The magmatism of Atienza (NW Iberian Chain, Spain): age, origin and architecture of the magmatic plumbing system

A sill covering an area of more than 15 km2, several dykes up to 5 m thick and a volcanoclastic deposit crop out with similar petrology in the sector of Atienza (NW Iberian Chain, Spain). They consist of calc-alkaline porphyritic andesites with phenocrysts of plagioclase, amphibole, biotite, garnet and orthopyroxene. Based on U–Pb zircon analysis, an age of 290 ± 3 Ma (Sakmarian-Artinskian in the Cisuralian) has been calculated for this magmatism. The chemistry and geothermobarometry on amphibole crystals revealed crystallisation at different depths between 31 and 16 km, involving several events of magma recharge and fractional crystallisation. Magma ascent led to destabilisation of the amphibole crystals, their replacement by biotite, and the formation of thick microcrystalline coronas. Whole-rock trace element and isotopic compositions support a strong crustal influence in the origin of the magma. Crustal melting was produced by heating generated after lithospheric thinning, delamination, and asthenospheric rise produced after the uplift of the Variscan Orogen and the oroclinal folding of the Iberian Massif.


Introduction
During the Cisuralian, intense calc-alkaline magmatism of intermediate-felsic composition took place throughout the Iberian Massif (Fernández-Suárez et al., 2000;Gutiérrez-Alonso et al., 2011;López-Gómez et al., 2019). The magmatism developed in a context of transtensive tectonics during the end of the Variscan Orogeny and the beginning of a rifting stage, related to the Pangea break-up and the beginning of the opening of the Neo-Tethys ocean (Torsvik & Cocks, 2016;Wilson et al., 2004). In the north-eastern sector of the Iberian Massif, Permian calc-alkaline magmatism is widely represented throughout the Iberian Chain by hundreds of intrusions and volcanic deposits with compositions ranging from rhyolitic to basaltic andesites (Lago et al., 2004(Lago et al., , 2005. Their emplacement was related to directional faults and pull-apart basins and was mainly subvolcanic in the form of dykes and sills. However, block and ash pyroclastic deposits and a plutonic body, the Loscos gabbro, are also exposed (Lago et al., 2004) (Fig. 1). The sector of Atienza ( Fig. 1, 2) hosts the highest volume of calc-alkaline Permian igneous rocks in the Iberian Chain. These rocks have been the subject of several studies about their position within the Permian stratigraphic series Hernando, 1980;Pérez-Mazarío, 1990;Virgili et al., 1973). The only radiometric dating of the Atienza magmatism was carried out by Hernando et al., (1980), with a result of 287 ± 12 Ma (Asselian-Kungurian) based on the 40 K/ 40 Ar whole-rock analysis. Ancochea et al. (1981) related the origin of this magmatism with the collision between continents during the Variscan Orogen. Finally, the geochemical composition of some of these rocks was part of a general study about the Permian calc-alkaline magmatism in the Iberian Chain (Lago et al., 2004).
None of these previous studies focused in detail on the evolution of the magmas from their origin to their emplacement close to the surface, nor the magmatic processes recorded on these rocks, except for the assimilation of xenoliths (Aparicio & García-Cacho, 1984;Lago et al., 2004). The rocks of Atienza contain a complex phenocryst assemblage, including calcic amphibole that shows textures and zoning patterns suggesting a complex evolution in the P-T conditions of crystallisation. Recent studies highlight that amphibole composition is a very useful tool for unravelling magmatic processes such as magma storage, mixing and ascent since this mineral crystallises in a wide range of physicochemical and compositional conditions (Gorini et al., 2018 andreferences therein, Ubide et al., 2014).
In this work, we present a detailed petrological, mineralogical and compositional study of the Permian calc-alkaline magmatism of the Atienza sector. The geothermobarometric study of the amphiboles of these rocks unravels the processes that occurred during the crystallisation of the magmas and makes it possible to reconstruct, for the first time, the architecture of a Permian magmatic system in the Iberian Chain. The new geochemical data and the isotopic compositions from Lago et al. (2004) allowed us to approach the origin of the melts. Finally, we present the first U-Pb zircon age of these rocks, which constrains the age of the magmatism to the Sakmarian-Artinskian.
In the study area, three lithostructural units can be recognised. A first Paleozoic unit, in the southern part, formed by Ordovician and Silurian rocks arranged in a large NNE-SSW oriented syncline (Fig. 2a) and belonging to the extreme SE part of the Ollo de Sapo Domain of the CIZ of the Iberian Massif . This syncline is part of the Hiendelaencina anticlinorium, a Variscan structure resulting from several stages of deformation (Gabaldón et al., 1978). The second unit comprises Permian rocks that crop out in the central part of this area (Fig. 2a) and are disposed discordantly on the Silurian units ( Fig. 2b). They form a syncline (Cañamares Syncline) with an NW-SE orientation except in the easternmost part (Fig. 2). The third unit comprises the Triassic succession, discordantly overlapping any of the previous units (Fig. 2).
The outcrops of igneous rocks locate in the confluence of two important faults of regional scale: the prolongation of the Somolinos fault, with an NW-SE strike, and the Bornova Fault of NNE-SSW strike (Fig. 2a). According to Hernando (1980), these two faults controlled the sedimentation during the Permian and allowed the rise of andesitic magmas, locating the thickest outcrop (outcrop of La Miñosa, Fig. 2) at the confluence between this two faults. The igneous outcrops of Atienza comprise one sill, a pyroclastic deposit and several dykes (Fig. 2).
The sill (Fig. 3a) measures more than 15 km 2 and intruded through the Permian base unconformity (Majarena et al., 2015b). Its thickness increases towards the east (Fig. 2b), exceeding 100 m at the contact between the prolongation of the Somolinos Fault (NW-SE strike) and the Bornova Fault (NNE-SSW strike) (Fig. 2a).
