREE-enriched (e.g., Fig. 10).In this sense, the available ε Nd (2.7 Ga) of -4.1 to -6.2 and Nd model age (T DM ) of 3.9-3.3Ga for our samples (Dantas 1996, Dantas et al. 2013) support such an interpretation of an earlier crustal component that provided, by partial melting, the mixing component to the upper mantle peridotite (e.g., Jayananda et al. 2000, 2018, Nebel et al. 2018).
Moderate-to-low contents of Sc (25-15 ppm) in the least evolved sample (ES407, ES467) suggest the presence of minor garnet and amphibole in the source as these minerals have K D Sc >> 1 (Nielsen 2022).Through investigation of high-pressure experiments that produced saturated tholeiites, andesitic, and dacitic melts, Nicholls and Ringwood (1972) and Green (1973) suggested that they were generated by partial melting of hydrous mantle peridotite (with pargasitic amphibole) followed by different degrees of fractional crystallization of olivine at mantle or crustal depths.Moreover, petrological constraints from high-pressure experimental runs discussed by Pilet (2015) led to assume a lithosphere metasomatized by amphibole-bearing veins as the most probable source of sodic (Na 2 O/K 2 O > 1), mantle-derived magmas.This agrees with the less evolved samples having Na 2 O/K 2 O = 3.3-2.0wt.%.
Figure 14 summarizes the polybaric evolution of the studied rocks.The first stage is the partial melting of a hydrous metasomatized mantle peridotite (probably with garnet and/ or spinel, pargasitic amphibole, and phlogopite), forming a basaltic to andesitic magma at pressures of 2.5-3.0GPa (~85 to 102 km) and temperatures of 1,000-1,200°C.In the second stage, these magmas underwent olivine fractionation at mantle and/or crustal depths giving origin to our less evolved samples (ES407 and ES467).The third stage comprises fractional crystallization (± crustal assimilation) under pressures of ~400 to 600 MPa (~15 to 23 km) and temperatures > 800°C, under reducing to slightly oxidizing conditions that has left olivine gabbro-norite or olivine monzonite cumulates (Table 1).

