The Late Cretaceous igneous rocks of Romania (Apuseni Mountains and Banat): the possible role of amphibole versus plagioclase deep fractionation in two different crustal terranes

Abstract

We provide new whole-rock major and trace elements as well as ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd isotopic data of a suite of samples collected in the Late Cretaceous volcanic and plutonic bodies of the Apuseni Mts. (Romania) that belong to the Banatitic Magmatic and Metallogenic Belt, also called the Apuseni–Banat–Timok–Srednogorie belt. The samples define a medium- to high-K calc-alkaline differentiation trend that can be predicted by a three-step fractional crystallization process which probably took place in upper crustal magma chambers. Published experimental data indicate that the parent magma (Mg# = 0.47) of the Apuseni Mts. trend could have been produced by the lower crustal differentiation of a primary (in equilibrium with a mantle source) magma. The Late Cretaceous magmatic rocks of the Apuseni Mts. and Banat display overlapping major and trace element trends except that Sr is slightly lower and Ga is higher in the Apuseni Mts. parent magma. This difference can be accounted for by fractionating plagioclase-bearing (Apuseni Mts.) or amphibole-bearing (Banat) cumulates during the lower crustal differentiation of the primary magma to the composition of the parent magma of both trends. This, together with results obtained on the Late Cretaceous igneous rocks from the Timok area in Eastern Serbia, further suggests variation of the water content of the primary magma along and across the belt. The Apuseni Mts. versus the Banat samples display different isotopic compositions that likely resulted from the assimilation of two distinct crustal contaminants, in agreement with their emplacement in two separate mega-units of Alpine Europe.

Introduction

The Apuseni Mts. are part of the 1500-km-long Late Cretaceous Banatitic or Apuseni–Banat–Timok–Srednogorie Magmatic and Metallogenic belt that stretches across south-eastern Europe from Romania, through Serbia and Bulgaria (Berza et al. 1998; Popov et al. 2002). Late Cretaceous magmatic rocks are also known from drillings in the Tatra Mts. of Slovakia (Kohut et al. 2013) and in the north-eastern part of the Pannonian Basin (NE Hungary–NW Romania) (Berza and Ilinca 2014). In outcrop, a NE–SW orientation in the Apuseni Mts. and the western end of South Carpathians (Banat) turns to N–S in Eastern Serbia (Timok and Ridanj-Krepolijn) and finally to E–W in Bulgaria (Srednogorie) (Fig. 1). This present-day orientation is due to a 80° clockwise rotation of the Tisza and Dacia units during the Cenozoic (Panaiotu 1998), turning the Tatra–Apuseni–Banat part from an original E–W orientation to a N–S one. The Banatitic belt contains abundant ore deposits ranging from Cu–Au–Ag epithermal deposits to Fe–Pb–Zn–Mo skarn and Cu–Mo–Au porphyry (Berza and Ilinca 2014). Precise Re–Os geochronological data acquired on molybdenite from these ore deposits (Zimmerman et al. 2008; Kohut et al. 2013) and U–Pb LA-ICPMS ages obtained on the magmatic zircon (Nicolescu et al. 1999; Von Quadt et al. 2005; Georgiev et al. 2012; Kohut et al. 2013; Kolb et al. 2013) indicate the following bracketing ages for intrusion and/or molybdenite crystallization: Tatra (U–Pb: 81 Ma—Re–Os: 81 Ma), Apuseni (Re–Os: 79–80 Ma), Banat (Re–Os: 72–83 Ma—U–Pb: 76–79 Ma), Ridanj-Krepolijn (U–Pb: 71–76 Ma), Timok (Re–Os: 81–88 Ma—U–Pb: 79–89 Ma) and Srednogorie (Re–Os: 87–92 Ma—U–Pb: 78–92 Ma).

Fig. 1

Schematic geological map of Roumania displaying the location of the studied intrusions

The geodynamic setting of the Late Cretaceous magmatic rocks, named “banatites” (Von Cotta 1864), was mostly considered as subduction-related because of their calc-alkaline composition, their enrichment in LILE and depletion in Nb–Ta and the presence of a major oceanic remnant in the Carpathian–Balkan orogen, namely the Eastern Vardar Ocean (a branch of the Tethys Ocean) that is supposed to have subducted northward under the European margin during the Jurassic to Lower Cretaceous and to which the ophiolites of the Mureş zone (or Transylvanides) are linked (Schmid et al. 2008) (Figs. 1, 2). However, their precise geodynamic setting is still debated (Berza and Ilinca 2014). Based on their occurrence in Upper Cretaceous Gosau-type sedimentary basins that post-date the Mid-Cretaceous nappe empilement, several authors consider that they were emplaced in an extensional regime caused by orogenic collapse (Giuşcǎ et al. 1969; Antonijević et al. 1974; Popov 1981; Berza et al. 1998; Popov et al. 2002; Georgiev et al. 2012) or slab rollback (e.g. Von Quadt et al. 2005; Zimmerman et al. 2008; Schuller et al. 2009; Kolb et al. 2013). Neubauer (2002) interpreted this magmatism either as Andean-type or post-collisional and resulting from slab break-off or continuous subduction. Quantification of the magmatic processes that produced these igneous rocks provides independent constraints that help deciphering their origin. Detailed petrological informations are now available for the Banat region of southern Romania (Dupont et al. 2002) as well as for the Timok and Ridanj-Krepolijn regions of Eastern Serbia (Kolb et al. 2013). In this study, we analyse the petrogenesis of Late Cretaceous magmatic rocks outcropping in the Apuseni Mts., using new major and trace element, and isotopic data (Sr, Nd) on a series of samples collected in this region. We discuss their differentiation processes, their possible mantle-derived parent magmas and plausible contaminants. These data, together with the recent recognition of adakite-like magmas in the Timok and Ridanj-Krepolijn regions of NE Serbia (Kolb et al. 2013), enable us to propose that the H₂O content of the mantle-derived magmas from which the Late Cretaceous magmas differentiated varied along the belt.

