Introduction

The European Cenozoic Rift System (ECRIS; Prodehl et al. 1995) is a major NE–SW-trending rift zone in western and central Europe. It extends from the Spanish Valencia Trough across the French Massif Central (Rhône Depression, Limagne Graben, Bresse-Saône Graben system), Black Forest and Vosges (Rhine Graben with Kaiserstuhl), Hegau, Urach and in the area of the Rhenish Massif (Lower Rhine Embayment, Hessen Depression), Vogelsberg, Rhön, Heldburger Gangschar) to the Bohemian Massif (Ohře/Eger Graben) (Wimmenauer 1974; Kopecký 1978; Wilson and Downes 1991; Ziegler 1994; Prodehl et al. 1995; Ulrych and Pivec 1997; Dèzes et al. 2004; Lustrino and Wilson 2007). The ECRIS lies mostly (volcanism in removed Lower Rhine Graben is missing) in the northern foreland of the Alps–Carpathian arc and is closely linked spatially and temporally to the Alpine Orogeny (e.g., Ziegler 1990, 1994; Prodehl et al. 1995).

Similar to other places in the ECRIS, magmatic rocks of the Ohře rift (OR) are typically anorogenic mildly silica-undersaturated alkaline types (e.g., Ulrych et al. 2002, 2011; Lustrino and Wilson 2007). However, the OR contains, in addition to volcanic rocks, hypabyssal rocks of the high-level intrusions, which are rare in the other parts of the ECRIS. The presence of this kind of rocks can provide constraints on the genesis of the hypabyssal and subvolcanic level of volcanism and refine the ideas on the rift history. They can also contribute to a recent debate on whether intraplate magmatic activities in the ECRIS are related to mantle plumes or to passive adiabatic upwelling of sub-lithospheric mantle. A variety of processes have been suggested for the origin and development of the ECRIS magmatism (e.g., Lustrino and Carminati 2007 and references therein). They range from a model implying the presence of a “classic” mantle plume (Le Bas 1987; Wilson and Downes 1991), large-scale mantle plumes (Hoernle et al. 1995), channeled mantle plumes with the stem of the system located very far away beneath Cape Verde Island (Oyarzun et al. 1997) or Iceland (Wilson and Patterson 2001), a really limited deep mantle plumes (Goes et al. 1999), small plumes originated around the top of the mantle transition zone (“finger plumes”; Granet et al. 1995) or local passive diapiric upwelling (Lustrino and Wilson 2007), variably related to Alpine subduction processes (Foulger and Meyer 2007).

The purpose of this paper is to present whole-rock major and trace element concentrations, Sr–Nd isotope ratios and mineral chemical compositions of hypabyssal and associated dike rocks of the Roztoky Intrusive Complex (RIC), a major intrusive center of the České středohoří Mts. These data are used to constrain the petrogenesis, the mantle sources and tectonic geologic setting of the igneous activity.

Geological setting and characteristics of the Roztoky Intrusive Complex

Magmatism of the ECRIS in the Bohemian Massif (Czech Republic) represents the easternmost part of the Central European Volcanic Province (CEVP of Wilson and Downes 1991; Hoernle et al. 1995; Ulrych et al. 1999, 2011; Lustrino and Wilson 2007). In the Czech Republic, most voluminous Cenozoic magmatic activities occurred within the ENE–WSW-trending OR zone. The rift zone is thought to be a reactivated Variscan tectonic suture of the Bohemian Massif by the Alpine Orogeny. It is situated between the Moldanubian and Teplá-Barrandian blocks in the southeast and the Saxothuringian block in the northwest (Babuška and Plomerová 1992, 2001, 2010). Babuška et al. (2007) even invoke existence of the “triple junction” of the Saxothuringian—Teplá-Barrandian and Moldanubian lithospheric units in Western Bohemia.

In the Bohemian Massif, the magmatic rift system forms an arc-shaped belt extending over 500 km from the western to the easternmost parts of the massif (Franke 1989; Ziegler 1990, 1994; Fig. 1). The more pronounced western segment of the belt is a linear SW–NE-oriented structure bounded between the NW–SE striking Franconian and Labe-Odra wrench fault systems and contains the OR graben, which is about 180 km long and reaches a maximum width of 25–30 km. The OR contains a significant volume of volcanic rocks, which are locally associated with hypabyssal complexes. Further to the east and southeast, the volcanics occur in a more scattered form within the Labe-Odra wrench fault system (Fig. 1).

Fig. 1
figure 1

Overview map of the Bohemian Massif with graben structures and Cenozoic igneous rocks of the Eger-Ohře graben. The Roztoky Intrusive Complex is indicated by circled star. Inset diagram shows lithospheric thickness contour lines in km (a). Detailed map showing the occurrence of different rock types of the Roztoky Intrusive Complex (b). BM Bohemian Massif, ChB Cheb Basin, LF Lusatian Fault, MLF Mariánské Lázně Fault, OR Ohře (Eger) Rift

Full size image

The crustal thickness in the Bohemian Massif is well constrained by numerous reflection and refraction seismic profiles (Hrubcová et al. 2008 and references therein) as well as receiver functions studies (Heuer et al. 2007; Geissler et al. 2008, 2012). These geophysical studies indicate the presence of a relatively shallow Moho (H RF = 25–31 km) in the western part of the Bohemian Massif (e.g., Ziegler and Dézes 2006). This zone of thin crust, represented by the western OR graben, extends from the Western Bohemia/Vogtland region, where the Moho is shallowest (H RF = 25–28 km), to the north and northeast toward Eastern Bohemia. The zone in Western Bohemia shows geochemical evidence (based on the fumarolic 3He/4He and CO2 isotopic data) for ascending mantle-derived melts (Bräuer et al. 2005).

The distribution of Cenozoic volcanism associated with the OR is tectonically controlled. The two main volcanic centers within the OR are formed by the Doupovské hory Mts. and České středohoří Mts. complexes. They occur at the intersection of the OR with transverse NW–SE-trending faults (Fig. 1). The Doupovské hory Mts. Volcanic Complex, about 30 km in diameter, covering an area of nearly 600 km2, crops out at the western part of the graben (Holub et al. 2010 and references therein). The České středohoří Mts. Volcanic Complex, further to the east, spreads over about 500 km2 (Shrbený 1995).

The ages of igneous rocks emplaced in the OR range from late Cretaceous to Quaternary (~79 to 0.26 Ma; Ulrych et al. 2011). On the basis of K–Ar data and a paleostress chart, Ulrych and Pivec (1997) and Ulrych et al. (2011) defined three periods of Cenozoic magmatic activity in the OR: (1) pre-rift period (79–49 Ma), (2) syn-rift (42–16 Ma) and (3) late-rift period (16–0.26 Ma). This magmatic evolution in the OR is comparable to that of French Massif Central (Michon and Merle 2001). The volcanic rocks of the OR are mostly sodic SiO2-undersaturated types forming two synchronous alkaline rock series: the dominant mildly alkaline basanite–trachyte series (MAS) and minor strongly alkaline nephelinite–phonolite series (SAS; Ulrych and Pivec 1997).