Southwest of Cañamares village, there is a volcanic succession ( Fig. 2a) composed of tens of pyroclastic flows of block and ash (Fig. 3b). It has up to 80 m in thickness, shows irregular and very well-defined bedding surfaces (Fig. 3c), and contains numerous stems and leaves fossils of Callipteris Conferta carried by the pyroclastic flows (Lago et al., 2005, Fig. 3d). The deposit of each pyroclastic flow has a basal section with abundant blocks of cm to dm in size, and an upper part formed by ash-fall deposits with a very fine grain size that measures several centimetres thick or is completely eroded.
To the south of Alpedroches village, the sill is cross-cut by a set of dark grey dykes ( Fig. 2a) with an N-S strike, sub-vertical dip, between 0.5 and 5 m in thickness (Fig. 3e) and net contacts (Fig. 3f). Moreover, two dykes intrude the Silurian series, one of them to the W of La Miñosa village and the other in the north-eastern sector of the studied area ( Fig. 2a).
Polished thin sections were prepared at the Servicio de Preparación de Rocas y Materiales Duros, Universidad de Zaragoza (Spain). Mineral compositions were determined at the Unidad de Fluorescencia y Microsonda Electrónica, Universidad de Oviedo using a Cameca SX-50 electronic microprobe and at the Centro Nacional de Microscopía Electrónica, Universidad Complutense de Madrid (Spain) using a JEOL JXA-8900 M electronic microprobe. Backscattered electron images of the samples were used to guide electron microprobe analysis. Analyses from Cameca SX-50 electronic microprobe were performed using an accelerating voltage of 15 kV and an electron beam current of 15 nA, with a beam diameter of 3 μm. Elemental counting times were 10 s on the peak and 5 s on each of the two background positions. The standards used were orthoclase (K, Si), wollastonite (Ca), albite (Na, Al), pyrophanite (Mn, Ti), hematite (Fe), strontianite (Sr), barite (Ba) and synthetic MgO, Cr 2 O 3 , Rb 2 O and NiO. Analyses from JEOL JXA-8900 M electronic microprobe were performed using an accelerating voltage of 15 kV and an electron beam current of 20 nA, with a beam diameter of 5 μm. Elemental counting times were 10 s on the peak and 5 s on each of the two background positions. The standards used were albite (Na, Si), sillimanite (Al), almandine (Fe, Mn), microcline (K), kaersutite (Mg, Ca, Ti), fluorapatite (P, F, Cl) and pure metal (Ni, Cr). Precision was 0.8 to 5% for major and minor elements. All electron microprobe analyses were corrected for inter-elemental Whole-rock analyses were carried out at the SGIker-Geochronology and Isotopic Geochemistry Facility, Universidad del País Vasco (Spain). The samples were run for major oxides and selected trace elements using an inductively coupled plasma-optical emission spectrometer (ICP-OES Perkin Elmer Optima 8300) and an inductivelycoupled plasma mass spectrometer (ICP-MS X7 Thermo Scientific updated to XSeries 2). The procedures of analyses and standards are provided in Madinabeitia et al. (2008). Detection limits for major oxides were better than 0.0007 wt.% except for Na 2 O (0.0015 wt.%). Detection limits for trace elements were better than 0.03 ppm except for for Ba (0.19 ppm), Co (0.07 ppm), Cr (0.73 ppm), Cu (10.8 ppm), Ni (3.22 ppm), Rb (0.09), Sr (0.12), Pb (0.79 ppm), Zr (0.08 ppm) and Y (0.31 ppm). Measurements of geochemical reference standards were carried out to guarantee the quality and accuracy of the analyses.
The sample for zircon analyses (ATUZ,) was crushed in a steel jaw-crusher and sieved to select the fine fraction (< 250 µm) at the Servicio de Preparación de Rocas y Materiales Duros, Universidad de Zaragoza (Spain). Zircon concentrates were obtained by using the Wilfley table and heavy liquids (methylene iodide) and selected zircon grains were mounted in epoxy resin. Electron back-scattering images were used to reveal the internal structures of the zircon grains and target areas for analysis. Analyses of zircons were conducted at the SGIker-Geochronology and Isotopic Geochemistry Facility, Universidad del País Vasco (Spain). The samples were ablated with a 213 nm Nd: YAGbased laser ablation system (NewWave Research) coupled to a Thermo Fisher iCAP Qc quadrupole ICP-MS instrument with enhanced sensitivity through a dual pumping system. Analytical conditions involved spot diameters of 40 µm associated with repetition rates of 10 Hz and laser fluence at the target of ca. 2.5 J/cm 2 . The ablated material was carried into helium, then mixed with argon and stabilized with a smoothing device (Tunheng & Hirata, 2004) before injection into the plasma source. Single analyses consisted of 25 s of background integration with the laser off followed by 45 s integration with the laser firing. Data reduction was carried out with Iolite v. 3.32 (Paton et al., 2011) and VizualAge (Petrus & Kamber, 2011) using the GJ-1 zircon standard (Jackson et al., 2004) for calibration and Plešovice zircon as secondary standard (Sláma et al., 2008). The analyses of the standards are shown in the Supplementary material 2. Tera-Wasseburg diagrams (Tera & Wasserburg, 1972) and weighted average age were produced using Isoplot/Ex 3.0 (Ludwig, 2003). (2 vol.%) and orthopyroxene (< 1 vol.%) (modal fractions calculated by visual estimation). These phenocrysts show different degrees of alteration, some replaced by chlorite, calcite, or opaque minerals. The volcanic rocks are more altered and have all the amphibole crystals replaced by the secondary phases described above. We have not observed differences in the degree of alteration between the dykes and the sill. All the studied samples also contain xenocrysts of quartz (< 1 vol.%) and some of them also contain xenoliths of black slates and metapelites up to 8 cm in size (< 2 vol.%). The plagioclase phenocrysts reach 3.2 mm in size and have a subhedral habit and oscillatory zoning. Some crystals are strongly altered and completely replaced by secondary minerals. Occasionally, resorption horizons as those described by Ruprecht and Worner (2007), Smith et al. (2009) or Shcherbakov et al. (2011 truncate the layers of the oscillatory zoning. Some crystals show rounded cores surrounded by oscillatory zoning. Plagioclase microcrysts occur in the matrix and the amphibole coronas.