Fig. 2

Schematic tectonic map of the Apuseni Mts. (after Kounov and Schmid 2013) showing the location of the Late Cretaceous magmatic bodies that are sampled in this study. Vlad Vlădeasa, Bud Budureasa, Piet Pietroasa, B. Bih. Băiţa Bihor, Lunc Luncsoara, Bais Băişoara, Gil Gilau

Geological setting

The Apuseni Mts. are part of the Carpathians of Romania and represent a huge basement outcrop between the Neogene Transylvanian and Pannonian Basins (Fig. 2). The southern Apuseni Mts. essentially comprise ophiolites of Jurassic age (the Metaliferi Mts.) that represent a branch of Neotethys, known as the Mureş zone (Savu 1996) or Transylvanian nappes (Rǎdulescu and Sǎndulescu 1973). The northern Apuseni Mts. are made of a sequence of several nappe systems that was formed during the continental collision (e.g. Schmid et al. 2008; Kounov and Schmid 2013). Each nappe system comprises crystalline basement rocks that deformed and metamorphosed during the Variscan orogeny. These basement rocks were intruded by Variscan granitoids and later covered by a sequence of sedimentary rocks of Permian to Late Cretaceous age (e.g. Pană et al. 2002; Balintoni et al. 2009). The boundary between the Tisza and Dacia Mega-Units, the South Transylvanian fault, passes through the Apuseni Mts. The two southern nappe sequences are attributed to the Dacia Mega-Unit, whereas the northern ones are interpreted as belonging to the Tisza Mega-Unit. However, the exact limit between the two Mega-Units is still debated (Csontos and Vörös 2004; Schmid et al. 2008). During the Late Cretaceous post-orogenic collapse of the orogen, sedimentary Gosau-type basins were formed because of extensional processes (Willingshofer et al. 1999; Schuller 2004; Schuller et al. 2009). The Late Cretaceous magmatic rocks are observed as volcanics in the Gosau-type basins and as shallow intrusions and volcanic bodies that are spread all over the Apuseni Mts. mostly as small occurrences (e.g. Budureasa, Pietroasa, Băişoara, Băiţa Bihor) except the Vlădeasa volcano-plutonic complex, situated in the northern part, that covers an area of more than 600 km² (Istrate 1978; Stefan 1980; Stefan et al. 1982, 1992) (Fig. 2).

Detailed mapping of the Late Cretaceous magmatic rocks outcropping in the Apuseni Mts. was performed by Istrate (1978), Stefan (1980), Stefan et al. (1982, 1992) and Ionescu (1997). These authors recognized volcanic and plutonic facies that range in composition from andesites/quartz diorites to dacites and rhyolites/granites. Istrate (1978), Stefan (1980) and Stefan et al. (1982) have also described dark fine-grained dykes, referred to as lamprophyres, that are up to 1 m of thickness and intrude the Senonian deposits as well as the volcanic and eruptive Late Cretaceous igneous rocks. Field evidence indicates that Vlǎdeasa volcanic rocks cover sedimentary rocks of Upper Coniacian–Lower Maastrichtian age and are themselves included in conglomerates of Palaeocene–Ypresian age (Stefan et al. 1992). K–Ar ages of 61 ± 3 and 61.5 ± 5 Ma were obtained on the Vlădeasa rhyolite by Bleahu et al. (1984). These ages are similar to those obtained on pre-Mesozoic Bihor crystalline schists, pointing to the presence of a significant thermal aureole of Banatitic plutons, also expressed by aeromagnetic and gravimetric anomalies (Andrei et al. 1989). Important ore bodies are associated with the granodiorites and granites as, for example, the Mo–Cu skarn of Băița Bihor dated between 80.63 ± 0.3 and 78.69 ± 0.4 Ma (Re/Os on molybdenite) (Zimmerman et al. 2008). This is currently the most accurate isotopic age obtained on Late Cretaceous igneous rocks from the Apuseni Mts., but ongoing research from ETH Zürich will soon present U–Pb zircon ages.

Sampling and methods

Because of the abundant vegetation in the region of the Apuseni Mts., the number and extension of the outcrops are rather limited. This hampers detailed observations of the relationships between different facies within one intrusion and also of the relationships of one particular intrusion with its surroundings rocks. Sixty-one samples were collected in a series of intrusions [Cornet (2), Luncsoara (4), Băiţa Bihor (1), Pietroasa (15), Budureasa (8), Vlădeasa (21), Băişoara (9), Gilau (1)] and care was taken in the field to sample fresh rocks (Table 1; Fig. 2). This is witnessed by the absent to weak alteration observed in thin sections, except in a few samples that were not considered for geochemistry, and by the low loss on ignition (LOI) (0.78 to 3.56 wt% with an average of 1.7 wt%) (Table 2). Moreover, mobile elements such as Ba, Sr, Pb and Mn are not correlated with the LOI.

Table 1 Sample name and location

Table 2 Major and trace element concentrations in the Apuseni Mts. whole-rock samples

A selection of thirty-two samples were crushed with a hammer and pulverized in an agate planetary mill (Gesels 2003). Fused glass discs prepared with lithium tetra- and meta-borate and 0.35 g of rock powder previously dried at 1000 °C for 2 h were used to measure major elements by X-ray fluorescence (ARL 9400 XP wavelength-dispersive (WD)-XRF spectrometer, University of Liège). Some trace elements (Ba, V, Cr, Ni, Cu, Zn) were also measured by X-ray fluorescence on pressed powdered pellets. The rest of the trace elements were determined by inductively coupled plasma mass spectrometry (VG Elemental PQ2+, University of Liège) following the method described by Vander Auwera et al. (1998) (Table 2). Accuracy is 1–3 % for major elements and ≤5 % for trace elements (Table 2).

Electron microprobe analyses of feldspars, amphibole, clinopyroxene and biotite were performed on a selection of polished thin sections from the different intrusions (samples R6 from Cornet, R10 from Luncsoara, R23 from Pietroasa, R30 from Budureasa, R43 from Vlădeasa, and R63 and R64 from Băişoara) with the Cameca SX50 hosted at the Ruhr-Universiteit Bochum (Supplementary Material S1a to S1e). An accelerating voltage of 15 kV was used, and elements were counted for 10 s at a beam current of 10 nA. Silicate and oxide standards were used, and X-ray intensities were reduced with the Cameca PAP correction program.