The primitive mafic volcanic rocks from the OR have relatively uniform Sr–Nd–Pb isotope compositions with a common sub-lithospheric mantle source component (Blusztain and Hart 1989; Wilson et al. 1994; Lustrino and Wilson 2007; Ulrych et al. 2011). The isotopic and incompatible element characteristics of primitive mafic magmas of the OR are similar to those of the HIMU-type ocean island basalts (OIB) attributed to recycling of subducted oceanic crust in the upper mantle (Alibert et al. 1987; Lustrino and Wilson 2007) referred to the Common Magmatic Reservoir (CMR) by Lustrino and Wilson (2007).

The České středohoří Mts. represents an erosional relict within the OR graben. The largest center of magmatic activity of the České středohoří is the Roztoky Intrusive Complex (RIC; Fig. 1). The intrusive magmatic activity of the RIC lasted from ~33 to 25 Ma (Ulrych and Balogh 2000; Ulrych et al. 2011). The center is composed of an elliptical crater vent (3 × 1.5 km) filled with trachytic breccia and a cluster of trachytic to trachybasaltic and phonolitic dike intrusions. Large intrusion of olivine nephelinite occurs at northern margin of the RIC. Problematic carbonate dikes with Pb–Zn–Cu (Ag–Te) mineralization penetrating the monzodiorite intrusion at Roztoky (Pivec et al. 1998) show trace element and isotopic compositions similar to that known from late-carbonatites (Ulrych et al. 1997). Hypabyssal intrusions of monzodiorite, essexite, sodalite syenite as well as a radial dike swarm of lamprophyres and felsic differentiates occur in close proximity to the center (Ulrych 1983, 1998; Ulrych et al. 2006; Jelínek et al. 1989). The geophysical survey of Mrlina and Cajz (2006) found that essexite and monzodiorite intrusions at Roztoky form a single deep body that has a few protrusions reaching the surface.

The volcanic activities of the České středohoří Mts., stretching in the NE part of the Ohře Rift, are represented mostly by primitive alkali basaltic lavas, pyroclastics/volcaniclastics and rare intrusions of the similar composition. Four volcanostratigraphic formations of the latest Eocene to Late Miocene age (first two with an overlap of K–Ar ages) were recognized in this volcanic complex by Cajz et al. (1999), and Ulrych et al. (2002):

  1. 1.

    The Ústí Formation (age 36.1–25.5 Ma): Mg# 82–65, average 75; 87Sr/86Sr 0.70313–0.70353, 143Nd/144Nd 0.512738–0.512849, εNd 2.8–4.8 comprising lavas and volcaniclastics of primitive basanitic, rarely olivine nephelinite composition. The basanitic magma is products of the initial phase of rifting, representing the near-primary magma of mantle source. A relatively rapid magma ascent (presence of angular mantle xenoliths), probably along the main rift-forming fault structures, can be supposed. Rejuvenation of tectonic movements associated with the development of the Ohře Rift might have caused the periodically repeated activity of the magma chamber. The concept of pulsation character of magmatic chamber in a broader time span is broadly accepted (Bellon et al. 1998);

  2. 2.

    The Děčín Formation (age 30.8–24.7 Ma): Mg# 72–60, average 65, 87Sr/86Sr 0.70443–0.70465, 143Nd/144Nd 0.512979–0.512742 εNd 1.5–2.6 comprising trachybasaltic lavas (prevailing trachybasalts transiting to basaltic trachyandesites) and pyroclastites, and hypabyssal bodies of the RIC (33.1–25.6 Ma; Ulrych and Balogh 2000). The Labe/Elbe valley is the only part of the České středohoří Mts. where the deeper levels of the volcanic pile, including hypabyssal bodies, are exposed. The trachybasaltic magma represents product influenced by crustal material. Geochemical and rheological characteristics result in the longer evolution within the independent high-level chamber and slower ascent of more acid magma accompanied by stopping and/or zone refinement. Both magma chambers occurring in different deep levels have been coeval, and feeding channels have been active at the same time. Similar time–coeval existence of two magma sources was presented by Wedepohl (1985) from Hessian Basins, Germany. The hypabyssal intrusions of the RIC represent probably the relatively older deep-seated product of this formation;

  3. 3.

    The Dobrná Formation (age 24.0–19.3 Ma): Mg# 79–70, average 75, 87Sr/86Sr 0.70365–0.70376, 143Nd/144Nd 0.512845–0.512847, εNd 4.3–4.4 consisting of a primitive basanitic lava;

  4. 4.

    The Štrbice Formation (13.9–9.0 Ma): Mg# 80–69, average 75, 87Sr/86Sr 0.70365–0.70376, 143Nd/144Nd 0.512845–0.51247, εNd 4.4–4.3 consisting of intrusions and lavas of basanitic composition.

One of the largest and well-exposed hypabyssal intrusions of the RIC, which is composed of monzodiorite, is rather heterogeneous (Fig. 2). It shows a steeply dipping magmatic rhythmic layering of rocks with variable grain size, texture and clinopyroxene/biotite ratios as well as bands of leuco-diorite and mela-monzodiorite. A similar vertical magmatic layering was described from the Monteregian Hills, Quebec, Canada, and it was interpreted to be the result of flowage differentiation, a process characteristic of the magma emplacement in a rift setting (Bhattacharji and Nehru 1972; Philpotts 1974; Eby 1984, 1985a, b). The RIC intrusions and related lamprophyric dikes contain enclaves of hornblendite cumulates, up to 50 cm in diameter.

Fig. 2
figure 2

Vertical section of the central part of the Roztoky monzodiorite body showing rhythmic, partly diffuse, layering. Exposure at the railroad cut near Povrly-Roztoky stop at the Ústí nad Labem–Děčín railroad line, Northern Bohemia. Biotite–clinopyroxene facies: (1) dark fine-grained, (2) dark medium-grained, (3) light fine-grained, (4) dark medium-grained, (5) light medium-grained, (6) dark coarse-grained, (7) light coarse-grained. Clinopyroxene–biotite facies: (8) fine- to coarse-grained (dark in color), (9) fine- to coarse-grained (light in color), (10) coarse-grained to pegmatitic (light in color), (11) dike rocks, (12) soils and hillside slopes, (13) faults indicated and supposed

Full size image

The Doupovské hory Mts. Volcanic Complex is nearly coeval (~30 Ma) with the České středohoří Mts. Complex. The phaneritic rocks of this complex belong to the essexite–monzodiorite–monzosyenite association, with a composition similar to that of the RIC rocks. Here, clinopyroxenites and rocks of the melteigite–ijoilite–urtite series occur predominantly in the marginal parts of the complex (Holub et al. 2010).