Petrography
The amphibole crystals range between 0.2 and 2.2 mm in size and have subhedral to anhedral habits. Most amphiboles make up isolated phenocrysts and, in some cases, show cores (C-type) overgrown by mantles (M-type). Others appear grouped in glomeroporphyritic aggregates and show a zoning pattern (G-type). Most amphiboles have coronas formed by microcrysts of clinopyroxene, amphibole, plagioclase, biotite, ilmenite, magnetite, titanomagnetite and apatite. The rest of the amphiboles show resorbed textures.
The igneous biotite occurs as euhedral and homogeneous crystals up to 0.8 mm in size, as inclusions in amphibole, in pseudomorphs of amphibole and the coronas of the amphibole. On the other hand, some biotite crystals have anhedral habits and numerous inclusions, show darker colours in parallel polars and resorbed borders, and are similar to the biotite of the xenoliths.
Garnet can reach 1 cm. It shows cracked and mottled cores with inclusions of biotite, ilmenite, rutile and apatite. This textural feature is typical of the garnets from the metapelite xenoliths, and these cores have been interpreted as xenocrystic (Majarena et al., 2015a and2021). The xenocrystic cores are surrounded by homogeneous mantles, without inclusions and with scarce cracks, that have been interpreted as of igneous origin (Majarena et al., 2015a and2021).
The orthopyroxene and zircon are accessory minerals (< 1%). The orthopyroxene phenocrysts are between 0.2 and 1.2 mm long and have a subhedral habit. The zircons crystals are 150-400 μm long, euhedral to subhedral, with straight edges and mainly prismatic. Finally, quartz forms rounded xenocrysts up to 0.8 mm in size.
The subvolcanic samples (sill and dykes) show an equigranular microcrystalline matrix ( Fig. 4a) with quartz, plagioclase, minor apatite and, in a lower proportion, secondary opaque minerals.
The pyroclastic flows contain up to decimetre-sized juvenile fragments and isolated crystals (plagioclase, amphibole, biotite, quartz and garnet) immersed in an altered vitreous matrix with low contents in vesicles (Fig. 4b). The juvenile fragments are porphyritic andesites containing the crystals listed above in an equigranular microcrystalline matrix similar to that of the subvolcanic samples. The rocks corresponding to the fall deposits are made up of very fine-grained crystals and shards (Fig. 4c).