The isotopic composition of Sr and Nd was measured on a selection of 11 samples covering the different intrusions of the Apuseni Mts. with an upgraded VG54E thermal ionization mass spectrometer (Clermont-Ferrand, France). Samples were dissolved by acid digestion (including dissolution of any refractory residue in pressurized PTFE vessels), and Sr, Sm and Nd fractions were separated by using extraction chromatographic methods adapted from Pin et al. (1994) and Pin and Santos Zalduegui (1997). Samples were loaded on a Ta single filament with a droplet of 3 M H₃PO₄, and the Sr isotope ratios were measured in dynamic triple collection mode, with normalization to ⁸⁶Sr/⁸⁸Sr = 0.1194. After sample loading as a phosphate on single Ta filaments, Sm isotope ratios were measured in the single collection mode. The Nd separate was loaded on the side filaments of a triple Ta–Re–Ta assembly and analysed as the metal ion in dynamic triple collection mode, with normalization to ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219. Twenty-five analyses of the international NBS987 standard have given a Sr isotopic composition of 0.710248 ± 0.000026 (mean and standard deviation of 25 analyses). The French Ames Rennes Nd and the international JNdi-1 standards have given values of 0.511966 ± 0.000015 (eight analyses) and 0.512114 ± 0.000006 (six analyses), respectively (Table 3).

Table 3 Whole-rock Rb–Sr and Sm–Nd isotopic data

Results

Petrography and mineral composition

Our microscopic examination of the samples corroborates previous petrographic descriptions (e.g. Istrate 1978; Stefan et al. 1982, 1992). Samples collected in several discrete bodies display similar mineralogy and textures. The main rock-forming minerals are plagioclase, K-feldspar, quartz, amphibole and biotite. Opaques (magnetite, ilmenite, pyrite and chalcopyrite), apatite, zircon and titanite are ubiquitous accessory phases. Relic cores of clinopyroxene (augite to diopside (Morimoto 1989): Mg# = 0.73–0.85) (see Supplementary Material S1a) were frequently observed in the amphibole which is itself locally surrounded by biotite (Fig. 3a). Plagioclase usually occurs as strongly zoned phenocrysts with cores of labradorite and rims of oligoclase with the exception of sample R30 in which a core of bytownite (An₈₃) has been analysed (see Supplementary Material S1b). Plagioclase is mostly euhedral to hypidiomorphic, but crystals with higher aspect ratio also occur (sample R13). In porphyritic samples, smaller grains of plagioclase are present in the groundmass (Fig. 3b). Plagioclase contains some inclusions of amphibole, opaques and apatite as well as some rare zircon. K-feldspar is usually anhedral, locally perthitic and surrounding grains of plagioclase, opaques, apatite and biotite. K-feldspar is rich in orthose (Or₈₂ to Or₉₆) with a few core analyses ranging from Or₆₀ to Or₇₀ (see Supplementary Material S1c). At the contact with K-feldspar, plagioclase frequently displays partly corroded contours. Quartz is anhedral and fills the interstices between the other minerals. Amphibole, a magnesiohornblende (Leake et al. 2004) with a Mg# ranging between 0.64 and 0.82 (see Supplementary Material S1d), is usually hypidiomorphic (Fig. 3c), very locally included in plagioclase or displaying interstitial contacts with plagioclase phenocrysts. Biotite (Fe# = 0.27–0.64, moderate Ti content: 0.43–0.76 p.f.u.) (see Supplementary Material S1e) is hypidiomorphic to anhedral (Fig. 3d) and locally fills the voids between feldspars and amphibole. Small grains of biotite frequently surround the amphibole. Euhedral grains of apatite are ubiquitous as well as apatite needles dispersed in other minerals. Titanite usually surrounds opaque minerals and is not abundant. In volcanic facies, the matrix is microcrystalline to vitreous. These petrographic observations suggest early crystallization of apatite and opaques followed by plagioclase and clinopyroxene. Amphibole seems to have formed by reaction from the clinopyroxene and was earlier than biotite. K-feldspar and quartz are late crystallizing phases.

Fig. 3

Microphotographic pictures of selected samples from the Late Cretaceous magmatic rocks of the Apuseni Mountains. a Sample R10 (Luncsoara) displaying relic cores of clinopyroxene (Cpx) within the amphibole (Hbl) which is itself surrounded by biotite (Bt) (plane-polarized light). b Sample R53 (Vlădeasa) is porphyritic with large phenocrysts of plagioclase (Pl) and biotite (Bt) (crosspolarized light). c Sample R52 (Vlădeasa) displaying hypidiomorphic amphibole (Hbl) in contact with plagioclase (Pl). Please note the apatite grain (Ap) included in biotite (Bt) (plane-polarized light). d Sample R64 (Băişoara) with biotite (Bt) displaying interstitial contours surrounding plagioclase (crosspolarized light)

The extent of alteration varies among the different samples with some displaying deep alteration and others being totally fresh. Only weakly to non-altered samples have been considered for the geochemistry. Hornblende is transformed into an assemblage of fibrous amphibole (actinote), epidote, secondary biotite and chlorite. Biotite is frequently altered to chlorite and epidote. Alteration is also expressed by damouritization of feldspars. Pronounced alteration into carbonates was noted in sample R20. The microcrystalline groundmass is locally cut by fissures made of calcite, epidote and quartz.

Geochemistry

In the TAS diagram (Fig. 4a), samples from the Apuseni Mts. plot in the fields of basaltic andesites to rhyolites with some samples, higher in K₂O, in the fields of basaltic trachyandesites to trachytes. Most of the samples plot just below the limit between the alkaline and subalkaline series (Miyashiro 1978). The whole trend is calc-alkaline in the AFM (Fig. 4b) and MALI (Frost et al. 2001) (Fig. 4c) diagrams and more precisely, medium- to high-K in the K₂O–SiO₂ diagram of Peccerillo and Taylor (1976) (Fig. 4d). The Apuseni trend is magnesian in the Frost and Frost (2008) classification (Fig. 4e). Our data overlap with previous results (Istrate 1978; Stefan 1980; Stefan et al. 1982, 1992).