The RIC is associated with a dike swarm consisting of about 1,000 dikes. They are mostly radially arranged, and the swarm has a radius of about 15 km. The dike system of the RIC is composed of: (1) lamprophyres (monchiquite, camptonite; ~56 % of dikes), (2) medium derivatives (maenite of Rock 1991; local names “gauteite” and “bostonite”; ~28 %), (3) felsic derivatives (tinguaite; ~1 %), (4) basaltic rocks (~11 %) and (5) phonolites and trachytes (~4 %).

The hypabyssal rock series of the RIC exhibits many similarities in modal composition and geochemistry (e.g., the presence of essexites, sodalite syenites, both mafic and felsic dike rocks) with the Cenozoic volcanic complex of Kaiserstuhl in Germany (e.g., Wimmenauer 1974; Keller 1981).

Analytical methods

Twenty-four fresh samples were selected for this study. Whole-rock major element concentrations were determined by the wet chemical analysis methods at the Faculty of Science, Charles University in Prague, using wet chemical methods described in Johnson and Maxwell (1981) and Potts (1995). Analyses of the reference materials SY-4 CCRMP Canada and AGV-2 (USGS USA) yielded a total error of ±5 % (1σ).

The trace element analyses of whole-rock samples were carried out after modified total digestion in acids (HF + HClO4) and borate fusion (Na2CO3 + Na2B4O7) in platinum crucibles followed by solution nebulization using an ICP-MS PQ3 VG Elemental at Charles University in Prague. The analytical protocol and calibration strategy closely followed those described in Strnad et al. (2005). The relative standard deviation (RSD) of the ICP-MS data is 0.5–5 %. The accuracy as determined on the reference materials AGV-2 and BCR-2 (USGS, USA) measured as a ratio 100× sample/standard is 91–105 rel% for all elements.

Mineral analyses were carried out on a CAMECA SX 100 electron microprobe (EMP) at the Institute of Geology, v.v.i., Academy of Sciences of the Czech Republic, Prague. The analytical protocol used is the same of Ulrych et al. (2010).

The Sr–Nd isotope compositions were determined in the isotope laboratory at Universität München (Germany) according to the procedures outlined in Hegner et al. (1995). The 143Nd/144Nd ratios were determined with a Finnigan MAT 261 using a dynamic triple mass method and monitoring 147Sm; Sm isotopes were determined in static data collection mode. The 143Nd/144Nd ratios were normalized to 146Nd/144Nd = 0.7219 and 147Sm/152Sm = 0.56081. The 143Nd/144Nd ratio of the in-house Ames Nd standard solution was 0.512142 ± 12 (n = 35), corresponding to 0.511854 in the La Jolla Nd reference standard material. The εNd(t) values were calculated with the parameters of Jacobsen and Wasserburg (1980). Present-day ratios for the chondrite uniform reservoir (CHUR) were the following: 147Sm/144Nd = 0.1967, 143Nd/144Nd = 0.512638 (Jacobsen and Wasserburg 1980; 143Nd/144Nd re-normalized to 146Nd/144Nd = 0.7219). 87Sr/86Sr ratios were determined with a dynamic double mass method, monitoring 85Rb, and normalized to 86Sr/88Sr = 0.1194. The NIST 987 reference material yielded 87Sr/86Sr = 0.710230 ± 11 (n = 22).

Petrographic characteristics of hypabyssal and dike rocks of the Roztoky Intrusive Complex

The main petrographic characteristics of the hypabyssal rocks and dikes of the RIC are summarized in Table 8 of Appendix 1.

Hypabyssal rocks

Essexite (nepheline monzodiorite)

Three distinct petrographic types of the essexite were recognized as follows:

The dark fine- to medium-grained type, equigranular with hypidiomorphic texture is the most common (Fig. 16a of Appendix 2). It occurs in the central part of the intrusions. Fine grains of homogeneous plagioclase and natrolitized nepheline are the dominant felsic minerals. Next to them also larger laths of columnar zoned plagioclase and K-feldspar infilling of intergranular spaces occur rarely. Clinopyroxene is the predominant mafic mineral showing both sector and oscillation zoning. Biotite and amphibole (overgrowths or replacing of clinopyroxene) are also abundant.

The light fine- to medium-grained type, equigranular with prismatic granular texture, is common in marginal parts of the bodies, but also forms lenses in the central part (Fig. 16b of Appendix 2). The lath-shaped plagioclase is the dominant mineral. The relatively fresh columnar nepheline and K-feldspar fillings are relative less frequent. The most common mafic mineral is clinopyroxene, which is similar to that of the dark types. Biotite and amphibole are present in subordinate amounts.

The porphyritic type is characterized by the presence of 1–2-cm-long prismatic clinopyroxene phenocrysts enclosed in a medium-grained matrix composed of lath-shaped plagioclase and more rare alkali feldspar and nepheline.

Monzodiorite (“rongstockite”)

The monzodiorite shows distinct compositional and textural variations including a magmatic rhythmic layering:

The equigranular types with hypidiomorphic prismatic textures are prominent in the intrusions. They are predominantly fine-grained passing into medium-grained varieties, which contain typical lath-shaped plagioclase (Fig. 16c, d of Appendix 2). Alkali feldspar is present in minor amounts as infilling between the plagioclase laths. Short prismatic clinopyroxene and poikilitic biotite laths occur in subordinate amounts. Sector zoning of pyroxenes is rare. Hornblende occurs only rarely as overgrowths on clinopyroxene. The central part of the Roztoky monzodiorite body (“rongstockite” of Tröger 1935; Fig. 2) shows an unusual magmatic rhythmic layering. The banded rock contains numerous feldspathic (plagioclase 94 vol%, K-feldspar 4 vol%) medium-grained 2–5-mm-wide bands, which alternate with prevailing pyroxene–biotite dioritic to mela-monzodioritic bands (mafic/felsic bands ~3:2 vol%). A singular lens of clinopyroxene diorite to gabbro with relicts of olivine was also found in the central part of a rhythmic layered section. At the margins of the body, this type grades into a leucocratic, mostly fine-grained variety represented by sample S-1.

The porphyritic types with transitions to medium-grained types are characterized by large phenocrysts (up to 10 mm lengths) of clinopyroxene and biotite enclosed in a granular groundmass. Hornblende is very rare. Plagioclase forms equigranular grains displaying a concentric zoning and it prevails over K-feldspar occurring in intergranular spaces.

