Mafic microgranular enclaves (MMEs) are a ubiquitous feature of post-collisional magmatism, receiving much attention among earth scientists over the last decades. While recent advances point to the large-scale involvement of the lithospheric mantle in granite petrogenesis, MMEs have received less attention in such discussion. Because MMEs are commonly acknowledged to represent the mafic end member with a mantle affinity that is related to early-stage batholith petrogenesis, they constitute a good proxy for the mantle role in the process. Using MME data from Los Pedroches batholith in southwestern Iberia, we conduct a geochemical comparative study between MMEs and the mafic-intermediate (sanukitoid) suite of post-collisional batholiths. An accurate overlap between the two groups is revealed, implying a potential genetic link between MMEs and the sanukitoid suite. Together with evidence from experimental cotectic liquids, the link between the high-Mg signature of postcollisional magmas and the predominance of amphibole in the studied MME samples is used to account for the composition of post-collisional magmatism. Implications for post-collisional batholith petrogenesis is then discussed in a qualitative manner, suggesting a heterogeneous yet common two-stage origin for all post-collisional magmatism in which the relationship between MMEs, sanukitoid, and the host felsic magmas is a differentiation process, thus representing a major input of juvenile magma into the crust.

Fine-grained and dark mafic microgranular enclaves (MMEs) are a characteristic feature of post-collisional granodiorites and granites, together with the characteristic mafic-intermediate and felsic suites of post-collisional magmatism (e.g., Couzinié et al., 2016; Gómez-Frutos et al., 2023; Moyen et al., 2017). Although they share many geochemical features with their host granitic magmas, MMEs have a stronger mantle affinity (Clemens et al., 2017; Holden et al., 1987), implying extensive mantle-crust interaction in the petrogenesis of the granite magmas. However, even though the petrogenetic significance of MMEs has been studied for several decades (e.g., Barbarin, 2005; Barbarin and Didier, 1992; Paterson et al., 2004; Vernon, 1984; Wiebe, 2016), a satisfactory explanation that meets all observations for such mantle-crust interaction and for their relationship with host granites remains elusive (see Clemens et al., 2017, for a review). While different interpretations for the petrogenesis of MMEs exist (Barbey et al., 2008; Rodríguez and Castro, 2019; Vernon et al., 1988; Žák and Paterson, 2010), it is widely accepted that MMEs were involved in the earliest stages of magma generation, constituting direct evidence of the mechanism that formed the batholiths. Moreover, if MMEs represent melt generated in the mantle as commonly inferred, exploring their relationship with the modified mantle-sourced mafic-intermediate suite, also called sanukitoid, can provide useful insights. Sanukitoid magmas (including appinites and vaugnerites) have been described all around the world as minor mafic-intermediate intrusions associated with post-collisional granite batholiths, displaying a heterogeneous yet consistent geochemistry (e.g., Fowler and Rollinson, 2012; Heilimo et al., 2010). In this paper, we investigate the genetic link between post-collisional MMEs and host granitic rocks by comparing new data from the Los Pedroches batholith (southwestern Iberia) with a large sanukitoid magma database and existing experimental data. The potential significance of MMEs regarding the origin of post-collisional batholiths as fractionated liquids from an intermediate parental magma is established.

The Los Pedroches batholith is an aligned group of intrusions located in the southernmost sector of the Central Iberian Zone in the Iberian Massif. It is hosted mainly by low-grade Carboniferous strata, with only the northern part of the batholith intruding older metapelites of early Cambrian to Neoproterozoic age. The batholith comprises two large groups of intrusions (Fig. 1): (1) an elongated massif referred to as Los Pedroches granodiorite, in which abundant MMEs can be found, and (2) a peraluminous monzogranitic unit, occurring as small discontinuous intrusions that crosscut the granodioritic massif and are much scarcer in MMEs (Carracedo et al., 2009). The current study is focused on the granodiorite suite, which shows evidence of limited assimilation of crustal material during emplacement given the scarcity of host xenoliths, low values in alumina saturation index (ASI), and the isotopic equilibrium between the studied MMEs. A representative number of new enclaves was collected for this study and complemented with existing data from the bibliography (for further details see Table S1 in the Supplemental Material1).