Fig. 4

Nomenclature of the Late Cretaceous igneous rocks from the Apuseni Mts. and Banat. a TAS diagram: analyses have been recalculated to 100 % on a volatile free basis, and Fe is considered as total wt% Fe₂O₃ (Le Maitre 1989). Discrimination between the alkaline and subalkaline fields has been made with the boundary line proposed by Miyashiro (1978). b AFM diagram showing the boundary between the calc-alkaline and tholeiitic fields from Irvine and Baragar (1971). c MALI index from Frost et al. (2001). d Subdivision of the calc-alkaline rocks in the K₂O versus silica diagram with the boundary lines from Peccerillo and Taylor (1976). d Ferroan versus magnesian series from Frost and Frost (2008). Grey fields are data on the Late Cretaceous igneous rocks from the Apuseni Mts. from Istrate (1978), Stefan (1980), Stefan et al. (1982) and Stefan et al. (1992). Data for the Late Cretaceous magmatic rocks of Banat are from Dupont et al. (2002)

In the Harker diagrams (Fig. 5), samples from the different intrusions display overlapping trends of decreasing FeO_t, MgO, CaO, TiO₂ and P₂O5 and increasing K₂O with increasing SiO₂·Na₂O remains relatively constant (not shown). The mafic dykes analysed by Istrate (1978), Stefan (1980) and Stefan et al. (1982) overlap in composition with the least differentiated samples and are similar to the calc-alkaline dykes associated with the late Alpine intrusions (e.g. Traversella, Biella, Bergell, Adamello) (Dal Piaz et al. 1979; Venturelli et al. 1984; von Blanckenburg et al. 1992; Bogaerts 1998; Bogaerts and Vander Auwera 1999; Peccerillo and Martinotti 2006) (Fig. 5).

Fig. 5

Variation diagrams of major elements displaying data for the Late Cretaceous igneous rocks of the Apuseni Mts. and Banat (data from Dupont et al. 2002). Data of Kolb et al. (2013) for the Late Cretaceous igneous rocks of Eastern Serbia are shown as light grey fields. Data for the mafic dykes (calc-alkaline) are from Dal Piaz et al. (1979), Venturelli et al. (1984), von Blanckenburg et al. (1992), Bogaerts (1998), Bogaerts and Vander Auwera (1999) and Peccerillo and Martinotti (2006) for the western alpine intrusions and from Istrate (1978), Stefan (1980) and Stefan et al. (1982) for the Apuseni Mts. Data for SH-85-44 as well as experimental melts and cumulates (Run B668 and B704) are from Müntener et al. (2001). See text for explanation

Samples from the different massifs have similar trace element contents (Fig. 6). Among trace elements, Sr, Zn, V, Ga and Ni (not shown) decrease with differentiation, whereas Rb and Th increase and Ba and Zr are rather constant. Two samples from Budureasa (R27 and R29) display higher Ba content than the main trend suggesting some accumulation of biotite and/or K-feldspar, the main Ba carriers in the samples. Three samples (R29, R30 and R34) have significantly higher Zr content than the main trend suggestive of some zircon accumulation. Chondrite-normalized REE patterns (Fig. 7) display LREE fractionation [(La/Yb)N = 3.9–17.4 with an average of 11.7] and negative to slightly positive Eu anomaly ((Eu/Eu*)_N = Eu_N/(√(Sm_N*Gd_N)) =0.51–1.34). Compared to N-MORB in spiderdiagrams (Fig. 8), the Apuseni samples are enriched in incompatible elements (K, Rb, Ba, Th, LREE) and display negative anomalies in Nb, Ta and Ti. The Apuseni samples plot in the volcanic arc field (VAG) in the Pearce et al. (1984) discrimination diagrams (Fig. 9a–d) and partly overlap in composition with arc magmas in the Th/Yb and Nb/Yb diagram of Pearce and Peate (1995) (Fig. 9e).

Fig. 6

Variation diagrams of selected trace elements displaying data for the Late Cretaceous igneous rocks of the Apuseni Mts. and Banat (data from Dupont et al. 2002). Data of Kolb et al. (2013) for the Late Cretaceous igneous rocks of Eastern Serbia are shown as grey fields. Same symbols as in Fig. 4

Fig. 7

REE patterns normalized to the C1 chondrite of Sun and McDonough (1989) of the Late Cretaceous igneous rocks of the Apuseni Mts. Data for Banat (Dupont et al. 2002) and Eastern Serbia (Kolb et al. 2013) are shown as grey fields

Fig. 8

Spiderdiagrams normalized to MORB (Pearce 1983) of the Late Cretaceous igneous rocks of the Apuseni Mts. Data for Banat (Dupont et al. 2002) and Eastern Serbia (Kolb et al. 2013) are shown as grey fields

Fig. 9

a–d Discrimination diagrams (Pearce et al. 1984) of the Late Cretaceous igneous rocks of the Apuseni Mts. and Banat (Dupont et al. 2002). Data for Eastern Serbia (Kolb et al. 2013) are shown as grey fields. e Th/Yb versus Nb/Yb diagram of Pearce and Peate (1995) for the Apuseni and Banat (Dupont et al. 2002) samples. Same symbols as in Fig. 4

Sr and Nd isotopic data of the Apuseni samples have been recalculated back to an age of 80 Ma in agreement with the Re–Os geochronological data on the Mo-Cu skarn associated with the Băița Bihor granodiorite (Zimmerman et al. 2008). In the ε _Ndt versus (⁸⁷Sr/⁸⁶Sr)₈₀ (Fig. 10a), samples from the Apuseni display mildly negative epsilon Nd (−1 to −4) and relatively low Sr initial ratios (0.7055–0.7070) except one sample from the Cornet intrusion which has a positive epsilon Nd (2) and lower (⁸⁷Sr/⁸⁶Sr)₈₀ (0.7046). These Sr isotopic data agree with previous results of Pavelescu et al. (1985) (initial ⁸⁷Sr/⁸⁶Sr of 0.708) reported by Stefan et al. (1992). These data imply source materials that on a time-integrated basis were both weakly enriched to slightly depleted (Cornet sample) in LREE and in Rb relative to Sr.

Fig. 10

Initial Sr, Nd isotopic compositions of the Apuseni and Banat (Dupont et al. 2002) samples (a). Isotopic compositions have been recalculated back to an 80 Ma age. Sr and Nd isotopic data from Dupont et al. (2002) and this study. Evolution of bulk silicate Earth (BSE) is calculated back from present ratios of ⁸⁷Sr/⁸⁶Sr = 0.7047 and ⁸⁷Rb/⁸⁶Sr = 0.0850 (Faure 1986). Data of Kolb et al. (2013) for Serbia and of Wiesinger et al. (2007) for Banat are shown for comparison (b). Mantle array is from De Paolo and Wasserburg (1976)

Barometry and thermometry of the Apuseni samples

P₂O₅ is decreasing with differentiation (Fig. 5) indicating that apatite, an ubiquitous accessory mineral, is a fractionating phase. On the contrary, Zr remains rather constant. In agreement with these observations, apatite (792 to 1075 °C) (Harrison and Watson 1984) and zircon (622 to 752 °C) (Watson and Harrison 1983) saturation temperatures indicate their, respectively, early and late crystallization (Table 2). The Al-in hornblende geobarometer of Johnson and Rutherford (1989) was used on the plutonic samples and confirms the shallow level of emplacement of the intrusions (<0.2 GPa) (see Supplementary Material S4).