Sodalite (analcime) syenite

The syenitic rocks occur mainly in the outer parts of the RIC, as well as in some isolated intrusions up to 5–10 km from the center. They are predominantly fine-grained syenites or high-level intrusive trachytic rocks. Fine- to medium-grained rocks are usually equigranular, rarely with a hypidiomorphic granular texture (Fig. 16e of Appendix 2). Spaces between laths of feldspars are filled by isometric sodalite grains. Sodalite and/or analcime also occur as euhedral crystals. Micro-porphyritic rocks with phenocrysts of both plagioclases and sodalite are rare. Sodalite is nearly completely replaced by analcime. The most abundant mafic mineral is columnar clinopyroxene with common sector zoning. Hornblende occurs either as rare microphenocrysts or more commonly as phenocrysts (up to 10 mm) from disaggregated hornblendite enclaves (Fig. 16f of Appendix 2). The enclaves of hornblendite (up to 40 cm in size) are medium- to coarse-grained (Fig. 16g of Appendix 2). In addition to dominant kaersutite, the enclaves contain minor amounts of clinopyroxene (diopside), altered olivine and plagioclase with accessory amounts of titanite and apatite. Miarolitic cavities in the syenite are often filled by zeolites and carbonates.

Sodalite-bearing monzosyenite to syenite

The sodalite-bearing monzodiorite forms one small intrusion at Býčkovice (Ulrych and Novák 1989). It is located about 15 km from the RIC (Fig. 1) and associated with a minor lamprophyre dike swarm. The rocks are mostly fine-grained, equigranular and have mineral characteristics similar to those of sodalite syenite (Fig. 16h of Appendix 2).

Dike rocks

Camptonite and monchiquite

Camptonite and monchiquite lamprophyre dikes (sensu Rock 1991 and Le Maitre 2002) are the most frequent dikes of the RIC area. Camptonite is composed of phenocrysts of kaersutite and phlogopite set in a fine-grained groundmass free of glass. Monchiquite is composed of phenocrysts of diopside and kaersutite in groundmass containing glass and/or analcime. It forms not only independent dikes but also “chilled margins” of camptonite dikes.

Maenite

Maenite covers local rock names “gauteite” and “bostonite” representing more acid medium derivatives (sensu Rock 1991), and maenites are often intimately associated with the camptonite–monchiquite dike suite (Rock 1991). Le Maitre (2002) supposes that maenite is a variety of trachyte and “bostonite” a leucocratic alkali feldspar syenite. Maenite contains phenocrysts of hastingsite, diopside, biotite, oligoclase, rarely also sodalite, in groundmass with rare glass. Transitional maenite types to camptonite as well as “bostonites” are from the RIC mentioned by Jelínek et al. (1989). “Bostonite” is composed mainly of oligoclase phenocrysts (rarely also diopside, hastingsite, biotite and sodalite) in fine-grained groundmass rich in both feldspars and sometimes also sodalite.

Tinguaite

Tinguaite and porphyric tinguaite dikes of phonolitic composition represent the most evolved rocks in the RIC. They were studied in detail by Ulrych et al. (2006).

Mineralogy and mineral chemistry

Clinopyroxene

Clinopyroxenes of all hypabyssal and dike rocks of the RIC occur as phenocrysts and/or grains in the matrix and correspond, following the classification of Morimoto et al. (1988), to aluminian ferroan diopside with about half of the grains being subsilicic and/or ferrian (Fig. 3). Selected analyses of clinopyroxenes are shown in Table 1. The ferroan pyroxenes (median Fe3+/Fetot ~0.2) were found in monzodiorites and sodalite monzosyenites, the ferric types (median Fe3+/Fetot ~0.5) occur in the most silica-deficient magmas like sodalite syenites and maenites, and in part also in camptonites; pyroxenes in essexites and monchiquites display an intermediate trend (median Fe3+/Fetot ~0.4–0.45; see Fig. 4). Among hypabyssal rock types, some general compositional trends were observed: the contents of Ca, Al, and Ti increase from monzodiorite to sodalite monzosyenite, and sodalite syenite to essexite; Si and Mg follows the opposite trend. As a consequence, the contents of quadrilateral pyroxene end-members vary as well (Fig. 3). Many clinopyroxene phenocrysts show normal and/or sector zoning; in some cases, patchy zoning was observed as well. The sector zoning is mainly present in the dike rocks (Ulrych 1983). Concentric zoning usually displays cores enriched in Mg and partly Si, whereas Al, Fe and Ti are depleted; Ca is constant across the entire grains.

Fig. 3
figure 3

Chemical composition and classification of pyroxenes of the hypabyssal and dike rocks of the Roztoky Intrusive Complex. The criteria of Morimoto et al. (1988) classify the pyroxenes based on Q–J diagram as Ca-pyroxenes. Except for two pyroxene grains, all compositions correspond to diopside. Gray field in the En–Fs–Di–Hd diagram at top left is zoomed into show the compositional variability of pyroxenes in detail in three panels at the bottom. Note the large spread in the composition of pyroxenes in sodalite (monzo)syenites, whereas those in essexites are relatively uniform

Full size image
Table 1 Representative chemical analyses (in wt%) of clinopyroxene in hypabyssal and dike rocks of the Roztoky Intrusive Complex
Full size table
Fig. 4
figure 4

Box plots showing the composition (in apfu) of pyroxenes in the hypabyssal and dike rocks of the Roztoky Intrusive Complex. Median values for Ca, Al and Ti contents increase from monzodiorite to sodalite monzosyenite, and sodalite syenite to essexite, whereas Si and Mg follows the opposite trend. SMS sodalite monzosyenite, SS sodalite syenite, MD monzodiorite, E essexite, C camptonite, M monchiquite, G “gauteite,” H hornblendite

Full size image

Temperatures and pressures of pyroxene crystallization were calculated from their chemical composition and the chemical composition of the host rocks following Putirka (2008). Unfortunately, no reasonable estimate for composition of melt being at the equilibrium with crystallizing pyroxene has been available, so the numerical values can be used only for comparison purpose. The highest pressure and temperature have been calculated for pyroxenes in hornblendite enclaves. Slightly lower pT conditions can be assumed for hypabyssal rocks, and among them, the highest pressures and temperatures were calculated for essexite S-25. Pyroxenes in the dike rocks indicate even slightly lower temperatures and pressures of crystallization.

Amphibole

Amphiboles are abundant in lamprophyres and in sodalite syenite, where they occur as phenocrysts, groundmass and probably also xenocrysts from the disaggregated cumulates. They also occur in minor amounts in essexites and monzodiorites. According to the classification of Leake et al. (1997), amphiboles belong to the calcic series and predominantly correspond to potassian kaersutite or ferrokaersutite though a few analyses fall in the potassian titanian pargasite or ferropargasite (Fig. 5). Kaersutites in the matrix of monchiquites are typically enriched in TiO2 (up to 7.5 wt%). Contents of F vary between 0.5 and 2.1 wt%, with Cl contents always below the EMP detection limit. Representative analyses of amphiboles are summarized in Table 2.

Fig. 5
figure 5

Chemical composition and classification of amphiboles of the hypabyssal and dike rocks of the Roztoky Intrusive Complex adopting the scheme of Leake et al. (1997). The data correspond mostly to kaersutite though a few analyses represent ferrokaersutite, pargasite or ferropargasite

Full size image
Table 2 Representative chemical analyses (in wt%) of amphibole in hypabyssal and dike rocks of the Roztoky Intrusive Complex
Full size table

The chemical composition of amphiboles allowed the estimation of pressures and temperatures of their formation. Temperatures were calculated applying semi-empirical thermometer of Otten (1984), and they vary between ~950 and 1,100 °C. The geobarometer of Schmidt (1992) yielded pressure values ranging from ~6 to ~8.5 kb.

