2. Geological setting

The basement of the Balkan terrane comprises a Neoproterozoic ophiolite and a Cambro-Ordovician island arc association (Haydoutov, Reference Haydoutov1989; Savov et al. Reference Savov, Ryan, Haydoutov and Schijf2001), unconformably overlain by a Palaeozoic sedimentary sequence, which is intruded by a post-Variscan calc-alkalic volcanic suite (Cortesogno et al. Reference Cortesogno, Gaggero, Ronchi and Yanev2004).

On the southern margin of the Balkan terrane (Svoge region), mafic plutons have intruded at a relatively shallow depth (andalusite + biotite in the contact aureole) within the Middle–Upper Ordovician grey shales and sandstones of the Grohoten Formation (Vladykin, Grozdanov & Bonev, Reference Vladykin, Grozdanov and Bonev2001; Gutiérrez-Marco et al. Reference Gutiérrez-Marco, Yanev, Sachancki, Rábano and Lakova2003).

Polyphase plugs and dykes of syenite composition cut the plutons. A mafic pluton and related syenitic to quartz syenitic intrusions extending for about 2.5 km² have been studied in detail for the Svidnya river valley (Fig. 1b).

3. Petrography

3.a. The mafic pluton

The Svidnya main intrusion, described as shonkinite or lamproite in the regional literature (Grozdanov, Reference Grozdanov1965; Stefanova, Reference Stefanova1966; Vladykin, Grozdanov & Bonev, Reference Vladykin, Grozdanov and Bonev2001), shows essentially two compositions: (1) diopside–sanidine–phlogopite lamproite (formerly orendite) according to the IUGS nomenclature (Woolley et al. Reference Woolley, Bergman, Edgar, Le Bas, Mitchell, Rock and Scott Smith1996), characterized by lack of primary plagioclase, and (2) aegirine, Na–Ca and Na-amphibole, potassic and sodic feldspar melasyenite.

As a rule, both compositions show medium- to coarse-grained (1–10 mm) phaneritic texture and fine- to medium-grained (0.5–1 mm) porphyritic texture. Flow textures can develop, but generally the large K-feldspar phenocrysts do not exhibit a preferred orientation. Locally, the early-crystallized mafic phases develop cumulus features (Figs 2, 3a; a colour version of Fig. 3 is available in online Appendix at http://www.cambridge.org/journals/geo).

Figure 2. Intrusion relationships of lamproites, syenites and country rock at Svidnya. The present setting of contact surfaces is at high, nearly vertical dip. The field of view is about 5.85 m wide. A colour version of this figure is available at http://www.cambridge.org/journals/geo.

Figure 3. Microphotographs of Svidnya lamproites. Plane polarized light. Scale bar in the photographs. (a) Cumulus texture of clinopyroxene and biotite with K-feldspar as intercumulus phase. (b) Phaneritic medium-grained texture of clinopyroxene, biotite and K-feldspar. (c) Subhedral biotite, poikilitic on apatite and pyroxene, mantled by green amphibole aggregates. (d) Biotite coronas surrounding amphibole pseudomorphs after olivine. For a colour version of this figure see online Appendix at http://journals.cambridge.org/geo.

The lamproite is characterized by early-crystallized Fe-rich K-feldspar, diopsidic augite and high-Ti biotite (Fig. 3b). K-feldspar exhibits growth zoning and synneusis from a light-coloured core. Diopsidic augites occasionally show overgrowths of, or partial replacement by, richterite, winchite and actinolite. Biotite, frequently including pyroxene and apatite, is intergrown with and mantled by subhedral richterite (Fig. 3c). Sometimes biotite shows deformed reddish cores and undeformed orange-reddish rims with oxide inclusions. A few lamproites, lacking diopsidic augite, are instead characterized by amphibole and biotite pseudomorphs after olivine (Fig. 3d). Biotite coronas around these pseudomorphs suggest that decompression occurred during cooling. Very fine-grained mosaic biotite and K-feldspar crystallize in rounded aggregates up to 0.5 cm in olivine-bearing lamproites. Dark brown biotites, rimmed by reddish biotite, develop inside the patches (Fig. 4b, Table 3). Within amphibole and biotite, early-crystallized apatite, with pinkish cores and evident growth zoning, is abundant, and late-precipitated apatite is generally colourless. Zircon, titanite, Fe-oxides and scarce galena are accessory phases. The amphibole core is occasionally altered to calcite or, more rarely, barite and Th-, LREE-phosphates.