Comparison with other Late Cretaceous igneous rocks from the Banatitic belt and with Alpine intrusions

The major element composition of our samples from the Apuseni Mts. is overlapping with the composition of samples from Banat (Dupont et al. 2002) and from Eastern Serbia (Timok and Ridanj-Krepoljin) (Kolb et al. 2013) with, however, a tendency towards slightly higher TiO₂ content in the Apuseni samples (Figs. 4, 5). Samples from the different regions have overlapping FeOt/MgO (Fig. 4b), but samples from the Apuseni have a slightly higher FeOt/MgO than samples from Banat.

Samples from Romania (Apuseni, Banat) and Serbia display similar spiderdiagrams (Fig. 8) and REE patterns (Fig. 7). However, samples from Serbia are generally lower in U, Th, Zr, Hf, Nb, Ta and LREE (Figs. 6, 8) than samples from Romania. Interestingly, samples from the three regions mostly differ by their Ga and also Sr contents. The Ga content decreases from the Apuseni to Banat and then Serbia, but the least differentiated samples from Banat and Serbia have similar Ga content, whereas the Sr content is lower in the Apuseni trend than in the Banat and Serbia trends. These differences are better observed in a Ga versus Sr diagram (Fig. 11).

Fig. 11

Ga versus Sr for the Apuseni and Banat (Dupont et al. 2002) samples. Same symbols as in Fig. 4. Data for Serbia are from Kolb et al. (2013). The composition of Ga and Sr of sample SH 85-44 of Mt. Shasta (Baker et al. 1994) and of the calculated liquids (amphibole- or plagioclase-bearing cumulates) is also shown for comparison (see text for explanation)

Kolb et al. (2013) recognized that a series of samples from Serbia are characterized by high Sr/Y ratios and low Y contents and thus have adakite-like signatures (Defant and Kepezhinskas 2001). This is not observed in the Romanian samples as Sr/Y is lower than 40 except in four samples from Banat, La/Yb is lower than 20 and Y is higher than 18 except in a few samples of Banat.

Samples from Banat and Apuseni have significantly different isotopic composition. In the ε _Ndt versus (⁸⁷Sr/⁸⁶Sr)₈₀ diagram (Fig. 10a), samples from Banat have higher ε _Ndt (0–+4) and lower (⁸⁷Sr/⁸⁶Sr)₈₀ (0.7042–0.7058) than the Apuseni samples (ε _Ndt (−0.7 to −3.7), (⁸⁷Sr/⁸⁶Sr)₈₀ (0.70553–0.70702)). Note that the sample from the Cornet intrusion (Apuseni Mts.) (ε _Ndt = 1.6, (⁸⁷Sr/⁸⁶Sr)₈₀ = 0.70458) plots in the field of the Banat samples. The isotopic composition of samples from Serbia overlaps with that of samples from Banat (Fig. 10b).

As shown in Supplementary Materials Fig 1 and Fig 2, the Apuseni and Banat samples are similar in major and trace element composition to the Adamello and Bergell intrusions from the Alps.

Discussion

Differentiation processes within the Apuseni Mts. trend

Samples from the Apuseni Mts. have been collected in different intrusions/extrusions, but their comparable major and trace element compositions suggest that they differentiated from a similar parent magma. Consequently, in order to constrain the differentiation process(es), we selected samples from the Vlădeasa volcanics and Budureasa pluton because the largest series of samples is available from the former and the latter contains the least differentiated sample and is also close to Vlădeasa.

Mixing between the least and most differentiated magmas is a possible differentiation process as mingling between a microdioritic and a granodioritic facies has been observed in the Pietroasa intrusion. The mixing process has been tested with two sets of samples: samples R34 (56.85 wt% SiO₂) (mafic end member), R39 (71.25 wt% SiO₂) (acid end member) and R50 (64.63 wt% SiO₂) (intermediate facies) as well as samples R30 (52.05 wt% SiO₂) (mafic end member), R39 (71.25 wt% SiO₂) (acid end member) and R44 (62.82 wt% SiO₂) (intermediate facies). These compositions were selected because, in variation diagrams, they plot on the differentiation trends, thus providing the most favourable cases to test the mixing process. All major and trace elements have been considered in the test. The procedure proposed by Fourcade and Allègre (1981) was used:

$$x = \frac{{\left( {C_{\text{mix}}^{i} - C_{\text{fels}}^{i} } \right)}}{{\left( {C_{\text{maf}}^{i} - C_{\text{fels}}^{i} } \right)}}$$

where x fraction of the mafic component, $C_{\text{mix}}^{i}$ the concentration of element i in the intermediate sample, $C_{\text{fels}}^{i}$ the concentration of element i in the felsic end member of the mixing and $C_{\text{maf}}^{i}$ the concentration of element i in the mafic end member of the mixing.

In a C _mix–C _fels versus C _maf–C _fels diagram, the differentiation trend can be predicted by a mixing process if a good-fit regression line with a slope (=x, the fraction of the mafic component) between 0 and 1 can be drawn through the origin (Fig. 12). For both sets of samples, Ba, Zr, Sr and to a lesser extent Zn, V, Rb plot outside of a possible regression line (Fig. 12a, c). When Ba, Zr, Sr are not considered (Fig. 12b) in the first set of samples (R34, R39, R50), the regression line has a plausible slope of 0.65, but the value of the intercept is too low (−3.6) (Fig. 12b). Removing Zn, V, Rb from the regression still gives a too high intercept of −1.6. The intercept can be improved if additional elements such as La, Ce and Nd are also excluded (slope = 0.43, intercept = −0.8). For the second set of sample (R30, R39, R44), removing Ba, Sr, Zr as well as Zn, V, Rb gives a slope of 0.6 similar to the one obtained for the first set of samples with an intercept close to zero (−0.3) (Fig. 12d). The mixing model could thus explain the differentiation trend if a series of elements are not considered. Tests have been made with other samples selected as the intermediate facies, but Zr, Sr, Ba and sometimes Zn and Rb are outside the regression line. In the Zr–SiO₂ diagram (Fig. 6), data display some dispersion that is probably the cause of the poor correlation obtained for this element in the mixing test. However, Rb, Zn and Sr display good differentiation trends with a good alignment of samples R30, R39 and R44 in these variations diagrams. The negative test of the mixing process is thus not due to samples having their composition affected by mineral accumulation. We thus conclude that magma mixing cannot be fully excluded, but it was not a major process during the differentiation of the Late Cretaceous Apuseni magmas.