The collected MME samples display variations in grain size, texture, and modal abundance but have a consistent assemblage of plagioclase (Pl), biotite (Bt), amphibole (Amp), quartz (Qz), and K-feldspar (Kfs) with accessory apatite, allanite, zircon, titanite, and oxides, the same assemblage as the host granodiorite. Moreover, MMEs are devoid of chilled margins but show magmatic quenching textures affecting the whole enclaves and no evidence of reaction with the host granodiorite (Castro and Stephens, 1992; Fig. S2). Because mingling between two magmas results in chilled margins, these observations are explained only if MMEs and host rocks are two systems in chemical equilibrium, with MMEs being autoliths (Rodríguez and Castro, 2019). Two types of MMEs are distinguished: MME1 and MME2. Mesocratic and porphyritic textures characterize MME1, with phenocrysts of Pl and Amp clots as much as 5 mm in length, enclosed in a fine-grained groundmass of Pl, Qz, Kfs, Bt, and Amp. By contrast, MME2 is rich in Amp and shows varied textures from equigranular hypidiomorphic to porphyritic, with phenocrysts of Pl and Amp aggregates (clots). Transitional enclaves between the two groups are common. Amphibole (hornblende to locally subsolidus actinolite) clots are the most outstanding feature of both groups, a characteristic that is also common in post-collisional Caledonian-type plutons (Castro and Stephens, 1992). Considering their potential relevance as possible remnants of an ultramafic end member or magmatic cumulates, sampling focused on collecting enclaves containing variable amounts of Amp clots. Collected samples were analyzed for major and trace elements. A selected group was analyzed for isotopic ratios of the Sr and Nd systematics (rock analyses and analytical methods can be found in Table S1 and Supplemental Text S3 [see footnote 1]).

Comparative classification diagrams for the Los Pedroches MMEs, their granodiorite host, and sanukitoid series are represented in Figure 2. World sanukitoid series are represented in a kernel density diagram (Figs. 2A2D) and an average line (Fig. 2E) incorporating a database of 2557 analyses. In most diagrams, a clear overlap between all Los Pedroches samples and the maximum density areas of the sanukitoid series is evident. Compared to the Sierra Nevada (western USA) and Patagonian Andean-type batholiths (dashed red line and gray circles; compilations are available in Gómez-Frutos et al., 2023), significant differences can be observed in the geochemistry of MMEs and host granodiorites. A high-Mg# (Mg# = molar MgO/[MgO + FeOT], where T indicates total iron) signature is evident in the Fe*-silica diagram (Fe* = wt% FeOT/[FeOT + MgO]) (Fig. 2A), a common feature of the sanukitoid series. The ASI (ASI = Al2O3/[Na2O + K2O + {CaO - 3.3 × P2O5}]) diagram shows an overlap in the metaluminous field between all the plotted series (Fig. 2B). Most interestingly, a consistently depleted CaO but enriched K2O and MgO signature in all plotted groups of the the MME-granodiorite association and the sanukitoid series is evidenced in the modified alkali-lime index (MALI = wt% [Na2O + K2O - CaO]) (Fig. 2C) and CaO-MgO diagrams (Fig. 2D), coherent with the post-collisional geochemical signature (Gómez-Frutos et al., 2023). Furthermore, Los Pedroches MMEs also display the characteristic trace element enrichment of sanukitoid magmas, namely in Cr and Ni, and in contents of largeion lithophile elements (LILEs), such as Rb, Sr, and Ba (Fig. 2E). Major and trace element resemblance between MMEs and sanukitoid series supports that the two groups are analogous, implying that they share a common origin and that a modified mantle played an essential role in the origin of MMEs. See Figure S4 (see footnote 1) for additional Harker diagrams.

If derivation from a modified mantle accounts for the origin of the sanukitoid suite (Rapp et al., 2010), conditions during three stages in the evolution of the magmas will be the most influential in constraining magma crystallization conditions: (1) conditions in the mantle during segregation, (2) conditions at boundaries with large rheology contrasts (Moho), and (3) conditions at the final emplacement level. Among them, residence time in the Moho and at emplacement level is expected to be substantially higher than in the source, being the most significant when addressing crystallization conditions. Consequently, pressures of ~1.0 and 0.3 GPa have the most compositional effect in the magmas (de Oliveira et al., 2010). Previous experimental work using Los Pedroches MMEs as a starting material attempted to constrain the geochemical evolution of the sanukitoid series (Gómez-Frutos and Castro, 2022; Gómez-Frutos et al., 2023). Experimental liquids at 1.0 and 0.3 GPa follow a cotectic linear path in the CaO-MgO diagram (Fig. 2D), different from that followed by Andean-type batholiths (Castro, 2021). The implication of the existence of a common cotectic trend is that the composition of both sanukitoids and MMEs is thermodynamically constrained instead of resulting from magma mixing processes. Furthermore, their compositional similarity supports that they share the same origin and are controlled by a common thermodynamic system, following the same liquid line of descent (LLD). In other words, altogether, MMEs and sanukitoid represent the mantle component in post-collisional magma generation.

Cotectic differentiated liquids reproduce most of the geochemical features of granitic (sensu lato) post-collisional magmas (Fig. 2). Nevertheless, they do not fully reproduce the high MgO contents of the series. While Mg# is indeed enriched compared to that of Andeantype rocks, experimental liquids still have slightly lower MgO contents than typical postcollisional rocks. We attribute MgO enrichment to the entrainment of ultramafic material, also likely related to the heterogeneity of the series. However, MgO enrichment is a ubiquitous feature of the magmas, with previous work concluding MMEs obtained their geochemical signature at depth, prior to being incorporated into their host magmas (Clemens et al., 2017, and references therein). Moreover, Sm-Nd isotopes from Los Pedroches batholith indicate that MMEs, sanukitoids, and granites (sensu lato) originated from a common parental magma and that they evolved along a LLD in a mostly closed system where crustal contamination was limited. These considerations exclude contamination during ascent as a potential mechanism to account for the high-MgO signature, suggesting that, instead, it is an intrinsic characteristic of the system.