Figure 4. (a) Ae (NaFe³⁺Si₂O₆) – Di (CaMgSi₂O₆) – Hd (CaFe²⁺Si₂O₆) and Wo–En–Fs diagram for clinopyroxenes from lamproites and syenites. (b) Al₂O₃ v. TiO₂ diagram for phlogopites. For comparison in (b), compositional fields of brown-micas from minettes, Roman Province lavas (RPT-lavas) and lamproites are reported. Data from Mitchell & Bergman (Reference Mitchell and Bergman1991). Abbreviations: LH-Al – aluminous phlogopite in inclusions from Leucite Hills; LH-GM – groundmass/madupitic micas from Leucite Hills; LH-P – phenocrysts from Leucite Hills; M-MA – madupitic lamproite Murcia Almeria; M-WK – madupitic lamproite West Kimberley; PL-MA – phlogopite lamproite Murcia Almeria; PL-WK – phlogopite lamproite West Kimberley.

Melasyenites are found in the main intrusion and also as thin porphyritic dykes. In melasyenites, the abundant albite exhibits silicate and clouds of oxide microinclusions, whereas the potassic feldspar shows perthite exsolutions. Clinopyroxenes are zoned with aegirine–augite cores grading to aegirine rims. Na–Ca and finally Na-amphiboles crystallize cotectically with K-feldspar and aegirine. In turn, poikilitic biotite with included crystals of K-feldspar and richterite is partly intergrown with Na-amphibole. Amphibole also occurs as radiating clusters, characterized by needle-shaped ferri-eckermannite at the core, likely replacing biotite or clinopyroxene, and secondary fibrous green richterite at the rim. Accessory phases are zircon, titanite, apatite and rare Ba- and Nb-titanosilicates.

3.b. Plugs and dykes

Syenites and quartz syenites are medium-coarse (1–6 mm), with heterogranular, hypidiomorphic granular to porphyritic textures (Fig. 5a, b). They range from melanocratic to leucocratic, due to cumulus processes (e.g. Fig. 5b–d), and from primitive to evolved compositions. The more primitive compositions are characterized by early cotectic crystallization of K-feldspar + Na–Ca amphibole, followed by the K-feldspar + Na-amphibole + aegirine assemblage in intermediate compositions, and finally by K-feldspar + aegirine in more evolved ones. Prismatic K-feldspar frequently forms radiating sprays, or includes apatite and aegirine. Quartz is a common interstitial phase.

Figure 5. Microphotographs of Svidnya syenites and quartz syenites. Plane polarized light. Scale bar in the photographs. (a) Porphyric alkali feldspar, biotite and aegirine phenocrysts; the groundmass is formed by acicular aegirine, amphibole and K-feldspar. (b) Leucocratic assemblages of K-feldspar and aegirine phenocrysts with interstitial Na-amphibole and acicular aegirine. (c) Orthocumulate texture characterized by euhedral to subhedral zoned amphibole, sometimes poikilitic, cumulus clinopyroxene and intercumulus quartz. (d) Orthocumulate texture characterized by subhedral amphibole, clinopyroxene and titanite as cumulus phases, quartz and K-feldspar as intercumulus phases. For a colour version of this figure see online Appendix at http://journals.cambridge.org/geo.

Melanocratic syenites are characterized by orthocumulus textures (Fig. 5c, d). The cumulus phases are dominantly zoned richterite (up to 80% in volume), overgrown along cleavages or partially replaced by eckermannite and ferri-glaucofane. Na-clinopyroxene, K-feldspar and apatite precipitate as cumulus phases, K-feldspar and quartz as intercumulus phases. Na-clinopyroxene is cotectic with or poikilitic on K-feldspar, amphibole and apatite. Moreover, Na-clinopyroxene occurs as corroded inclusions in amphibole and zoned K-feldspar rims. Rare biotite is found as inclusions in amphibole and K-feldspar.

In leucocratic samples, biotite crystallized as anhedral grains cotectically with alkali feldspar and as inclusions in amphibole. It was replaced to a variable extent by ilmenite + K-feldspar, probably during emplacement at shallow depths. Amphibole occurs as euhedral to acicular and markedly zoned grains. Alkali feldspar is generally inverted to microcline, with exsolved lamellae of hematite, as a result of slow cooling. Acicular apatite, highly variable in abundance, is found as inclusions in alkali feldspar and amphibole. Quartz is a widespread interstitial phase, sometimes occurring as patches. Accessory minerals are zircon, titanite, rutile, Ba- and Nb-titanosilicates.