Fig. 12

Test of a mixing process according to Fourcade and Allègre (1981). C _maf, C _fels and C _mix are the concentrations in the mafic, felsic and mixed facies. The hypothetical proportion (X) of the mafic component in the mixing process is given by the slope of the regression line. See text for explanation

Fractional crystallization and batch partial melting can be discriminated in a log C _A –log C _B diagram using the most compatible element (A) versus the most incompatible element (B) (Joron et al. 1978; Martin 1987). Indeed, the slope of the trend is $\frac{{D_{B} \left( {1 - D_{A} } \right)}}{{D_{A} \left( {1 - D_{B} } \right)}}$ and $\frac{{D_{A} - 1}}{{D_{B} - 1}}$ for a partial melting and a fractional crystallization process, respectively. As D_A and D_B are significantly higher and lower than 1, respectively, the slope of the trend can be simplified and will be close to -D_A for a fractional crystallization process and -D_B for a partial melting process. Among trace elements, Th appears as strongly incompatible and Zn, as strongly compatible. For the Vlădeasa and Budureasa samples, the slope of the trend in a log_Zn–log_Th diagram is −1.1. A fractional crystallization process thus better predicts the differentiation trend.

Modelling of the fractional crystallization process

The fractional crystallization process has been modelled with the least square regression method in three steps that are summarized in Table 4. In the first step, subtraction of a gabbronoritic cumulate drives the liquid from 53.64 % SiO₂ to 56.95 SiO₂ leaving 73 % of residual liquid at this stage. Further differentiation to a proportion of 36 % of residual liquid (64.73 % SiO₂) is produced by subtraction of a dioritic cumulate containing apatite. The most evolved composition (72.43 % SiO₂) is finally reached when the proportion of residual liquid is 21 %. The third cumulate is also dioritic but with a more albitic plagioclase.

Table 4 Results of least-squares modeling calculations for major elements

The composition of the cumulates derived with the least square regression method on major elements can be tested using the trace element composition and the Rayleigh distillation law:

$$C_{L} = \, C_{0} . \, F^{(D - 1)}$$

where C ₀ and C _L are the concentrations of the trace element in, respectively, the starting composition and the residual liquid, F is the fraction of residual liquid that has been calculated with the least square regression method and D, the bulk partition coefficient, equals ΣD _i.X _i. D _i is the partition coefficient of the trace element between mineral i and the liquid, and X _i is the proportion of this mineral in the subtracted cumulate. X _i has also been calculated with the least square regression method.

Partition coefficients were selected from the literature for andesitic to dacitic compositions and are given in Table 5. In the first step of differentiation, the calculated composition of the residual liquid compares rather well with the composition of sample R27 except for Ba and Zn (Table 6). Ba is lower in the calculated liquid (781 ppm) than in R27 (1514 ppm). However, sample R27 as well as samples R29 and R34 have a Ba content significantly higher than the main trend suggesting some mineral accumulation. The calculated Ba content is within the range observed in the samples having a SiO₂ content close to that of R27 (referred to L1 samples below and as “Range L1” in Table 6). The calculated Zn content in L1 (93 ppm) is higher than the Zn content of sample R27 (64 ppm) but hereto in the range of L1 samples. In the second step, the composition of sample R27 has been considered as the starting composition except for Ba, Co and Zn. For these three elements, the starting content has been interpolated from the main differentiation trend displayed by the Budureasa and Vlădeasa samples. The calculated composition is very close to that of sample R44 (Rb, REE, V, Co, Zn) or in the range of the L1 samples (Sr, Ba). Similarly, in the third step, the calculated composition is very close to the trace element content of sample R39 (Sr, Ba, REE, V, Zn) or close to the L1 samples (Rb, Co). The trace element contents thus support the mineral proportions of the three cumulates derived from the least square regression method.

Table 5 Partition coefficients for trace elements

Table 6 Results of trace elements modelling

The parent magmas of the Apuseni and Banat Late Cretaceous igneous rocks

Because the Late Cretaceous igneous rocks from Banat and the Apuseni Mts. crosscut the Mid-Cretaceous nappe piles and are extruding in Gosau-type basins, we consider that they were emplaced during the post-collisional stage of the belt. Their high-K calc-alkaline character, that is typical of this geodynamic setting (Liégeois et al. 1998), supports this conclusion. Harris et al. (1986) pointed out that post-collisional and arc magmas are difficult to discriminate geochemically and further suggested that the mantle component of the post-collisional magmas is produced from a source above a subduction zone, i.e. a mantle wedge that has been enriched in LILE. In the following, we will thus discuss the petrology of the Late Cretaceous igneous rocks using experimental and geochemical data obtained on arc magmas.

As the least differentiated samples from the Apuseni Mts. and Banat have a low Mg# (below 0.5 in the Apuseni and below 0.55 in Banat) as well as Cr and Ni contents below 55 ppm and 25 ppm, respectively, these samples cannot be representative of primary magmas in equilibrium with a mantle source as these magmas are inferred to have an Mg# > 0.7 and high Cr and Ni contents (e.g. Grove et al. 2012). Compared to sample SH-85-44C, recognized as a near-primary magma of Mont Shasta (Cascade Range, USA) (Grove et al. 2012), the least differentiated compositions of the Banat and Apuseni Mts. trends are indeed significantly lower in MgO (Fig. 5). Consequently, the least differentiated samples, parent magmas of the Apuseni and Banat differentiation trends, were either produced by partial melting of a lower crustal source or are the result of lower crustal differentiation from a precursor mantle melt as interpreted for the Late Cretaceous magmatism from Serbia (Kolb et al. 2013) and for the Tertiary Alpine intrusions such as Bergell (von Blanckenburg et al. 1992, 1998) and Adamello (Ulmer et al. 1983; Macera et al. 1983; Kagami et al. 1991). Partial melting of a lower crustal source to produce mafic magmas would require extensive melting of an ultramafic source at very high temperature. This process seems unlikely as there is no evidence of an important thermal anomaly associated with the Late Cretaceous magmatism. We thus retain the second hypothesis, lower crustal differentiation from a mantle-derived melt. Moreover, as the parent magmas of the Banat and Apuseni Mts. differentiation trends have overlapping major and trace element composition, except for Ga and Sr (see discussion in the next section), we consider that they were derived from the same mantle melt.