Whole-rock geochemistry
The analytical results of the whole-rock (major oxides, trace elements, and Sr-Nd isotopic composition) of the representative samples of this magmatism are listed in Table 1. The loss on ignition (LOI) ranges between 1.17 wt.% and 5.72 wt.%. The rocks with LOI > 4 wt.% (ATU-3, ATU-8, ATU-10 and AT-12; Table 1) show the main variability in the whole-rock composition and, for this reason, are not taken into account for the geochemical study.
The samples are classified as andesite, dacite, trachyandesite and trachyte on the TAS diagram ( Fig. 5b) and as andesites on the Nb/Y-SiO 2 diagram (Fig. 5c). The rocks share a similar composition that is consistent with Permian calc-alkaline magmatism of the Iberian Chain (Lago et al., 2004 and references therein, Fig. 5). Table 1 Major (wt.%), trace element (ppm) and isotopic composition of the studied rocks     The chondrite-normalised rare earth element (REE) patterns of these rocks are very similar (Fig. 6a) (Peccerillo & Taylor, 1976). b Alkalinity vs SiO 2 (TAS) diagram (Le Bas et al., 1986). c Zr/TiO 2 vs SiO 2 diagram (Winchester & Floyd, 1977). The grey points correspond to samples of the Permian magmatism in the Iberian Chain, taken from Lago et al. (2004) and references therein, Majarena et al. (2017), Navidad (1983), and Torres et al. (1991).   Orejana et al., (2009 and2020), and Villaseca et al., (1998Villaseca et al., ( , 1999Villaseca et al., ( and 2009 Fig. 6a]. Primitive mantle-normalised multi-element patterns (Fig. 6b) are also very similar and characterised by enrichment in large ion lithophile elements (LILE) (e.g., Rb, Ba) and light rare earth elements (LREE) and depletion in some of the high field strength elements (HFSE) (e.g., Nb, Ta). They also show a depletion in Ti and P and a small positive anomaly in Sr. In addition, the rocks exhibit a slight variability in Zr, Hf and the HREE (Fig. 6b).
The isotopic ratios analysed by Lago et al. (2004) in four samples of the sill (AT-6, AT-8, AT-12 and AT-13, Table 1) exhibit initial 87 Sr/ 86 Sr ratios of 0.70818 and 0.70875 and negative ɛNd(t) values of − 5.9 to − 5.3 ( Fig. 7; Table 1), yielding a homogeneous Nd model-age of 1,6 Ga. The isotopic composition (Sr-Nd) of these rocks is similar to those of some Variscan and late-Variscan granites and granodiorites of the SCS (Moreno et al. 1995;Villaseca et al., 1999) and the Pyrenees (

Mineral chemistry
Mineral analyses were performed in samples of both subvolcanic and volcanic rocks.

Plagioclase
The composition of the plagioclase phenocrysts ranges from labradorite (An 58 , Ab 41 , Or 1 ) to andesine (An 36 , Ab 62 , Or 3 ; Table 2, Supplementary material 1). The plagioclase phenocrysts have oscillatory zoning of high amplitude (anorthite content variations up to 16%) and low frequency (HALF; Viccaro et al., 2010). They usually have resorption horizons and Na-rich rounded cores that, in both cases, are overgrown by more anorthite-rich composition. Aside from these overgrowths and the oscillatory zoning, there are no apparent core-to-rim compositional variations. Plagioclase microcrysts of the matrix range from andesine (An 45 , Ab 53 , Or 2 ) to oligoclase (An 20 , Ab 72 , Or 8 ) and those of the coronas range from labradorite (An 56 , Ab 41 , Or 3 ) to oligoclase (An 27 , Ab 70 , Or 3 ).

Amphibole
The amphiboles are pargasite, magnesium-hastingsite and tschermakite, according to Leake et al. (1997) (Supplementary material 1). The phenocrysts and the glomeroporphyritic crystals make up a heterogeneous crystal population with different compositions and zoning patterns. Therefore, three types of amphibole compositions have been defined based on their petrography and geochemistry: G, C, and M. In addition, amphibole microcrysts occur in the coronas.
M-type compositions constitute the mantle of the phenocrysts with C-type cores (Fig. 8b, c), but sometimes they also form entire phenocrysts. They are the most common population and have been identified in all analysed samples. They are more evolved than C-type, with mg* values between 47.7 and 71.3 and CaO between 9.2 and 12.4 wt.% (Fig. 9, Tables 2 and 3). The M-type is subdivided into two compositional zones, M 1 and M 2 . M 1 zone surrounds C-type compositions (Fig. 8c) and sometimes forms individual crystals. The mg* values of the M 1 zone range between 56.1 and 71.3, and their CaO contents are between 10.4 and 11.8 wt.% (Fig. 9, Tables 2 and 3). M 2 zone surrounds C N (Fig. 8b) and M 1 (Fig. 8c), and sometimes also forms entire crystals. This zone has the lowest values of mg* = 51.7-60.7 and CaO = 9.8-11.2 wt.%.
Finally, the microcrysts of the coronas have a wide compositional variability, with mg* values ranging between 53.4 and 67.8 and CaO between 9.5 and 12.4 wt.% (Fig. 9).