4. Analytical methods

Quantitative electron microprobe analyses of mineral phases were acquired using a scanning electron microscope equipped with an X-ray dispersive analyser (EDAX PV9100), installed at the Department for the Study of Territory and its Resources, University of Genoa. Operating conditions were 15 kV accelerating voltage and 2.20 nA beam current. Reference standards for the elements (in brackets) were: jadeite (Na), forsterite (Mg), albite (Al), augite (Si, Ca), microcline (K), ilmenite (Ti), chromite (Cr), rhodonite (Mn) and fayalite (Fe). Other elements were below detection limits. The natural standards were analysed by WDS microprobe at Modena University. Na₂O and MgO contents analysed in silicates by means of an EDAX microprobe are generally underestimated if the analysis is processed with current automatic methods. To overcome this problem, the background for Na (1.040 keV) and Mg (1.252 keV) was manually corrected and considered to be between 0.9 and 4.2 keV. The estimated uncertainties are: 0.1 wt% for SiO₂, TiO₂, Al₂O₃, Cr₂O₃, FeO, MgO, MnO, P₂O₅; 0.05 wt% for CaO and K₂O; 0.25 wt% for Na₂O.

Feldspar analyses, on the basis of eight oxygens, were recalculated to total cations = 5. Clinopyroxene analyses were calculated according to the stoichiometric method of simultaneous normalization to 4 cations and 6 oxygens, and Fe³⁺ = (12 − total cation charge) was considered for clinopyroxene. The allocation of cations to sites T, M1 and M2 was performed according to Morimoto (Reference Morimoto1988). End-members were calculated in the sequence: wollastonite, enstatite, ferrosilite, aegirine, jadeite, CaAl₂SiO₆, CaFeAlSiO₆, CaCrAlSiO₆ and CaTiAl₂O₆. The nomenclature of Morimoto (Reference Morimoto1988) and Rock (Reference Rock1990) was adopted. The brown mica cation sum was normalized to 7 + Ti − (Na + K) on the basis of 11 oxygens. The amphibole cation sum was normalized to 13 − (Ca + Na + K), as suggested by Leake et al. (Reference Leake, Woolley, Arps, Birch, Gilbert, Grice, Hawthorne, Kato, Kisch, Krivovichev, Linthout, Laird, Mandarino, Maresch, Nickel, Rock, Schumacher, Smith, Stephenson, Ungaretti, Whittaker and Youzhi1997); Fe³⁺ = (46 − total cation charge); Fe²⁺ = (Fe_tot − Fe³⁺); Al^IV = (8 − Si); Al^VI = (Al_tot − Al^IV). The nomenclature of Leake et al. (1997, 2003), revised by Hawthorne & Oberti (Reference Hawthorne and Oberti2007), was adopted. Mineral abbreviations are after Kretz (Reference Kretz1983).

Whole-rock major and trace element abundances for Svidnya ultrapotassic rocks (21 samples) were measured by X-ray fluorescence spectrometry (XRF) at the X-RAL Laboratories (SGS Canada Inc.), Toronto, Canada. Losses on ignition (LOI) were determined by gravimetry. Rare earth elements (REE) were analysed by inductively coupled plasma-mass spectrometry (ICP-MS) at the X-RAL Labs.

Two ⁴⁰Ar–³⁹Ar age determinations were carried out on amphibole and biotite separates from the lamproite BL42. The separated fractions (95% amphibole and 5% biotite) were analysed by ⁴⁰Ar–³⁹Ar incremental heating at the Actlab Laboratories (Canada). The samples wrapped in Al foil were loaded in evacuated and sealed quartz vials with K and Ca salts and packets of LP-6 biotite interspersed with the samples to be used as a flux monitor. The samples were irradiated in a nuclear research reactor for 24 hours. The flux monitors were placed between every two samples, thereby allowing precise determination of the flux gradients within the tube. After the flux monitors were run, J values were calculated for each sample, using the measured flux gradient. The neutron gradient did not exceed 0.5% of sample size. LP-6 biotite has an assumed age of 128.1 Ma.

Sample dissolution and isotopic analysis were carried out at the Earth Science Department, University of Trieste. Samples were dissolved for isotopic analysis in Teflon^® vials using a mixture of HF–HNO₃ and HCl purified reagents. Sr and Nd were collected after ion exchange and reversed-phase chromatography, respectively; total blank for Sr was less than 20 pg. The Sr and Nd isotopic compositions were obtained using a VG 54E mass spectrometer and ‘Analyst’ software (Ludwig, Reference Ludwig1994) for data acquisition and reduction. The ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd were corrected for fractionation to ⁸⁶Sr/⁸⁸Sr = 0.1194 and ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219, respectively, and the measured ratios were corrected for instrumental bias to NBS 987 and La Jolla standard values of 0.71025 ± 0.00002 (n = 12) and 0.511860 ± 0.000021 (n = 9). The reported errors represent statistics at the 95% confidence level. Neodymium-model ages were calculated with respect to a depleted mantle evolution curve given by ϵNd(T) = 0.25T² − 3T + 8.5 (T in Ga) as reported in Ludwig (Reference Ludwig1994).