Lower crustal differentiation of the mantle-derived melt and the question of Ga and Sr

It has already been pointed out that in a Ga–Sr diagram (Fig. 11), the Apuseni Mts. magmatic rocks display a significantly higher Ga content and a slightly lower Sr content than the Banat magmatic rocks. This also holds for the least differentiated samples (parent magmas) of the two trends. These differences could result either from precursor mantle melts with different Sr and Ga contents or from the differentiation processes that produced the parent magmas of both trends from the mantle melts.

In a subduction-related geodynamic setting, the composition of primary magmas reflects contributions from the slab (including subducted sediment: Plank and Langmuir 1993) and the overlying mantle wedge (e.g. Gill 1981; Plank and Langmuir 1993). These magmas are enriched in LILE (Rb, Cs, Th, Pb) relative to the HFSE (Nb, Ta, Zr, Hf, TiO₂) that display negative anomalies compared to primary MORBs or OIBs, and these geochemical patterns are considered as resulting from the slab contribution (e.g. Gill 1981; Plank and Langmuir 1993). In a Th/Yb versus Nb/Yb diagram, arc magmas have thus notably higher Th/Yb ratios than the MORB array (Pearce 1983; Pearce and Peate 1995) (Fig. 9e). Ga and Sr have different geochemical behaviours: Ga is not contributed from the slab (in a Ga/Yb versus Nb/Yb diagram arc magmas plot in the MORB array), whereas Sr is (Pearce and Peate 1995). Consequently, variable trace element content of the mantle source would not produce high Ga and low Sr contents (Apuseni parent magma) versus low Ga and high Sr contents (Banat parent magma). Moreover, we note that in the Th/Yb versus Nb/Yb diagram (Fig. 9e), the Apuseni Mts. and Banat samples overlap in composition suggesting similar contribution of LILE from the slab.

We have tested the hypothesis that the different Ga and Sr contents were produced during the lower crustal differentiation of a mantle-derived magma using the experimental data of Müntener et al. (2001) performed at 1.2 GPa on sample SH-85-44, a near-primary magma composition at Mont Shasta. These experimental data indicate that when the H₂O content of the melt is above 3 wt%, amphibole crystallizes early and the liquidus temperature of plagioclase is lowered. More particularly, when their starting composition has an H₂O content of 5 wt% (run B704 of Müntener et al. (2001)), the cumulate contains amphibole (45 %), orthopyroxene (19 %), clinopyroxene (36 %) and trace amount of garnet, whereas in their run B668 with a lower H₂O content (2.5 wt% in the starting composition), the melt is in equilibrium with a cumulate made of plagioclase (27 %), orthopyroxene (34 %) and clinopyroxene (39 %). The experimental melts of runs B704 (with amphibole) and B668 (with plagioclase) have a major element composition very close to that of samples 98R12 (Dupont et al. 2002) and R30, parent magmas of the Banat and Apuseni differentiation trends, respectively (Fig. 5) supporting the hypothesis that these mafic magmas were derived by differentiation from a precursor mantle melt. Using appropriate partition coefficients (amphibole: 1.77 (Ewart and Griffin 1994), orthopyroxene: 0.32 (Ewart and Griffin 1994), clinopyroxene: 0.69 (Blundy et al. 1998), plagioclase: 0.08 (Blundy et al. 1998)), a Ga content of 17 ppm for the primary magma (sample 85-44 of Mount Shasta: Baker et al. 1994) and the Rayleigh distillation law, the Ga content of the residual melt is 24 ppm and 16 ppm when subtracting the plagioclase-bearing and amphibole-bearing cumulates, respectively. This agrees very well with the observed Ga contents of samples R30 (29 ppm) (parent magma of the Apuseni trend) and 98R12 (16.2 ppm) (parent magma of the Banat trend). Similarly, using the Sr content of sample 85-44 (295 ppm: Baker et al. 1994), the calculated Sr content of the residual melt is higher when subtracting the amphibole-bearing cumulate (524 ppm) than the plagioclase-bearing cumulate (445 ppm). The calculated values are lower than the observed Sr contents [766 ppm in sample 98R12 (Banat); 639 ppm in sample R30 (Apuseni)], but precise constraints on the Sr content of the mantle-derived magma are not available and subduction-related primary magmas have rather variable Sr contents (41 up to 745 ppm: Grove et al. (2012) and references therein). When considering 431 ppm Sr in the primary mantle melt, in the range of the Sr content of primary magmas, the calculated Sr content of the residual liquid is 766 ppm and 651 ppm with the amphibole- and plagioclase-bearing cumulates, respectively. The slightly higher TiO₂ content of the Apuseni differentiation trend (Fig. 5) compared to the Banat trends can also be predicted with these two different cumulates as amphibole has a higher TiO₂ content than plagioclase. It is thus plausible that the different Ga and Sr contents in the parent magmas of both trends (Banat versus Apuseni) are due to a different H₂O content of the mantle-derived magma. A higher H₂O content in the primary magma of Banat induces early crystallization of amphibole, thus decreasing the Ga content and increasing the Sr content in the derivative liquids, whereas a lower H₂O content in the Apuseni induces early crystallization of plagioclase, thus decreasing the Sr content and increasing the Ga content in the derivative liquids. Interestingly, the Late Cretaceous magmatic rocks from Serbia (Timok, Ridajn-Krepolijn), for which amphibole was also considered as a high- and/or low-pressure fractionating phase (Kolb et al. 2013), display similar Sr, Ga and TiO₂ contents than the Banat trend (Fig. 11).