Mica

Following IMA classification (Rieder et al. 1998), most micas correspond to ferroan phlogopite with few exception of magnesian–siderophyllite. In the diagram of Tischendorf et al. (2004, 2007), the majority of analyses fall into the field of biotite. A typical feature of all analyzed micas is elevated content of titanium; up to 10.5 wt% TiO2 (~1.25 apfu in formulae based on 22 cations) was found in essexites (Table 3). Titanium enrichment is achieved mostly via [6]R2+ + 2(OH)− → [6]Ti4+ + 2Ο2− and [6]R2+ + 2[4]Si4+ → [6]Ti4+ + 2[4]Al3+ substitutions though other substitution vectors occur as well (Fig. 6). The temperatures of biotite crystallization, based on the thermometer of Henry et al. (2005), vary between 750 and 800 °C. Though the calculated values for temperature are semi-quantitative due to calibration of the geothermometer used, the overall trend is well-defined: the lowest temperatures were observed for monzodiorites, intermediate for sodalite syenites and the highest yet quite scattered for essexites.

Table 3 Representative chemical analyses (in wt%) of mica in hypabyssal and dike rocks of the Roztoky Intrusive Complex
Full size table
Fig. 6
figure 6

Substitution vectors describing the composition of the micas found in the hypabyssal and dike rocks of the Roztoky Intrusive Complex

Full size image

Feldspar

In hypabyssal rocks, plagioclase is the major feldspar only in monzodiorite and essexite. It is distinctly zoned with An26–66 in monzodiorite and essexite, and An2–30 in sodalite syenite. Plagioclases in the dikes are compositionally variable with An19–65 in mafic types of maenite (“gauteite”) and An33–44 in more felsic types (“bostonite”). Alkali feldspar occurs in monzodiorite and essexite both as phenocrysts and dispersed grains in matrix. Phenocrysts in monzodiorite range in composition between Or55Ab38 and Or67Ab29. Those in essexite are, compared to previous, enriched in potassium (Or78Ab19 to Or65Ab33). In a matrix, contrary to phenocrysts, feldspars correspond to solid solution varying from Or83Ab16 to Or52Ab44. In sodalite syenite, the chemical composition of alkali feldspar varies between Or1Ab97 and Or56Ab41. Hornblendite contains both plagioclase (An40–50) and alkali feldspar (Or45Ab50). Compositional variability of studied feldspars is shown in Fig. 7 and Table 4.

Fig. 7
figure 7

Chemical composition of feldspars of the hypabyssal and dike rocks of the Roztoky Intrusive Complex

Full size image
Table 4 Representative chemical analyses (in wt%) of feldspar in hypabyssal and dike rocks of the Roztoky Intrusive Complex
Full size table

Using the calibrations of Putirka (2008), temperatures of crystallization were calculated from coexisting feldspar pairs: for essexites, the values range in a broad interval from ~900 to ~1,200 °C; for monzodiorites, temperatures vary between ~900 and ~1,000 °C; and for sodalite syenites, the procedure yields temperatures of ~1,000 and ~1,100 °C. Crystallization temperatures were further estimated from feldspar–liquid pairs; liquid was approximated as a respective host rock bulk chemical composition (Putirka 2008). Resulting temperatures are as follows: essexite 1,050–1,180 and 1,270–1,340 °C (sample S-25); monzodiorite 1,060–1,100 and 1,250 °C; sodalite syenite 1,020 and 1,270–1300 °C; hornblendite 1,130–1,280 °C; camptonite 1,030 °C; maenite 1,080–1,100 °C; monchiquite 1,010–1,050 °C. Though the temperatures partially overlap, feldspars of dike rocks crystallized at generally lower temperatures.

Foid and zeolite

Nepheline is a common constituent of essexite and a minor phase in sodalite syenite. In tinguaites, it forms reddish orange phenocrysts and cloudy grains in matrix (with hauyne and sodalite; Ulrych et al. 2006). Nepheline of essexites, sodalite syenites as well as tinguaites corresponds to a Si-poor type of solid solution in the Ne–Ks–Qz–H2O system at 700 °C and 1 kbar PH2O (Wilkinson and Hensel 1994). Sodalite was probably the dominant primary mineral in the sodalite syenite but was almost completely replaced by analcime, which in addition also forms euhedral crystals in miarolitic cavities. In essexite, sodalite is rare. Natrolite is a common mineral of the matrix in sodalite syenite; it is formed by the decomposition of sodalite.

Carbonate

There are three groups of carbonates in the monzodiorite intrusion at Roztoky partly differing in chemical composition and in particular carbon and oxygen isotopic composition: (1) disputed primary calcite, found only in carbonatized monzodioritic xenolith in trachytic breccia by Kopecký (1987a, b), (2) hydrothermal rhodochrosite dolomite and calcite related to epithermal tertiary Pb–Zn–Cu (Ag,Te) mineralization present in veins penetrating the monzodiorite (Pivec et al. 1998) interpreted by Kopecký (1987a, b) as products of residual solutions derived from a supposed carbonatite intrusion (late magmatic carbonatites C4 of Le Bas 1977), and (3) secondary chemically pure calcite in fissures associated with the alteration of the parental rock.

Kopecký (1987a, b) published isotopic data of the carbonate sample set. From this set, only one analysis of dolomitic carbonate (SrO ~1.0 wt%) of carbonatitized monzodiorite plots within the field of primary igneous carbonatites (PIC) as defined by Taylor et al. (1967). All other samples possess higher δ13C and δ18O values, plotting in the field of “carbonatite-associated carbonates,” which generally do not represent primary igneous carbonate. The isotopic data of carbonates (Pivec et al. 1998) from the vein with epithermal mineralization in Roztoky show similar δ13C but lower δ18O values when compared with data reported by Kopecký (1987a, b). If a temperature of the Roztoky deposition between 200 and 280 °C (Pivec et al. 1998) for the hydrothermal vein carbonates is accepted, δ13Cfluid values from ca. −3.0 to −5.0 ‰ and δ18Ofluid values from ca. −3 to −7 ‰ (SMOW) can be calculated. The hydrothermal carbonates were formed during relative shallow hydrothermal circulation of low to medium salinity, and low δ18O fluids supported by CO2 influx of deep-seated origin. These carbonates are characterized by moderate REE contents (100–130 ppm), La/Yb ratios (49–63) and SrO (0.12–0.15 wt%) and Ba (0.22–0.88 wt%) contents and by 87Sr/86Sr ratios (rhodochrosite 0.7052; calcite and dolomite 0.7051; cf. parental monzodiorite (~0.7044) reflecting some crustal contamination). Data of δ13C indicate probably minor admixture of sedimentary C in carbon of deep-seated origin, while the O isotope data point to participation of a δ18O fluid of meteoric derivation (Ulrych et al. 1997).