5. Mineral chemistry

5.b. Clinopyroxene

The early-crystallized clinopyroxene in lamproite (Fig. 4a; Table 2; Fig. A1 in online Appendix at http://journals.cambridge.org/geo) is Al- and Ti-poor (TiO₂ < 0.8 wt%) diopsidic-augite, a typical feature of clinopyroxene in lamproites (Conticelli, Reference Conticelli1998), with increasing aegirine solid solution towards the rim (Ae₅). Fine-grained groundmass crystals are augite to aegirine-augite. Corroded clinopyroxene, overgrown by amphibole, has a higher aegirine content (up to 24 mol.%). Diopsidic-augite (Ae_5–14) to aegirine (up to Ae₇₆) precipitates in melasyenites.

Table 2. Selected electron microprobe analyses of clinopyroxenes

In the monzodiorite, pyroxenes are diopsides with relatively high-Al cores (Fig. 4a). Early-crystallized aegirine-augite up to acicular aegirine occurs in syenites and quartz syenite.

5.c. Mica

Micas from lamproites (Table 3) are solid solutions between the phlogopite and annite end-members, characterized by high titanium contents (1.29–6.28 wt% TiO₂). Rarely they show weak zoning, that is, a core-to-rim MgO increase at decreasing FeO, Al₂O₃ and TiO₂. A reversed zoning profile was observed in biotites from melasyenites (Fig. 4b).

Table 3. Selected electron microprobe analyses of micas

Biotites from the monzodiorite inclusion and the rounded aggregates in olivine-lamproites are Al-rich (> 13 wt% Al₂O₃) with significant eastonite solid solution (Fig. 4b).

In syenites and quartz syenite, micas are relatively homogeneous solid solutions between predominant phlogopite and subordinate annite end-members.

5.d. Amphibole

The sodic–calcic amphiboles from lamproites (Table 4) are mostly richterite, magnesiokatophorite and winchite (Leake et al. Reference Leake, Woolley, Arps, Birch, Gilbert, Grice, Hawthorne, Kato, Kisch, Krivovichev, Linthout, Laird, Mandarino, Maresch, Nickel, Rock, Schumacher, Smith, Stephenson, Ungaretti, Whittaker and Youzhi1997, Reference Leake, Woolley, Birch, Burke, Ferraris, Grice, Hawthorne, Kisch, Krivovichev, Schumacher, Stephenson and Whittaker2003; Hawthorne & Oberti, Reference Hawthorne and Oberti2007). Rarely, their rims evolve to ferri-glaucophane. In melasyenites, from core to rim the amphiboles grade from richterite/winchite to ferri-eckermannite and ferri-glaucophane. The amphiboles from monzodiorite inclusions have magnesiohornblende compositions with an exceptionally high Al₂O₃ content (up to 6.7 wt%).

Table 4. Selected electron microprobe analyses of amphiboles

In syenites the early-crystallized amphiboles are richterite and winchite, rimmed by ferri-eckermannite, eckermannite and, finally, ferri-glaucophane.

6. Crystallization temperature of the xenolith assemblages

The andalusite–biotite pair in the country rock places the upper pressure limit for emplacement of the Svidnya main intrusion at 0.37 ± 0.02 GPa (Spear & Cheney, Reference Spear and Cheney1989).

Plagioclase–amphibole thermometry (Holland & Blundy, Reference Holland and Blundy1994) was only possible in the monzodiorite inclusion, where rims of hornblendes and plagioclases yielded temperatures in the range 741–685 °C for a likely maximum pressure of 0.4 GPa, and 766–706 °C for a pressure of 0.2 GPa.

The two-feldspar geothermometer of Fuhrman & Lindsley (Reference Fuhrman and Lindsley1988) for coexisting alkali feldspar and plagioclase in the xenolith assemblage, yielded a temperature of 787–780 °C for pressures between 0.5 and 0.4 GPa. The pressure interval was assumed on the stability curve of biotite (Wones & Eugster, Reference Wones and Eugster1965; Huebner & Sato, Reference Huebner and Sato1970; Wones, Reference Wones1972) and is consistent with experimental results for the stability of Fe–Al biotites (Rutherford, Reference Rutherford1973). Based on this, Buda & Dobosi (Reference Buda and Dobosi2004) suggest 800 °C and 0.5 GPa for the crystallization of Mg-rich biotites from high-K mafic enclaves within Variscan granitoids.

The equilibration temperatures were determined with the SOLVCALC 1.0 software of Wen & Nekvasil (Reference Wen and Nekvasil1994), using the feldspar site mixing model of Fuhrman & Lindsley (Reference Fuhrman and Lindsley1988) with assessed uncertainties of ± 30 °C. Temperatures were calculated for pressures between 0.5 and 0.4 GPa at a compositional uncertainty of 0.020 (molar end-member composition). The result is a smooth curve with temperature rising with pressure at about 7 °C per 0.1 GPa.