Crustal contamination

As mentioned above, samples from the Apuseni Mts. and Banat have different isotopic compositions except that the sample of Cornet (Apuseni Mts.) plots in the field of samples from Banat (Fig. 10), maybe because, before its final emplacement in calcareous marbles, this latter intrusion has crosscut the formations of the Mureş zone (Fig. 2) that are essentially made of Jurassic mafic rocks of MORB affinity overlain by calc-alkaline lavas (Bortolotti et al. 2002). The sample from Băiţa Bihor is also close to the Banat field. Most Banat samples plot on the mantle array and two of them (Oraviţa) are also close to the composition of near-primary arc magmas from North America (Mt-Shasta: Grove et al. 2002) and Adamello (Kagami et al. 1991). The rest of the Romanian banatite samples are displaced towards higher initial Sr isotopic composition and lower epsilon Nd (Apuseni Mts.) compared to the mantle array. Instead of displaying a continuous trend of decreasing epsilon Nd and increasing Sr_i, the Late Cretaceous igneous rocks of Romania more or less define three discrete groups (Fig. 10). One group plots close to the composition of primary arc magmas. A second group, comprising most of Banat samples has mildly positive epsilon Nd and clusters around an Sr_i of 0.705. The third group includes most of the Apuseni samples and displays negative epsilon Nd and high Sr_i. This split in three groups is in agreement with the lack of correlation with wt% SiO₂. The first group likely witnesses the isotopic composition of the mantle source. The other two groups probably evidence crustal contamination. As the Banat and Apuseni Mts. igneous rocks were emplaced in two different crustal segments, respectively, the Dacia and Tisza blocks, the occurrence of distinct isotopic composition probably results from the interaction with two different crustal contaminants. The Sr and Nd isotopic compositions of the geological formations that make the Tisza and Dacia blocks have not been extensively studied, but data are available for the granitoids belonging to the basement of the Getic nappes in the South Carpathians (Duchesne et al. 1997, 2008) (Dacia block) (Fig. 13a), the pre-Alpine crust in the Apuseni Mountains (Fig. 13b) [northern and southern terranes as well as the Highis–Biharia zone; Permian volcanics (Nicolae et al. 2014); Muntele Mare granitoid and surrounding rocks (Somes Series) (Anton 2000; Balintoni et al. 2014; Balintoni, pers comm) and the Variscan Mecsek granitoid from the Tisza block in Hungary (Klötzli et al. 2004)]. Available isotopic data show rather overlapping composition in the two blocks with, however, higher Sr_i and more negative epsilon Nd in the Tisza block, especially for the granito-gneisses analysed by Pană et al. (2002) for which no Sr isotopic composition is given. Crustal contamination has been tested with an AFC model as described by De Paolo (1981) and Ersoy and Helvaci (2010). As in the epsilon Nd_t versus Sr_i, the Romanian samples seem to cluster in three discrete groups, and within each of these groups, the isotopic composition (Sr_i, epsilon Nd) is not correlated with the SiO₂ content, and we suggest that assimilation occurred very early, probably when the primary mantle-derived magma differentiated towards the composition of the parent magmas of the Apuseni and Banat trends. We used the average isotopic composition of the Mecsek (Klötzli et al. 2004) and Sichevița (Duchesne et al. 2008) granitoids (Fig. 13c, d) as possible contaminants of the Apuseni and Banat magmatic rocks, respectively. The Sichevița intrusion indeed belongs to the Getic nappes that have been crosscut by the Late Cretaceous magmas. The Sr and Nd contents of the bulk continental crust (Sr = 320 ppm; Nd = 20 ppm) (Rudnick and Gao 2003) were selected for the element composition of the contaminants, and the experimental amphibole- (Fig. 13c) and plagioclase-bearing (Fig. 13d) cumulates of Müntener et al. (2001) were used to model the fractional crystallization process. Results are displayed in Fig. 13c, d and show that with an r value (fraction of crust assimilated versus crystals formed) of 0.3–0.4, it is possible to predict the isotopic composition of the Banat and Apuseni magmatic rocks, thus supporting the hypothesis of two distinct contaminants.

Fig. 13

Epsilon Nd_t versus initial Sr isotopic compositions of the Apuseni and Banat (Dupont et al. 2002) samples compared with possible contaminants of the Getic nappes (a) (Duchesne et al. 1997, 2008) and of the Apuseni Mountains [Permian volcanics (Nicolae et al. 2014); Muntele Mare granitoid and surrounding rocks (Somes Series) (Anton 2000; Balintoni et al. 2014; Balintoni, pers comm); granito-gneisses from the Apuseni Mts. (Pană et al. 2002)] and the Mecsek granitoid from the Tisza block in Hungary (Klötzli et al. 2004) (b). AFC model (De Paolo 1981; Ersoy and Helvaci 2010) using the Sr and Nd isotopic composition as well as the Sr and Nd content (Grove et al. 2002) of sample 85-44 (Mt-Shasta) as the starting composition and, as the contaminant, the average Sr and Nd isotopic composition of the Sichevița (Duchesne et al. 2008) (c), or the Mecsek granite (Klötzli et al. 2004) (d)

Conclusions

The geochemical and isotopic data presented here provide constraints on the petrology of the Late Cretaceous igneous rocks of Romania and lead to the following conclusions:

• 1. The parent magma of the Apuseni Mountains Late Cretaceous magmatism is a basaltic andesite that differentiated to rhyolitic composition mainly by a fractional crystallization process. The Al-in hornblende geobarometer indicates that this process occurred in upper crustal storage chambers.

• 2. Because of their too low Mg#, the parent magmas of the Apuseni and Banat differentiation trends cannot be considered as primary magmas. The experimental data acquired by Müntener et al. (2001) on a primary magma at 1.2 GPa indicate that they could have been produced by lower crustal fractionation from a primary mantle melt.

• 3. The Apuseni and Banat magmatic rocks display overlapping differentiation trends in variation diagrams except for Ga and to a lesser extent Sr. The parent magma of the Apuseni trend has a higher Ga and slightly lower Sr content than the parent magma of the Banat trend. This difference can be taken into account by a higher H₂O content in the Banat mantle-derived magma which induces early crystallization of amphibole at the expense of plagioclase. Amphibole has indeed a high and low partition coefficient for Ga and Sr, respectively. As the Timok and Ridanj-Krepolijn Late Cretaceous magmatic rocks (Kolb et al. 2013) overlap with the Banat samples in the Ga–Sr diagram, this suggests that the H₂O content of the mantle-derived magmas varied along the belt.

• 4. The Banat and Apuseni samples have contrasting isotopic compositions that can be taken into account by contamination with two different contaminants in agreement with the emplacement of the Banat and Apuseni magmatic rocks in two different crustal terranes, respectively, Dacia and Tisza.