Fe–Ti oxide

The ternary plot Ti–Fe2+–Fe3+ (Fig. 8a) shows two major series, the ilmenite–hematite (rhombohedral s.s.) and magnetite–ulvöspinel (spinel s.s.; Buddington and Lindsley 1964). The position of the latter lies markedly off the ideal Usp-Mt line; this reflects the substantial isomorphic substitution of other spinel group end-members.

Fig. 8
figure 8

Chemical composition of Fe–Ti oxide minerals of the hypabyssal and dike rocks of the Roztoky Intrusive Complex (a). Ternary diagram ilmenite–geikielite–pyrophanite (b) displays the compositional variation of ilmenites. A modification of Haggerty’s spinel prism (c, d) is used to present the composition of spinel group minerals. Dike rocks display notably higher Mg content compared to hypabyssal ones. Ternary diagram magnetite–hercynite–ulvöspinel (e) illustrates a wide variation in the Usp content among titanium-rich magnetites

Full size image

Three types of ilmenite were recognized (Fig. 8b): a pure ilmenite in monzodiorite and “bostonite,” an ilmenite with low pyrophanite (MnTiO3) content, hematite end-members in essexite, monzodiorite and sodalite syenite, and a Mg-bearing ilmenite containing geikielite (MgTiO3) and hematite end-members in essexite, maenite, monchiquite and hornblendite cumulates.

Magnetite occurs as (Fig. 8c–e): (1) pure magnetite associated with the decomposition of mafic minerals, (2) magnetite with medium contents of ulvöspinel end-member (Fe2TiO4) and low in magnesioferrite (MgFe2O4) and jacobsite (MnFe2O4) end-members in monzodiorite and essexite, (3) Ti-magnetite rich in the ulvöspinel end-member together with medium contents of magnesioferrite and jacobsite end-members in monzodiorite, essexite and sodalite syenite, (4) Ti-magnetite very rich in ulvöspinel and magnesioferrite end-members in monchiquite, camptonite and maenite. As illustrated in Fig. 8e, however, a few exceptions exist from this general trend. Ti-magnetite with a higher content of ulvöspinel and magnesioferrite end-members occurs in monchiquite.

Accessory minerals

Apatite compositionally corresponds to fluor-hydroxylapatite in monzodiorite, hornblendite and maenite, and to fluorapatite in sodalite syenite, essexite, camptonite and monchiquite. Its CO2, SiO2 and REE contents (Table 5) resemble the composition of apatites of carbonatite suites (Sommerauer and Katz-Lehnert 1985). Titanite forms at the expense of Ti-rich biotite in essexite. It commonly occurs also in hornblendite cumulates where it has high contents of REE (4,500–11,700 ppm).

Table 5 Representative chemical analyses (in wt% and ppm, respectively) of apatite in hypabyssal and dike rocks of the Roztoky Intrusive Complex
Full size table

Geochemistry of the hypabyssal and dike rocks of the Roztoky Intrusive Complex

Major and trace elements

Major and trace element concentrations of the main types rocks are listed in Table 6, and the data are plotted in the diagram of Cox et al. (1979) in Fig. 9. The samples show a range in SiO2 from 41 to 53 wt% and plot at high Na2O and K2O in the alkaline field; their Mg values fluctuate from 35.1 to 61.3 and alkali index A.I. (Na2O + K2O/Al2O3 mol%) 1.21–2.09:

Table 6 Major and trace element concentrations (in wt% and ppm, respectively) of rocks of the Roztoky Intrusive Complex
Full size table
Fig. 9
figure 9

Composition of hypabyssal and dike rocks of the Roztoky Intrusive Complex plotted to the classification diagram of Cox et al. (1979). Dark gray and light gray fields represent the composition of basanitic and trachybasaltic volcanic formations, respectively, from the České středhoří Mts. (Ulrych et al. 2002). These fields are shown to demonstrate the chemical affinity of studied RIC rocks to these formations

Full size image

(1) hypabyssal mildly alkaline (tephritic) series (essexite–monzodiorite–sodalite syenite) and hornblendite cumulate in sodalite syenite; (2) mildly alkaline dike series (MAS) (camptonite/monchiquite–maenite); and (3) strongly alkaline dike series (SAS) represented by tinguaite (Ulrych et al. 2006).

The major element variation with respect to MgO content is illustrated in Fig. 10. Positive correlation of TiO2, FeO, CaO and negative correlations of SiO2, Al2O3 and Na2O with MgO suggest variable degrees of fractional crystallization as the main evolutionary process. Also, positive correlation exists between MgO and some compatible elements (e.g., Sc, Co; not shown).

Fig. 10
figure 10

Major element variation diagrams for hypabyssal and dike rocks of the Roztoky Intrusive Complex. Trends (arrows) of possible fractional phases from Guo et al. (2004)

Full size image

The trace element variation in the hypabyssal intrusive rock series essexite–monzodiorite–sodalite syenite includes progressive enrichment in incompatible elements (e.g., rare earth elements—REE, large ion lithophile elements—LILE) and depletion of compatible elements (e.g., Cr, Co, Ni, Sc) and slight increase in light REE (LREE) relative to heavy REE (HREE).

Primitive mantle-normalized REE patterns (Fig. 11) of the hypabyssal rocks show a general enrichment in LREE with (La/Yb)N ratios varying from ~25 for monzodiorite to ~33 for sodalite syenite; hornblendite has the lowest LREE/HREE ratio of ~10. The REE patterns of lamprophyre dikes (Fig. 11) are similar to those of the hypabyssal rocks with the (La/Yb)N varying in the range of 18–22. Almost all the rocks exhibit a small negative Eu anomaly (Eu/Eu* = 0.9–1.0). Primitive mantle-normalized incompatible element diagrams (Fig. 12) display a progressive enrichment from incompatible elements in monzodiorite to sodalite syenite. The slight depletion of P is typical of mafic rocks (essexite, monzodiorite), whereas a strong negative anomaly is typical of felsic types (e.g., sodalite syenite). When compared to the patterns of hypabyssal rocks, the hornblendite cumulate has substantially lower contents of LREE (Fig. 11) and other incompatible elements (Fig. 12), while it is enriched in Ti consistent with the accumulation of kaersutite/titanian magnetite. Among the hypabyssal rocks, the hornblendites have the highest K/Rb ratio (~635); other rocks display a broad variation between ~230 and ~480.