By using the two-feldspar geothermometer, and based on the crystallization order, the near-solidus temperatures can be estimated at about 555–493 °C for pressures between 0.2 and 0.4 GPa, and correspond to the final stage of crystallization of the inclusion, in accordance with data from the Buhovo-Seslavtzi peralkaline rocks (Dyulgerov & Platevoet, Reference Dyulgerov and Platevoet2006).

7. Geochemistry

The Svidnya main intrusion, according to Le Bas et al. Reference Le Bas, Le Maitre, Streckeisen and Zanettin(1986), is peralkaline (molar (K+Na)/Al > 1.0, Table 5) and generally perpotassic (molar K/Al between 0.6 and 1.0). Following Foley et al. (Reference Foley, Venturelli, Green and Toscani1987), it may be considered ultrapotassic (MgO > 3 wt%; K₂O > 3 wt% and K₂O/Na₂O > 3). On the basis of chemical analyses by Stefanova (Reference Stefanova1966), Foley et al. Reference Foley, Venturelli, Green and Toscani(1987) referred the Svidnya main intrusion to group IV of ultrapotassic rocks, transitional between lamproites, kamafugites and rocks of orogenic areas. However, they plot in the lamproite field in the K₂O, MgO and Al₂O₃ classification diagram (Fig. 6a; Bergman, Reference Bergman, Fitton and Upton1987), as well as in the Foley et al. (Reference Foley, Venturelli, Green and Toscani1987) variation diagrams (Fig. A2 in online Appendix at http://journals.cambridge.org/geo). The normative classification of syenite plugs and dykes is consistent with alkali-feldspar syenite and alkali-feldspar quartz syenite (Fig. 6b).

Table 5. Whole rock major (%) and trace element (ppm) analyses

Figure 6. Whole rock classification diagrams. Small circles are literature data of Svidnya ultrapotassic rocks (Vladykin, Grozdanov & Bonev, Reference Vladykin, Grozdanov and Bonev2001). (a) K₂O, MgO and Al₂O₃ classification diagram for lamproites (Bergman, Reference Bergman, Fitton and Upton1987). (b) CIPW normative AQP composition of syenites (Le Bas & Streckeisen, Reference Le Bas and Streckeisen1991). For legend see Figure 4.

The lamproites are silica-saturated with SiO₂ in the range 49.7–56.4 wt%, Mg-number up to 44, Al₂O₃ contents between 10.7 and 12 wt%, TiO₂ about 1.3 wt%, Cr and Ni up to 309 and 86 ppm, respectively (Fig. 9a; Table 5). Melasyenites have higher SiO₂ contents, up to 60.8 wt%, Mg-number from 33 to 47, Al₂O₃ between 6.2 and 10.5 wt%, TiO₂ in the range 1.0–1.5 wt%, Cr and Ni abundances are lower than 250 and 70 ppm, respectively (Fig. 9a).

Compared with anorogenic lamproites, the Svidnya lamproites and melasyenites are characterized by: (1) higher Ba/Sr (in the range 4.5–6.7) than Roman Province-type (RPT) lavas (0.5–2.0), kimberlites, alkali basalts, lamprophyres (1–1.4) and primitive mantle (0.3); (2) significantly higher Zr/Nb (7.5–18.3) than kimberlites, alkali basalts and lamprophyres (0.4–4), but similar to primitive mantle (13) and (3) lower Ni contents and Nb/Th (0.2–1.0).

On the whole, SiO₂, low TiO₂, low Ni and Nb/Th make the Svidnya main intrusion similar to orogenic ultrapotassic lavas (TiO₂ < 2 wt%; Foley et al. Reference Foley, Venturelli, Green and Toscani1987) and to Mediterranean Cenozoic lamproites (Prelevic et al. Reference Prelevic, Foley, Romer and Conticelli2008; Fig. 7a, b), particularly the peralkalic Sisco, Corsica (Conticelli et al. Reference Conticelli, D'Antonio, Pinarelli and Civetta2002) and Cancarix and Calaspara, Spain (Prelevic et al. Reference Prelevic, Foley, Romer and Conticelli2008) examples. Compared with Carboniferous Mg–K granites from the French Central Massif, Svidnya lamproites have lower SiO₂ and higher TiO₂, but similar MgO/(MgO + FeO_t) and K₂O/K₂O + Na₂O (Sabatier, Reference Sabatier1980; Fig. A3 in online Appendix at http://journals.cambridge.org/geo).