Tectonic setting and geodynamic implication
Figure 15 shows the investigated samples in ocean ridge granite (ORG)-normalized multi-element diagram.Group 1 sample ES407 (Fig. 15A) displays a spectrum with an inclination (Rb N /Y N ) of 29.4,Yb N = 0.02, and Y N = 0.28, and no significant anomalies.These signatures share similarities with basaltic andesites of the Andean Central Volcanic Zone (Thorpe et al. 1984).
Figures 15B and 15C present four representative samples of group 2 and three of group 3.Both sets have steeper patterns and similar Rb N /Y N ratios (averages of 35.4 ± 29.2 and 35.9 ± 12.6, respectively).Group 2 may display positive anomalies for Ba (3.3 ± 2.9) and Ce (7.2 ± 5.3).These anomalies are slightly lower in group 3. Group 2 patterns can be compared to those found in the Andean Costal Batholiths (Petford and Atherton 1996) and Volcanic Arc Granites (Pearce et al. 1984), whereas group 3 has slightly higher normalized contents of Ce to Yb.
Whole rock compositions of the studied SJCgr samples indicate that these rocks correspond to slightly oxidizing to reduced A 2 -type granites (Figs.7A-7C) with some degree of crustal contamination that is reflected in their normative corundum <0.5 wt.%, and A/NK and A/CNK ratios of > 1.1 and < 1, respectively.Their trace element contents are akin to post-collisional granites, with some samples showing geochemical characteristics of volcanic arc granites (Fig. 7D).This agrees with the preservation of igneous features and rock relationships (Fig. 3), as well as microscopic textures (Fig. 4).Furthermore, the REE patterns point to a subduction tectonic setting for their genesis and evolution.In this tectonic environment, the subducting oceanic crust and the overlying mantle wedge are the main magma reservoirs.We explored this hypothesis in a SiO 2 vs. mg# binary diagram by plotting both our samples and experimental results from the literature (Fig. 16).Group 1 (sample ES407; SiO 2 53.9 wt.%, mg# 54.2) is the closest to the field of mantle-derived melts.The less evolved group 2 sample ES467 (SiO 2 58.5 wt.%, mg# 55) lies in between low-silica adakites and high-silica adakites and almost overlaps the average of sanukitoids (SiO 2 58.8 wt.%, mg# 56.8).Group 2 samples ES543A and ES572 have SiO 2 and mg# of ~65 wt.% and 46, respectively and plot between high-silica adakites and modern arc granitoids (SiO 2 ~65 to 68 wt.%, mg# ~48 to 81).Group 1 sample ES407 and the least evolved sample from group 2 (ES467)   Table 1 and Figs. 13A and 13B).
scatter among the field of hybridized melts (slab-derived melts mixed in different proportions with mantle-derived melts; Rapp et al. 1999).The remaining samples have SiO 2 > 67 wt.% and mg# < 20, falling out of the field of slab melts.This means that this kind of source cannot explain the compositional variation of the studied rocks.Hence, they may be explained as late differentiates from basic to intermediate parental liquids, as highlighted in Fig. 16.
The timing of the magma generation and the subduction/collisional process are difficult to assess (Condie 1989, Davidson andArculus 2006).Nevertheless, delamination of the continental lithospheric mantle (Rudnick 1995), underplating of basaltic magma at the base of the lower continental crust (Huppert et al. 1988), and even the action of mantle plumes (Abbott 1996, Condie 2001) could not be discarded.Nebel et al. (2018) consider that heat from ascending hot, high-Mg sanukitoid magmas, formed earlier through interaction of TTG melts with the asthenospheric mantle, triggers melting of lower crust material to generate late Archean high-K granites, the entire evolution extending for > 200 Ma.Based on the Nd model age (T DM ) of clinopyroxene from peridotite xenoliths in alkaline and kimberlitic magmas, Menzies et al. (1987) stated that the metasomatism of upper mantle peridotite by silicate melts and hydrous fluids is a very early event.According to Menzies et al. (1987), this event could take place ~0.5 to 1 Ga before the crystallization age, and even much older than 1 Ga beneath cratons.This time lapse estimation lies within the SJCgr intrusion age interval of ~2.7 Ga (Fig. 12C) and within the Nd model ages of 3.9-3.3Ga (Dantas 1996, Dantas et al. 2013).
Several episodes of juvenile accretion and crustal recycling have been suggested based on zircon U-Pb data for the São José do Campestre Massif (SJCM) by Dantas et al. (2013) and Souza et al. (2016): 3.5-3.4 Ga, 3.36 Ga, 3.25 Ga, 3.18 Ga, 3.12 Ga, 3.03 Ga, and 2.69 Ga.In an exhaustive compilation of zircon U-Pb and Nd model ages, Ganade et al. (2021) stated that the SJCM is one smaller cratonic blocks in northeastern Brazil that resulted from fragmentation of the São Francisco -Congo -West Africa cratons followed by successive collisions during the Proterozoic.In this regard, somewhat analogous events of crust formation are depicted for the Mairi complex, northern São Francisco Craton, by Oliveira et al. (2020) and Moreira et al. (2022), where banded gneisses (TTG-like felsic layers alternating with metamorphosed gabbro-diorite bands) have zircon U-Pb ages of 3.7-3.6Ga.For the same complex, Moreira et al. (2022) described successive younger events of crustal growth at 3.55-3.52,3.49-3.33,and 2.74-2.58Ga, with main crustal recycling events in the Neoarchean.Similar magmatic events and post-collisional granite emplacement are also documented in southeastern Brazil (Marimon et al. 2022) and in the Dharwar craton (e.g., Jayananda et al. 2006, Sebastian et al. 2021, Kumar et al. 2022).In the latter case, however, they assumed that the last Neoarchean episode was marked by juvenile, mantle-derived magmatism with some degree of crustal component.This could be accommodated by convergent tectonic settings (island arcs or continental margin-types) and the widespread amalgamation of earlier Archean units.Therefore, the data here obtained for the SJCgr complex show that global plate tectonics was the main process involved in the generation of continental crust in the SJCM, as also proposed for the Southern Brasília Orogen and the Southern São Francisco Craton (Marimon et al. 2022).

CONCLUSION
Integration of the data reported here and those in the existing literature led to the main conclusions listed below.
• This magmatism has major and trace element contents analogous to late Archean sanukitoid and modern arc granitoids.• The SJCgr complex represents the last Neoarchean episode of mantle-derived magma (and is related to some kind of plate tectonics) and marks the amalgamation of all earlier Archean blocks of the studied cratonic block.

ACKNOWLEDGMENTS
This article is a tribute to Professor Hervé Martin, who sadly passed away 1 year ago.ZSS is particularly indebted to Professor Hervé as his supervisor of the DESS (Diplôme d´Études en Sciences, Université de Rennes, 1991) and a post-doctoral sabbatical stage (Université Blaise Pascal, Clermont-Ferrand, 1997).In both situations, the focus was on the acquisition and interpretation of whole rock chemistry and isotopic data from Archean to Early Proterozoic juvenile granitoids of NE Brazil.Professor Hervé always made complex topics simpler with joy, good humor, and fine intelligence.He was really a great scholar and scientist.Thank you very much for all the learning you gave us.The authors acknowledge the Programa de Pos-Graduação em Geodinâmica e Geofísica and the Departamento de Geologia of the Universidade Federal do Rio Grande do Norte (PPGG/ UFRN and DGeo/UFRN) for supporting field and laboratory works.The Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) is acknowledged for financial support (grant number 408607/2018-1).The authors also thank Nilson Francisquini Botelho (Instituto de Geociências, Universidade de Brasília) for helping with mineral chemical data acquisition and fruitful discussions and Alexandre H. Santos Filho for microscopic image acquisition.

Figure 2 .
Figure 2. Simplified geologic map of the study area.

Figure 9 .
Figure 9. Trace element Harker-type plots for the studied rocks (SiO 2 in wt.%, and trace elements in ppm).

Table 1 .
Major and trace element fractional crystallization modeling of the Neorchean São José do Campestre granite complex, NE Brazil.