Table 2
Major element composition of representative mineral phases (see supplementary material 1 for structural formulae)

Pyroxene
The orthopyroxene has a very homogeneous composition (

Geothermobarometry
The crystallisation pressures and temperatures of the amphiboles have been determined by applying the equations of Ridolfi (2021), that have an error of ± 22 °C and 0.2-2 kbar, and are based on the major elements of the amphibole and being independent of the whole-rock composition. The microcrysts of the coronas have not been considered for the P-T calculations as they are likely related to destabilisation processes.
We have obtained the highest values of pressure and temperature of crystallisation for the G-type amphiboles (Fig. 10a, b). They have very similar pressure (ca. 7.8 ± 1.8 kbar) and temperature (ca. 1005 ± 23 °C; Table 3), with a small increase in the temperature values from G 1 to G 2 . The pressure and temperature obtained for C N cores are 6.1 ± 1.8 kbar and 957 ± 26 °C ( Fig. 10; Table 3), while the pressure and temperature obtained for C R cores are lower (C R1 : 5.3 ± 0.6 kbar and 922 ± 22 °C; C R2 : 5.8 ± 1.4 kbar and 949 ± 24 °C). The pressure obtained for the M 1 zone is 4.9 ± 0.9 kbar, and its temperature is 922 ± 22 °C (Table 3). Finally, we have obtained the lowest values of P (3.8 ± 0.8 kbar) and T (881 ± 22 °C) for the M 2 zone ( Fig. 10; Table 3).
The igneous garnets are Ca-rich (CaO > 4 wt.%) and Mnpoor (Mn < 3 wt.%). This type of almandines could crystallise from hydrous andesitic magmas at more than 7 kbar and 900-950 °C and are unstable at shallow depths (Harangi et al., 2001 and references therein). Moreover, the experimental results of Green (1992) suggest that the mineral assemblage of plagioclase + amphibole + garnet in dacitic melts is stable at pressures above 8 kbar and temperatures above 800 °C.

LA-ICP-MS U/Pb geochronology
A total of 27 zircons from the sample ATUZ (Table 4) were analysed to determine the age of magma crystallisation. Only 16 zircons (11 analyses were discarded) with a degree of concordance of 95-105% were used for data reduction and 206 Pb/ 238 U age calculation (Table 4). To date the last crystallisation event, the analyses were performed in the outermost area of each crystal (Fig. 11a). They show oscillatory concentric zoning and, in some cases, ovoidal cores truncated by the oscillatory zoning (Fig. 11a). A Concordia age of 290 ± 3 Ma from a group of 11 concordant analyses was calculated (Fig. 11b, Table 4). A weighted average age of 288 ± 3 Ma was also obtained. The Th/U ratio of concordant analysis ranges from 0.12 to 0.30 (Table 4).