Figure 7. Binary correlations: (a) Ni v. Cr; (b) Nb v. Zr. Data source from Mitchell & Bergman (Reference Mitchell and Bergman1991). Compositional fields are: M-LH – madupitic lamproite Leucite Hills; M-MA – madupitic lamproite Murcia Almeria; OL-WK – olivine lamproite West Kimberley; PL-LH – phlogopite lamproite Leucite Hills; PL-MA – phlogopite lamproite Murcia Almeria; PL-WK phlogopite lamproite West Kimberley; RPT-lavas – Roman-province-type lavas; SB – Smoky Butte lamproite. For legend see Figure 4.

The Svidnya lamproites and melasyenites show homogeneous REE distribution patterns normalized to chondrite (Sun & McDonough, Reference Sun, McDonough, Saunders and Norry1989; Fig. 8a) with lower REE abundances (ΣREE in the range 317–477; Table 6) and very low La/Yb (15 < La_N/Yb_N < 24) compared with anorogenic lamproites. They show a weak negative Eu anomaly (Eu/Eu* on average 0.59), as in the Sisco lamproites, and patterns similar to those of the Murcia–Almeria lamproites. The monzodiorite inclusion shows higher ΣREE than the host rocks, with Eu negative anomaly and HREE fractionation. The syenite and quartz syenites are characterized by ΣREE in the range 142–591, with evident HREE fractionation (14 < La_N/Yb_N < 50) and small negative Eu anomaly (Eu/Eu* on average 0.61; Fig. 8b).

Figure 8. Chondrite normalized REE-patterns (Sun & McDonough, Reference Sun, McDonough, Saunders and Norry1989) for (a) lamproites and monzodiorite intrusion, (b) syenites and quartz syenites. For legend see Figure 4.

Table 6. Whole rock rare earth element concentrations (ppm)

The multi-element diagram normalized to primitive mantle (Sun & McDonough, Reference Sun, McDonough, Saunders and Norry1989; Fig. 9a) shows a common LILE enrichment with HFSE fractionation, Nb, Ta, Sr, and Ti troughs, Th and U spikes. Syenites and quartz syenites show strongly variable Th, U, Hf and Zr abundances, and Nb, Ta, Sr and Ti troughs (Fig. 9b). Fractionated syenites tend to be Zr- and Hf-enriched.

Figure 9. Primitive mantle-normalized multi-element diagram (Sun & McDonough, Reference Sun, McDonough, Saunders and Norry1989) for (a) lamproites and monzodiorite intrusion, (b) syenites and quartz syenites. For legend see Figure 4.

8. Emplacement age

8.a. Results

Two ⁴⁰Ar–³⁹Ar age determinations were carried out on amphibole and biotite separates from the lamproite BL42.

The argon release spectra are given in Figure 10. The biotite fraction (Fig. 10a) yielded a plateau age of 339.1 ± 1.6 Ma for 79.9% of the released³⁹ Ar. The total fusion age is 335.1 ± 3.1 Ma and is therefore concordant with the plateau age.

Figure 10. ⁴⁰Ar/³⁹Ar incremental release spectra for biotite (a) and amphibole (b) of the Svidnya lamproites (sample BL42).

The amphibole fraction (Fig. 10b) yielded a near-plateau age of 337 ± 4 Ma. At lower temperatures the oldest age of 365.5 ± 36.3 Ma represents almost 0.5% of the released gas. A cluster of ⁴⁰Ar/³⁹Ar increments contributes to about 16% of the released gas. This cluster corresponds to the youngest and apparent ages, and approaches the total fusion age (313 ± 3 Ma). Two following incremental heating steps correspond to 337 ± 4 Ma for 35% of the gas released. At higher temperatures, ages drop to 303 ± 3 Ma and then again rise to 336 ± 4 Ma, that is, 24.9 and 10.1% of the released gas, respectively.

While the apparent ages for both fractions overlap within uncertainties, the total fusion ages for the amphibole and biotite fractions are slightly different.