Fig. 11
figure 11

Primitive mantle-normalized rare earth element (REE) patterns of hypabyssal and dike rocks of the Roztoky Intrusive Complex. Normalizing values from McDonough and Sun (1995). E essexite, MD monzodiorite, LMD leuco-monzodiorite, SS sodalite syenite, SMS monzosyeníte, H hornblendite

Full size image
Fig. 12
figure 12

Primitive mantle-normalized incompatible trace element patterns of hypabyssal and dike rocks of the Roztoky Intrusive Complex. Normalizing values from McDonough and Sun (1995). ES essexite, MD monzodiorite, LMD leuco-monzodiorite, SS sodalite syenite, SMS monzosyenite, H hornblendite

Full size image

The lamprophyre dikes have similar concentrations of incompatible elements compared to the hypabyssal rock series, except Rb which is significantly enriched (Fig. 12). The trace element patterns show depletions in Cs, Rb, P, Ti and enrichment in Zr and Ba relative to elements with similar compatibility. Enrichment in incompatible elements and depletion in compatible elements are even more pronounced in felsic rock types. Their patterns also show depletion in P and Ti and enrichment in Nb and Zr.

Most of the hypabyssal rocks and associated dikes are distinctly differentiated as indicated by their low MgO contents and Mg#. However, those with MgO contents over 7 wt% and/or Mg# >56 show PM-normalized trace element patterns consistent with primitive volcanic rocks of the OR and resemble those of HIMU-OIB with enrichment in high-field-strength elements (HFSE) and frequently a negative K anomaly.

The Zr/Hf ratios of the RIC rocks vary in a range from of 33 to 55 with the exception of the hornblendite cumulate (~26). The lower ratios in felsic types might be due to the late-magmatic (or post-magmatic) fluids producing Zr minerals. Various Zr-bearing minerals were identified at other localities in the region by Ulrych et al. (1992).

87Sr/86Sr and 143Nd/144Nd isotopes

Initial 87Sr/86Sr ratios have been recalculated to 30 Ma (Table 7). Hypabyssal rocks range from 0.70375 (sodalite syenite) to 0.70503 (sodalite monzosyenite). The 143Nd/144Nd ratios show a large variation and initial εNd values ranging from +0.7 to +3.4. Except for the sodalite monzosyenite samples, the samples plot at the lower end of the field for 42–16-Ma syn-rift volcanic rocks of Ulrych et al. (2011) and are in some cases less radiogenic in their Nd isotopic composition (Fig. 13). However, there are distinct differences among similar rock types. For example, sodalite syenites show a large variation in εNd from +1.3 to +3.3. In contrast, sodalite monzosyenites exhibit very homogeneous Sr–Nd isotopic compositions, but strong decoupling between 87Sr/86Sr and 143Nd/144Nd ratios in that their Sr isotopic composition departs from the mantle array (Fig. 13). Lamprophyric dikes have Sr–Nd isotopic compositions overlapping those of hypabyssal rocks.

Table 7 Rb–Sr and Sm–Nd isotope data for hypabyssal and dike rocks of the Roztoky Intrusive Complex
Full size table
Fig. 13
figure 13

Initial 87Sr/86Sr and 143Nd/144Nd isotope ratios for ca. 30-Ma-old hypabyssal and dike rocks of the Roztoky Intrusive Complex. The gray field of 42–16-Ma syn-rift volcanic rocks uses data of Cenozoic volcanic rocks from Ulrych et al. (2011), Lustrino and Wilson (2007) and Ulrych (unpublished data). BSE Bulk Silicate Earth. Symbols as in Fig. 10

Full size image

Discussion

The RIC suite belongs to the volcanic series of the České středohoří Mts., which represents an integral part of the ECRIS. The RIC hypabyssal rock series (essexite, monzonite, sodalite syenite, questionable carbonatites) accompanied by dike rock suite (camptonite–tinguaite) and other intrusions as olivine nephelinite, tephrite/basanite, phonolite have the only analogue within the ECRIS: similar rock series of Kaiserstuhl in the Rhine Graben, Germany (Wimmenauer 1974; Ulianov et al. 2007). Compositionally, the extrusive volcanic rocks of the České středohoří Mts. resemble those of the Massif Central in France (Wilson et al. 1995) and the German Volcanic Province (Lustrino and Wilson 2007). Other partly analogous rift-related rock associations in Europe are known from Oslo Graben in Norway (Dons and Larsen 1978) and Serra de Monchique, Portugal (Rock 1978). The RIC rock suite shows close textural as well as chemical similarities to that of the Monteregian Province rift (St. Lawrence rift system), Quebec, Canada. The origin of all these alkaline rock suites is usually explained by fractional crystallization of a mantle-derived magma (Eby 1984, 1985a, b, 1987).

The hypabyssal rock suites of the RIC formed by mildly alkaline essexite–monzodiorite–sodalite syenite series and dikes, which appear to be closely genetically related. This is indicated by: (1) Radial orientation of a dike swarm around the core composed of hypabyssal intrusions of monzodiorite, essexite and partly also sodalite syenite (Hibsch 1926; Shrbený 1969; Ulrych et al. 1983; Kopecký 1978); (2) The comparable ages of hypabyssal rock series (33–28 Ma) and dike rocks (30–25 Ma; Ulrych and Balogh 2000); (3) Covariations of major and trace element contents of the hypabyssal intrusions and dikes (Ulrych et al. 1983; Ulrych 1998; Jelínek et al. 1989); and (4) Overlap of the Sr–Nd isotope composition of the hypabyssal rocks and dikes.

Tinguaites representing the singular member of the strongly alkaline series were studied in detail by Ulrych et al. (2006). The rocks are rich in alkali and other incompatible elements and have high contents of Cl and SO3 concentrated in sodalite and nosean. The large variation of 87Sr/86Sr30 from 0.70335 to 0.70419 and εNd30Ma values of +2.1 to +4.3, indicating a moderately depleted mantle source, are comparable to that of other dike rocks of the RIC. The chemically evolved composition of tinguaites was attained via melt differentiation at shallow crustal levels with limited assimilation of crust. Their K–Ar ages of 31–28.5 Ma and chemical geochemical characteristics support their synchronous origin with the rocks of the RIC.

The intrusions of RIC are texturally and compositionally heterogeneous. The significant heterogeneity of individual intrusions is also demonstrated by alternation in bands of various grain sizes, texture (porphyritic vs. equigranular), contents of mafic minerals (leucocratic vs. mesocratic) and clinopyroxene/biotite ratio. These variations are probably due to flowage differentiation during the emplacement of the intrusion at a deeper crustal level. The steeply dipping magmatic layering is indicative of rocks crystallizing in extensional rift conditions. Bhattacharji (1966), Bhattacharji and Nehru (1972), Philpotts (1974) and Eby (1984, 1985a, b) suggest that such inhomogeneity (vertically dipping rhythmic layering) and flowage differentiation of hypabyssal rocks are characteristic form of their emplacement under the riftogenic conditions, e.g., in the Monteregian Hills, Canada).