Architecture of the magmatic plumbing system
The P-T crystallisation conditions, the observed compositional variations and the equilibrium/disequilibrium relationships between the minerals and the whole-rock, allow us to unravel a complex magmatic history for the Permian magmatism at Atienza.
The analysed amphiboles constitute a trend of decreasing Al 2 O 3 , CaO and Na 2 O with decreasing mg* from the glomeroporphyritic crystals (G) to the outermost parts of the phenocrysts (M 2 ) (Fig. 9); in this variation trend, SiO 2 concentrations are fairly constant. On the other hand, the microcrysts of the coronas plot scattered and do not follow the same trend (Fig. 9).  (Table 4). b Tera-Wasseburg plot (Tera & Wasserburg, 1972) of the analysed zircon crystals Fig. 12 a Equilibrium between the amphiboles and the melt. Solid and dashed lines (after Rhodes et al., 1979) represent the range of equilibrium between mineral and melt using an iron-magnesium distribution coefficient of 0.28 ± 0.11 (Putirka, 2016). The whole rock composition of the sample containing the mineral has been used as a proxy of the melt. b Relation between the estimated pressure of crystallisation of the amphibole population on the basis of the equations of Ridolfi et al. (2021), and the SiO 2 content (wt.%) of the melt in equilibrium with each amphibole composition (Putirka, 2016) The mineral-melt equilibrium between the amphibole and the whole-rock composition was assessed by applying the Rhodes diagram (Fig. 12a). They also show a continuous trend from compositions slightly more mafic than the whole-rock (G-Type) to more evolved than the whole-rock (M 2 ). The C N amphiboles are in equilibrium with the wholerock, whereas C R1 amphiboles are more evolved and C R2 amphiboles are close to the equilibrium (Fig. 12a). The M 1 amphiboles show a larger range of mg*, from values near to the equilibrium to values clearly in disequilibrium.
Moreover, there is a direct relationship between the estimated pressures of crystallisation and the predicted SiO 2 content of the melt in equilibrium with the amphiboles (Fig. 12b). As the estimated pressure decreases, the melt in equilibrium with the amphiboles becomes more felsic (Fig. 12b). The compositional trend observed in the whole of the amphibole dataset (from G-Type to M 2 zones, Fig. 9) and the progressive descend in the SiO 2 of the melt in equilibrium with them (Fig. 12b) suggest that all of them belong to the same magmatic system and can be related by fractional crystallisation processes.
The glomeroporphyritic amphiboles (G-type), with a more primitive composition than the whole-rock (Fig. 12a), could correspond to primitive antecrysts. Regarding their internal zoning, the G 2 mantles have higher mg* and CaO wt.% than the G 1 cores. In addition, the limit between the G 1 and G 2 zones is very clear, and the change of composition occurs abruptly. This compositional change from G 1 to G 2 suggests that the G 2 amphiboles grew after a more mafic and hotter magma recharge, which took place in the reservoir where G 1 type amphiboles were crystallising (Table 3, Fig. 13). Finally, the glomeroporphyritic amphiboles (G-type) evolved by fractional crystallisation from G 2 to G 3 forming normal zoning in the crystals. The P-T conditions obtained for the different zones of the G-Type amphiboles are very similar and show that they crystallised around 7.8 kbar and 1000ºC (Table 3, Fig. 10b). These P-T conditions are consistent with those of the igneous garnets (P > 7 kbar and T = 900-950 ºC; Harangi et al., 2001 and references therein). Moreover, the stability of the mineral assemblage of plagioclase + amphibole + garnet agrees with these pressures (Green, 1992). Considering the density of the crust (between 2.3 g/cm 3 at the surface and 2.9 g/cm 3 at a depth of 31 km) and the crustal structure in the CIZ defined by Villaseca et al. (1999), the estimated pressure of 7.8 kbar corresponds to a depth of 31 km (Table 3, Fig. 13).
Regarding the cores of the amphiboles, C N crystallised at 6.1 kbar (24 km) and C R1 at 5.3 kbar (21 km). The C R2 have higher values of mg* than C R1 (Fig. 9). This suggests, for G 2 zones, that C R2 grew after a more mafic and hotter magma recharge in the reservoir where C R1 were crystallising (Fig. 13). The observed pressure increase in the geothermobarometric results from C R1 to C R2 is unrealistic and only responds to the compositional change caused by the recharge. G-type amphiboles, C N cores and C R cores never occur together, suggesting that their crystallisation occurred in independent magma reservoirs (Fig. 13).
The magmatic recharges of a more basic melt inferred for the G 2 and C R2 amphiboles are consistent with the observed overgrown in the plagioclase crystals of the Na-rich rounded cores and the resorption horizons by a more anorthiterich composition (Cashman & Blundy, 2013;Ruprecht & Wörner, 2007;Shcherbakov et al., 2011;Smith et al., 2009).
The M 1 amphiboles crystallised at 4.9 kbar (20 km). They surround both the C N and C R cores, indicating that the magmatic chamber where M 1 crystallised incorporated cores (C N and C R ) formed at higher depths (Fig. 13). The crystallisation pressure of M 2 amphiboles was 3.8 kbar (16 km). They grew surrounding M 1 , C N or C R and evolved from M 1 by fractional crystallisation (Figs. 9, 12).
All amphibole crystals show resorbed rims or have microcrystalline coronas. The coronas reach a thickness of up to 100 µm and contain microcrystals of amphiboles (R-type), biotite, clinopyroxene, plagioclase, apatite, ilmenite, Fig. 13 Conceptual approach to the magmatic system and processes involved in the Atienza magmatism. The hexagons represent the different amphibole crystals (same legend as Figs. 5, 6 and 7). The depth values have been calculated according to the density and crustal structure defined by Villaseca et al. (1999) 1 3 magnetite and titanomagnetite. The origin of this coronas has been related to decompression processes during a rapid rise of the magma (Buckley et al., 2006) that is coherent with the occurrence of Ca-bearing almandine, unstable at low pressures (Harangi et al., 2001), in these rocks. On the other hand, the biotite crystals of the coronas have a very similar composition to those that grow as amphibole pseudomorphs. This fact suggests a general destabilisation process in the amphibole that caused the formation of the coronas and the replacement of some amphibole crystals with biotite.
The range of crystallisation pressures of the amphiboles (3.8-7.8 kbar) agrees with the metamorphic conditions inferred for the metamorphic xenoliths carried by these magmas (2-8 kbar).