9. Sr–Nd isotopic data

9.a. Lamproites

The measured ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd range from 0.71278 to 0.71599 and 0.512155 to 0.512219 in lamproites and melasyenites (Table 7). The Nd and Sr isotope ratios were corrected back to 339 Ma on the basis of the ⁴⁰Ar–³⁹Ar dating on biotite (Fig. 11). The initial ϵ_Nd values range between −4.87 and −5.88, suggesting a source relatively enriched in LREE with respect to Bulk Earth. The Nd model ages calculated (Ludwig, Reference Ludwig1994) with respect to a depleted source (initial ϵNd between +4.4 and +5.0) range between 1.3 and 1.5 Ga, thus representing the age of the enrichment event, assuming a pristine depleted mantle source. The relatively radiogenic Sr isotopic composition (⁸⁷Sr/⁸⁶Sr_i = 0.70694–0.70769) agrees with a time-integrated enriched mantle source. Despite the restricted dataset, the lack of a positive correlation of ⁸⁷Sr/⁸⁶Sr with lithophile elements, the enrichment of most incompatible elements (e.g. Ti, Th, Nb, La), and high Cr and Ni compared with continental crust abundances (Rudnick & Gao, Reference Rudnick, Gao, Holland and Turekian2004) suggest that crustal contamination alone is insufficient to account for the radiogenic signature of these magmas. The isotopic data are instead consistent with an origin of lamproites from a depleted mantle, contaminated by an enriched component, probably related to metasomatic processes by sediment dehydration during an ancient subduction event. In addition, the Sr and Nd isotope compositions of the Kfs–Cpx–Plg–Bt–Hbl-bearing monzodiorite inclusion and of its host rock, suggest a common origin from the same mantle source.

Table 7. Whole rock Sr and Nd isotope data

Initial Sr and Nd are recast to 339 Ma (⁴⁰Ar–³⁹Ar dating on biotite, present work).

Figure 11. Initial ϵNd v. initial ⁸⁷Sr/⁸⁶Sr. The ⁸⁷Sr/⁸⁶Sr of the ‘Bulk Earth’ has been calculated at 339 Ma (⁴⁰Ar–³⁹Ar dating on biotite, present work), using a present-day Sr isotope composition of 0.7045 and Rb/Sr = 0.029 (DePaolo & Wasserburg, Reference DePaolo and Wasserburg1976). For comparison are reported the fields of Variscan ultrapotassic plutons from southern Hungary (Buda & Dobosi, Reference Buda and Dobosi2004), the French Central Massif (Livradois area: Solgadi et al. Reference Solgadi, Moyen, Vanderhaeghe, Sawyer and Reisberg2007) and Corsica (Cocherie et al. Reference Cocherie, Rossi, Fouillac and Vidal1994), the Cenozoic Mediterranean potassic province and the main lamproite provinces (Wilson, Reference Wilson1991). For legend see Figure 4.

10. Igneous evolution

The geochemical data suggest the origin of Kfs–Cpx–Plg–Bt–Hbl-bearing monzodiorite inclusions and lamproites by fractionation from a common parental magma. In particular, major, trace and rare earth element contents of lamproites are in accordance with SiO₂-rich lamproitic compositions (Carlier & Lorand, Reference Carlier and Lorand2003; Mitchell & Edgar, Reference Mitchell and Edgar2002), but the monzodiorite has a higher total ΣREE content.

Conversely, unequivocal cogenetic correlations between lamproites and syenites based on major and trace element chemistry are lacking. The rare occurrence of phlogopitized biotite in lamproites (Table 3) and metastable titanian phlogopite in syenites may suggest hybridization processes (Prelevic et al. Reference Prelevic, Foley, Cvetkovic and Romer2004). However, the mineralogy, the whole rock compositions and the crystallization order in syenites suggest a melt richer in Na₂O, SiO₂ and H₂O. Different accessory minerals have also precipitated out; for example, Ba- and Nb-titanosilicates are lacking in lamproites, likely due to element partitioning in other silicates in spite of high contents in the bulk composition. For this reason, the mineral chemistry and microtextures are reliable evidence for two distinct hybridizations. The positive correlation between MgO and P₂O₅ and the variable Zr, Nb, Ce and Th in lamproites and syenites could reflect the melting of a source containing residual apatite, but with a heterogeneous distribution of other accessory minerals, possibly a veined mantle producing variably hybridized melts (Venturelli et al. Reference Venturelli, Mariani, Foley, Capedri and Crawford1988).

Syenites and quartz syenites show apparent HREE fractionation, likely an effect of cumulus plus fractional crystallization that also explains their ΣREE. Both in lamproites and in syenites, the negative Eu anomaly is inferred as inherited from the source, as the precipitation of sodic plagioclase only starts from syenite compositions.

Two evolved syenites deviate from the general Sr–Nd isotopic trend towards more ⁸⁷Sr radiogenic values, attributed to selective contamination with circulating fluids of crustal origin. These samples are also characterized by almost chondritic Nd isotopic values, implying a mantle source enriched in incompatible elements and LREE. This could be interpreted as the effect of melting of an isotopically heterogeneous source, characterized by different domains, which underwent distinct evolutionary trends from a pristine depleted mantle, as revealed by the initial Nd isotopic ratio obtained from model-ages.

For Mediterranean Cenozoic Fo-rich olivine-, Cr-rich spinel-bearing lamproites, olivine phenocrysts and xenocrysts are assumed as a proxy of a depleted component in the mantle source. In this case, an origin in an ultra-depleted lithospheric mantle affected by supra-subduction magmatism (boninites), followed by subduction of crustal-derived metasediments during the Alpine collision, has been suggested (Prelevic & Foley, Reference Prelevic and Foley2007).