Experimental evidence highlights how the products from melting of a metasomatized mantle are subjected to the nature of the premetasomatism mantle (Rapp et al., 2010), pointing to a dominantly harzburgitic mantle when addressing the high-Mg signature (Wood and Turner, 2009). More specifically, orthopyroxene (Opx) was identified as the liquidus phase of the sanukitoid system (Gómez-Frutos and Castro, 2022). Orthopyroxene constitutes both an available phase in the source and a potential cumulate in a putative crustal magma chamber. Moreover, entrainment of residual Opx in the magmas can account for the high-Mg signature and does not modify the apparently closed system Sm-Nd signature. These inferences point directly to the Amp clots, assumed to come from Opx alteration by rehydration during crystallization (Beard et al., 2005). In Los Pedroches MMEs, the most mafic type (MME2) has the largest abundance of Amp clots and plots toward the Mg pole in the CaO-MgO diagram, showcasing the largest scatter and accounting for the high-Mg signature (Fig. 2D). Textural variations in MME2 are representative of such self-contamination process, with the most equigranular enclaves resembling cumulates and the most porphyritic resembling parental liquids. Together with Opx, crystal accumulation of accessory phases is also plausible, accounting for the observed deviations in Los Pedroches MMEs compared to the sanukitoid array (e.g., accumulation of titanite produces excess of TiO2). Potential resulting disequilibrium textures are likely masked by the magmas re-equilibrating during emplacement, as given by the low-pressure cotectic predicted in the experiments.

On the relationship between MMEs and host magmas, the most common explanation is that MMEs are the result of hybridization between a mantle-sourced material and melt generated by heating of the lower crust during magma emplacement (e.g., Barbey et al., 2008; Perugini et al., 2003). This hybridization model, however, cannot explain the lack of compositional modifications in the host granitic rocks due to the presence of MMEs, field relationships pointing to independent petrogenetic processes between MMEs and their granitic hosts (Clemens et al., 2017), equilibrium textures between systems that presumably come from two contrasting sources (mantle and crust), or the clear resemblance in major and trace elements between MMEs and mantle magmatism revealed by this study. If instead differentiation is invoked, most of these contradictions are immediately discarded. Inferences from other post-collisional batholiths have already pointed to a common origin for the mafic and felsic magmas (Fowler et al., 2008; Fowler and Henney, 1996; Moyen et al., 2017). Experimental cotectic liquids and the geochemical link between all post-collisional rocks (Figs. 2 and 3) provide consistent evidence that a mantle-derived parental magma can differentiate granodiorites, excluding both the crust in the process and the two-end-member model. This interpretation is supported by horizontal patterns in the 143Nd/144Nd–Hf/Nd diagram, indicating partial melting followed by magmatic differentiation (Fig. 3A), whereas mixing between two different end members would result in vertical patterns (Nielsen and Marschall, 2017). Close similarities for all samples in Nd and Sr initial isotopic ratios (Donaire et al., 1999) support the idea that these magmas are part of the same system, regardless of the whole-rock composition. Minor differences in Sri (where i means initial) are due to local assimilation processes that have been previously described for the Scottish Caledonian (Holden et al., 1987, 1991). Further evidence provided by Sr/Y-Y relations support a two-stage magma fractionation model (Fig. 3B), likely resembling source and lower crust underplating conditions (Fig. 4). Quenching textures in MMEs and MMEs representing fragments of ascent conduits support such conclusions, given that mafic-intermediate magma is the first to rise due to density contrast and lower viscosity and undercooling is expected to happen during ascent (Rodríguez and Castro, 2019), resulting in the observed quenching textures. Quenching also precludes contamination in the enclaves, pointing to pristine trace elements and isotopic relations. Subsequent pulses of more differentiated magma (the granodiorites) are then able to extract fragments from the ascent conduit during ascent, transporting them in suspension and ultimately resulting in the observed autolithic MMEs (Fig. 4). In conclusion, based on the studied MMEs from Los Pedroches batholith, we propose that post-collisional granite petrogenesis can be explained by partial melting and/or differentiation from an intermediate parental magma sourced in the lithospheric mantle, excluding the crust as a potential source.

1Supplemental Material. Rock analyses, sample photographs, sampling and analytical methods, additional Harker diagrams, and extended discussion for geochemical modeling. Please visit https://doi.org/10.1130/GEOL.S.23063099 to access the supplemental material, and contact editing@geosociety.org with any questions.

This work was supported through the Spanish Research Agency (AEI) grant IBERCRUST II/PID2021-126347NB-I00/AEI/10.13039/501100011033/FEDER, UE. We want to thank Rob Strachan for his handling of this manuscript, and Brendan Murphy and Mike Fowler for their constructive feedback.

Gold Open Access: This paper is published under the terms of the CC-BY license.