Origin of the magmatism
These rocks have nearly identical compositions and do not show any correlation of SiO 2 with other major oxides (Supplementary material 3). Nevertheless, fractional crystallisation is recorded by the evolution of the amphibole composition (Fig. 9,). The rocks of Atienza are similar to those of the Permian calc-alkaline magmatism of the Iberian Chain but have higher LREE-and lower HREE-enrichment with respect to the chondrite composition (Fig. 6a). As amphibole has D mineral/melt > 1 for HREE in andesites and dacites, the fractionation of this phase could be responsible for the differences in the REE-normalised patterns respect to the rest of the Permian calc-alkaline of the Iberian Chain.
Moreover, the moderate variability shown by Zr and Hf might be related to the zircon fractionation (Fig. 6b). On the other hand, the small positive anomaly of Sr in some rocks (Fig. 6b) could indicate slight plagioclase accumulation.
The accumulation of plagioclase and amphibole also could explain respectively the high content in LREE and HREE of some samples. In this sense, a correlation between the concentration in REE and the content in plagioclase and amphibole of the samples is observed. The sample with the highest content in REE (AT-11) have a slightly higher proportion of phenocrystals (ca. 45 vol.%) respect to the matrix (55 vol.%). The sample ATU-A9, with the lowest content in phenocrystals (ca. 35 vol.%) have the lowest concentration in REE. Finally, the sample ATU-A13, with a slightly higher concentration in plagioclase (ca. 60 vol.%) and a lower proportion of amphibole (ca. 25 vol.%), has in relation with the other samples, higher content in LREE and Sr, and lower content in HREE.
The negative anomalies in Nb, Ta, P and Ti and the enrichment in LILE (Fig. 6b) shown by these rocks suggest the involvement of crustal elements in the origin of this magmatism (Lago et al., 2004;Orejana et al., 2020). Their multi-elemental normalised patterns (REE and spider diagram, Fig. 6) are also very similar to that of the upper crust (Rudnick and Gao, 2003) in spite of their lower HREE content (that could be a result of amphibole and mainly garnet fractionation) and Zr and HF (that could be a result of zircon fractionation), and a higher content in Sr and P (that could be related, respectively, with the accumulation of plagioclase and apatite). Moreover, the occurrence of almandine garnet of igneous origin in these peraluminous rocks and metapelitic xenoliths is typical of melts generated by anataxis of metapelitic rocks (Harangi et al., 2001 and references therein). Finally, the Sr-Nd isotopic composition is markedly enriched (ƐNd = − 5.3 to − 5.9; 87 Sr/ 86 Sr (290 Ma) = 0.7082 to 0.7088; Table 1) and similar to some granites and granodiorites of the SCS and the Pyrenees, of crustal origin (Fig. 7).  The grey points correspond to samples of the Permian magmatism in the Iberian Chain, taken from Lago et al. (2004) and the references therein, Majarena et al. (2017), Navidad (1983), and Torres et al. (1991) All these elemental and isotopic criteria suggest a strong crustal imprint for the origin of this magmatism. The high concentration of partially reabsorbed enclaves of metapelite is consistent with this origin. The Sr-Nd isotopic ratios are very similar to the rest of the Permian calk-alkaline rocks of the Iberian Chain (Fig. 7), supporting a similar petrogenesis. These rocks are close to the Gb2 magmatism of the SCS characterised by calc-alkaline microdioritic dyke swarms accompanied by felsic granitic porphyries of 292 Ma (Orejana et al., 2020;Villaseca et al., 2022). The microdiorite dyke swarms have been interpreted as the melting products of an enriched lithospheric mantle mixed locally with felsic magmas (Orejana et al., 2020). Assuming a mixing model between mantle-derived magmas and the crust (i.e. assimilation, anataxis) for the origin of the Permian magmatism as suggested for the rest of the Iberian Chain (Lago et al., 2004) and the coeval SCS magmatism (Orejana et al., 2020;Villaseca et al., 2022), in Atienza, the mantle signature is negligible, and the source of the melts could be close to the crust (Fig. 7).
The composition of these rocks plots in the field of collisional rocks (Fig. 14). This tectonic setting agrees with the palaeogeography of the Atienza area being part of the Gondwana continental plate, which collided with the Armorica, Avalonia and Laurussia plates during the end of the Variscan orogeny and led to the formation of the Pangea supercontinent (Pérez-Estaún et al., 2004;Wilson et al., 2004). This collision favoured the cortical thickening during the lifting of the Variscan orogen and the orocline folding of the Iberian Massif (Gutiérrez-Alonso et al., 2011). During the Cisuralian, a heat increase related to the thinning, delamination and subsidence of the lithosphere, and the asthenospheric rise in the Iberian Massif allowed melting in the crust and the lithospheric mantle, and the development of the Permian magmatism.
The depleted mantle (DM) model ages obtained with the isotopic compositions from Lago et al. (2004) range from 1.56 to 1.63 Ga (Table 1). These model ages are very similar to those of the granitoids and gneisses from the CIZ (Bea et al., 2010;Castro et al., 1999;Montero et al., 2017;Villaseca et al., 1998Villaseca et al., , 2009). These ages can be interpreted as an estimation of the time of crust formation (Rollinson & Pease, 2021). Nevertheless, no significant continental crust formation events in the SW of Europe and the N of Africa are known at this time, and the obtained ages could be due to the mixing of components with very different crustal residence times (Bea et al., 2011).