In the Balkan orogen, an island-arc association with a 572 Ma boninite protolith, tholeiites and flysch deposits in the East Rhodope (central-eastern Bulgaria: Haydoutov et al. Reference Haydoutov, Kolcheva, Daieva, Savov and Carrigan2004) and in the Moldanubian zone were subducted during the Variscan collision. Based on the isotopic ratios (ϵNd_339Ma = −4.87 to −6.35), lamproites, melasyenites and syenites share a similarly enriched mantle source. In this regard, the timing of subduction of the ultra-depleted lithospheric mantle is later than in the model of Prelevic & Foley (Reference Prelevic and Foley2007), however, the mechanism and the lithospheric components might be still comparable in the petrogenesis of Svidnya lamproites.

11. Geodynamic implications

The post-collisional high-K plutonism recorded in the Bohemian, Austro-Alpine, Vosges, French and Corsica domains likely has a genetic link with the subduction setting in evidence along the Variscan collision front. In particular, the European Variscan ultrapotassic plutons are generally interpreted as having originated in a post-collisional setting (e.g. Finger et al. Reference Finger, Roberts, Haunschmid, Schermaier and Steyrer1997; Bonin, Reference Bonin2004). However, in detail, different mechanisms are invoked to account for their emplacement, such as (1) strike-slip tectonics (central Alps; Schaltegger et al. Reference Schaltegger, Gnos, Küpfer and Labhart1991), (2) extensional tectonics and uplift of deep crust (southern Hungary; Buda & Dobosi, Reference Buda and Dobosi2004), and (3) magma ascent along the uplift channels of the associated high-pressure rocks, due to heating of an enriched mantle by asthenosphere, in a slab break-off setting (Moldanubian domain; Finger et al. Reference Finger, Gerdes, Janousek, René and Riegler2007).

In the Bulgarian sector, ultrapotassic magmatism is coeval with the main Variscan tectonic and metamorphic stage and follows the exhumation from at least 40 km depth, of a migmatite- and eclogite-bearing gneiss–amphibolite complex dated at 398 ± 5.2 Ma by ³⁹Ar–⁴⁰Ar on hornblende, which represents the high-pressure record in the Sredna Gora terrane (≈ 1.2 GPa; Gaggero et al. Reference Gaggero, Buzzi, Haydoutov and Cortesogno2008). Thermal re-equilibration, following break-off of the subducting plate, would induce the partial melting of those domains within the lithospheric mantle metasomatized during the oceanic subduction and/or earlier events (Davies & von Blanckenburg, Reference Davies and Blanckenburg1995).

The strong enrichments in K, Rb, Ba and Th of Svidnya lamproites, and particularly the high Rb/Sr and Rb/Ba values, support the contribution of subducted sediments to the magma composition (e.g. Rogers et al. Reference Rogers, Hawkesworth, Mattey and Harmon1987), whereas the low Nb/Zr values indicate that a component from the partial melting of subduction-modified mantle cannot be excluded (Thompson & Fowler, Reference Thompson and Fowler1986). The Nd model ages may date a Mesoproterozoic metasomatic event (1.3–1.5 Ga) and overlap with the 1.4–1.7 Ga indicated for the lithospheric mantle roots of the European Variscan fold belt (Liew & Hofmann, Reference Liew and Hofmann1988).

During Carboniferous times, deformation propagated eastwards with the progressive closure of the Palaeotethys ocean. During this process, the western parts of the Variscan orogen were characterized by Himalayan-type continent–continent collision, whereas an Andean type ocean–continent setting is inferred in its eastern parts (Bonin, Reference Bonin2004). The roll-back of the subducting Palaeotethys slab has been invoked to account for the post-collisional ultrapotassic igneous products, such as in the External Crystalline Massifs of the Alps (von Raumer, Stampfli & Bussy, Reference von Raumer, Stampfli and Bussy2003). The model encompasses a stepwise development through: (1) lithosphere stacking and rapid uplift of the overriding plate, (2) decompression melting of an ultra-depleted mantle metasomatically enriched to originate ultrapotassic melts (Cocherie et al. Reference Cocherie, Rossi, Fouillac and Vidal1994), (3) break-off of the subducted slab and partial melting in the overriding lithosphere at shallower depth, with production of alkaline and calc-alkaline melts (Finger et al. Reference Finger, Gerdes, Janousek, René and Riegler2007; Holub, Cocherie & Rossi, Reference Holub, Cocherie and Rossi1997; (4) lithospheric thinning through delamination.