The variations in major elements, petrography and mineral chemistry indicate that the RIC suite, including hypabyssal rocks of intrusions and both dike series, may be derived from parent magmas of alkali basalt affinities by fractional crystallization. For RIC suite, whole-rock TiO2, CaO, FeO, Ni and Co contents show good positive correlation with MgO, while SiO2, Al2O3, Na2O, Ba and Sr are negatively correlated with the same oxide (Figs. 10, 14). All these trends could be explained by the fractionation of clinopyroxene, plagioclase, olivine and apatite (e.g., Guo et al. 2004). The presence of kaersutite + diopside ± olivine, fluorapatite, labradorite, titanite cumulates (e.g., sample S-6) in sodalite syenites, and some lamprophyres partly support this model. However, fractionation of amphibole (kaersutite) from the parental melt would lead to decrease in Ba, Sr or MREE (medium rare earth elements) with decreasing Mg content, which is not the case of RIC suite (Fig. 14).

Fig. 14
figure 14

MgO (wt%) versus selected trace element (ppm) variation diagrams for hypabyssal and dike rocks of the Roztoky Intrusive Complex

Full size image

However, fractional crystallization processes alone cannot explain all compositional variations within the RIC suite, particularly the variations in trace elements and Sr–Nd isotopes. For example, the high contents of several incompatible trace elements (e.g., LREE, Th, Ba) in the sodalite syenites suggest that it is unlikely these rocks are simply products of continuous fractional crystallization of alkaline basaltic magma and may point to assimilation of crustal material. To provide more constraints on the role of crustal contamination, the importance of fractional crystallization (FC), assimilation–fractional crystallization (AFC) and mixing processes was examined. For FC process, the Rayleigh fractionation model was used (Allègre et al. 1977), the AFC model was tested by a procedure from DePaolo (1981), and for mixing of two components (parent magma and continental crust), a model of Powell (1984) was used. To test these models, elements with different degrees of compatibility during magma fractionation were selected. Cobalt is assumed as a typical compatible element (note its positive correlation with MgO; Fig. 14), and La is assumed as a typical incompatible element. The results were discussed in combination with Sr–Nd isotopic composition of the studied rocks. The essexite sample S-25 is assumed to be the composition closest to the expected parental magma on the basis of the highest temperature and pressure estimate, the highest MgO and the lowest SiO2 contents. The composition of bulk continental crust (Rudnick and Gao 2003) was chosen as hypothetical assimilant. The mass ratio of assimilant/mass fractionated used in our AFC model is 0.7. The modeled FC, AFC and mixing paths constructed using FC–AFC–FCA Mixing modeler (Ersoy and Helvaci 2010) are shown in Fig. 15. Most of the hypabyssal rocks and lamprophyres dikes of the RIC suite plot along FC and AFC calculated curves (Fig. 15), and it is clear that the Co–La variation in RIC suite cannot be explained by simple mixing of two different components (e.g., S-25 and bulk continental crust). The essexite, monzodiorite and most of the lamprophyre dikes (except “bostonite” S-32 and camptonite S-43) can be accounted for by ~8–45 % of suggested parent magma crystallization without any contribution of crustal material. In contrast, monzosyenite and leuco-monzodiorite composition with variable Sr isotopic composition can be explained either by FC process with ~50 % of parental magma crystallization or by AFC process with lower ~35–40 % crystallization of parental magma. However, the latter is not supported with Nd isotopic data of these rocks overlapping with essexite and most of the lamprophyre dikes. The Co–La variation in sodalite syenites may suggests significant contribution of crustal material to their magmas, but this is not consistent with highly variable Sr–Nd isotopic composition with much of the sodalite syenite characterized by the highest εNd among all studied rocks (Table 7; Fig. 13). Therefore, the fractional crystallization and/or assimilation–fractional crystallization was likely accompanied by other processes such as the late-magmatic transfer of volatile fluids containing Zr, U, Th and REE as implied by the occurrence of rare minerals, which concentrate these elements (Zr minerals) in the RIC concentrated in evolved differentiates as sodalite syenite and tinguaites (Ulrych et al. 1992).

Fig. 15
figure 15

Cobalt versus lanthanum diagram for samples of the Roztoky Intrusive Complex. Shown are calculated data trends for fractional crystallization (FC; solid gray line), assimilation–fractional crystallization (AFC; solid black line), and mixing of parental melt with assimilant (dashed line). As a parental magma composition, essexite sample S-25 (Mg#56) was chosen. Bulk continental crust (Rudnick and Gao 2003) represents an assimilant (indicated as asterisk). For details see the text

Full size image

The Sr–Nd isotopic compositions (Fig. 13) provide further constraints on the evolution of the RIC suite. Except monzosyenites with highly radiogenic 87Sr/86Sr, all rocks fall in the depleted mantle field, and most of them plot in the least depleted portion of a data field for syn-rift Cenozoic volcanic rocks of the Bohemian Massif (Ulrych et al. 2011). However, large variations in the Sr–Nd isotopic compositions of individual rock types in spite of their similar petrography and whole-rock chemistry suggest variable crustal contributions and/or a heterogeneous mantle source. However, the former is not consistent with trace element AFC model employed above, in spite of the presence of crustal xenoliths (carboniferous and upper cretaceous sediments, Saxothuringian gneisses and altered granites, Ulrych et al. 2000) in the RIC hypabyssal rocks and dikes. In contrast, the latter is likely as 87Sr/86Sr ratios of 0.70314–0.70399 in mantle xenoliths from Cenozoic volcanic rocks (Blusztajn and Shimizu 1994; Ackerman et al. 2007; Ackerman, unpubl. data), and the presence of metasomatized mantle xenoliths in the volcanic rocks of the České středohoří Mts. Complex (Ackerman et al. 2013) may indicate heterogeneities in the subcontinental mantle. However, the RIC rocks with high 87Sr/86Sr (e.g., monzosyenite) may also have been modified by the late-stage magmatic fluids associated with evolution of the RIC.

Conclusions

Petrographic characteristics, major and trace element composition, as well as Sr–Nd isotopic data on monzodiorite–essexite–sodalite syenite suites and the associated dike swarms of the RIC suggest that they are genetically linked. The magmatic rhythmic layering of the RIC intrusions defined by the alternation in bands with different textures, different proportions of felsic to mafic minerals and different clinopyroxene/biotite ratios is probably due to flowage differentiation of the magma at deeper crustal levels. Trace element modeling shows that fractional crystallization (FC) and assimilation–fractional crystallization (AFC) processes could have played an important role in the evolution of the RIC rocks. The essexite, monzodiorite, monzosyenite, leuco-monzodiorite and most of the lamprophyre camptonite, monchiquite and maenite dikes can be best explained by ~8–50 % crystallization of essexite parent magma (the most depleted sample of essexite in terms of trace element composition) without no or limited contribution of crustal material. On the other hand, trace element composition of sodalite syenite is not consistent either with FC or AFC models, suggesting that their composition was most likely affected by the late-magmatic fluids. Large variations in the Sr–Nd isotopic compositions of the individual RIC rock cannot be explained by variable crustal contributions to the RIC parental melts, but rather reflect heterogeneous mantle source.