Age of the magmatism
The obtained age for the Atienza magmatism in this study is Sakmarian-Artinskian (290 ± 3 Ma, Fig. 11) and consistent with the previously obtained age by Hernando et al., (1980) (287 ± 12 Ma, Asselian-Kungurian). As the age estimated in this work is within the error margin of Hernando et al., (1980) for the same sill but is more precise (± 4 My), we consider the age of 290 ± 3 Ma as the best estimation for the sill crystallisation age. In the Castilian Branch of the Iberian Chain, only two other outcrops of calc-alkaline magmatism have been dated: the pyroclastic flows of Orea and the Noguera de Albarracín sill. Both are located in the Albarracín massif and yield ages of 294 ± 3 Ma and 293 ± 2 Ma (Asselian-Sakmarian), respectively (Majarena et al., 2018). All these ages point to a short time range (290-294 Ma, ca. Sakmarian) for the development of this magmatism in the Castilian Branch.
Moreover, the Permian calc-alkaline magmatism in the Iberian Chain can be considered as a short-live event (283-294 Ma) immediately after the formation of the Variscan orogen and the orocline folding of the Iberian massif (Gutiérrez-Alonso et al., 2011), that agrees with that observed in the rest of the WALZ (285-295 Ma;Fernández-Suárez et al., 2000). The calc-alkaline magmatism in the Iberian Cordillera is also contemporary with the calcalkaline dykes of the SCS (292 Ma, Orejana et al., 2020) and the calc-alkaline volcanism of the Pyrenees (291-292 Ma, Pereira et al., 2014).

Conclusions
The magmatism of Atienza is characterised by calc-alkaline intermediate rocks with homogeneous compositions between andesite, dacite, trachyandesite and trachyte. They are porphyritic and contain phenocrysts of plagioclase, amphibole, biotite, garnet, orthopyroxene, xenocrysts of quartz and metapelitic xenoliths.
We report for the first time a zircon U-Pb age of this magmatism (290 ± 3 Ma, Sakmarian-Artinskian). This age agrees with and better constrains the previous K-Ar in the whole-rock age of 287 ± 12 Ma (Asselian-Kungurian, Hernando et al., 1980) and coincides with the Cisuralian age of 1 3 the calc-alkaline magmatism in the Iberian Chain (Majarena et al., 2018 and references therein). This magmatism is coeval with other calc-alkaline magmatisms in the Pyrenees (Pereira et al., 2014), the Spanish Central System (Orejana et al., 2020) and the Iberian Massif (Gutiérrez-Alonso et al., 2011).
The negative anomalies in Nb, Ta, P and Ti, the enrichment in LILE, and the enriched Sr-Nd isotopic composition (ƐNd = − 5.3 to − 5.9; 87 Sr/ 86 Sr (290 Ma) = 0.7082-0.708 8) suggest a strong crustal imprint for the origin of this magmatism. Moreover, the whole rock compositions are close to the upper crust. These rocks are also very similar to the rest of the Permian calc-alkaline rocks of the Iberian Chain, suggesting a similar origin. In a mixing model between mantle-derived magmas and the crust, previously proposed for the rest of the Iberian Chain and the coeval magmatism of the Spanish Central System, the source of the melts of the Atienza magmatism is close to the crust. At the beginning of the Permian, in the Iberian Massif, the lithospheric thinning and asthenospheric rise favoured the melting of the crust and the lithospheric mantle (Gutiérrez-Alonso et al., 2011).
The petrography and mineral chemistry and the geothermobarometric estimates led us to unravel the possible architecture of a magmatic system in the Iberian Chain for the first time. The deepest identified crystals are glomeroporphyritic amphiboles (G 1 ) crystallised around 31 km. This depth coincides with the stability conditions for the igneous garnet observed surrounding xenocrystic cores. At this level, recharge with more primitive magma induced reverse zoning (G 2 ) followed by fractional crystallisation (G 3 ). Most amphiboles have cores formed at 24 km (C N ), whereas the other cores are related to a magmatic mafic recharge and crystallised at 21 km (C R ). At a similar depth, the growth of the amphibole continued by forming mantles (M 1 ) that surrounded the cores and evolved by fractional crystallisation (M 2 ) up to 16 km depth. Finally, the rapid ascent of the magma produced the amphibole destabilization favouring the formation of microcrystalline coronas of amphibole, clinopyroxene, plagioclase, biotite, ilmenite, titanomagnetite and apatite, and the replacement of the amphibole by biotite. The magma and its crystal cargo were emplaced close to the surface as a subvolcanic sill and later dykes. Magma reaching the surface produced pyroclastic flows of blocks and ash.