magmatic evidence for carbonate metasomatism in the ... · magmatic evidence for carbonate...

Magmatic Evidence for Carbonate

Metasomatism in the Lithospheric Mantle

underneath the Ohre (Eger) Rift

Philipp A. Brandl1,2*, Felix S. Genske1,3,4, Christoph Beier1,

Karsten M. Haase1, Peter Sprung5 and Stefan H. Krumm1

1GeoZentrum Nordbayern, Friedrich-Alexander-Universitat Erlangen–Nurnberg, Schloßgarten 5, 91054 Erlangen,

Germany, 2Research School of Earth Sciences, The Australian National University, 142 Mills Road, Acton, ACT

2601, Australia, 3CCFS, GEMOC, Department of Earth and Planetary Sciences, Macquarie University, Sydney,

NSW 2109, Australia, 4Institut fur Mineralogie, Westfalische Wilhelms-Universitat Munster, Corrensstr. 24, 48149

Munster, Germany and 5Institut fur Geologie und Mineralogie, Universitat zu Koln, Zulpicher Strasse 49b, 50674

Koln, Germany

*Corresponding author. Telephone: þ61 (0)2 6125 4301. Fax: þ61 (0)2 6125 8253. E-mail:

[email protected]

Received September 12, 2014; Accepted August 18, 2015

ABSTRACT

Magmas erupted in intracontinental rifts typically form from melting of variable proportions of as-

thenospheric or lithospheric mantle sources and ascend through thick continental lithosphere. Thisascent of magma is accompanied by differentiation and assimilation processes. Understanding the

composition of rift-related intracontinental volcanism is important, particularly in densely popu-

lated active rift zones such as the Ohre (Eger) Rift in Central Europe. We have sampled and ana-

lysed nephelinites from Zelezna hurka (Eisenbuhl), the youngest (<300 ka) Quaternary volcano

related to the Ohre Rift where frequent earthquake swarms indicate continuing magmatic activity

in the crust. This nephelinite volcano is part of a larger eruptive centre (Mytina Maar) representing

a single locality of recurrent volcanism in the Ohre Rift. We present a detailed petrographic, min-eralogical and geochemical study (major and trace elements and Sr–Nd–Hf–O isotopes) of Zelezna

hurka to further resolve the magmatic history and mantle source of the erupted melt. We find evi-

dence for a highly complex evolution of the nephelinitic melts during their ascent to the surface.

Most importantly, mixing of melts derived from different sources and of strong chemical contrast

controls the composition of the erupted volcanic products. These diverse parental melts originate

from a highly metasomatized subcontinental lithospheric mantle (SCLM) source. We use a com-bined approach based on mineral, glass and whole-rock compositions to show that the mantle

underneath the western Ohre Rift is metasomatized dominantly by carbonatitic melts. The nephel-

inites of Zelezna hurka formed by interaction between a carbonatitic melt and residual mantle peri-

dotite, partial crystallization in the lithospheric mantle and minor assimilation of upper continental

crust. Thermobarometric estimates indicate that the stagnation levels of the youngest volcanism in

this part of the Ohre Rift were deeper than the focal depths of recent earthquake swarms, indicating

that those are not directly linked to magma ascent. Furthermore, close mineralogical andgeochemical similarities between the Zelezna hurka nephelinite and fresh kimberlites may point

towards a genetic link between kimberlites, melilitites and nephelinites.

Key words: continental rift; nephelinite; carbonated peridotite; mantle metasomatism; assimilation

VC The Author 2015. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected] 1743

J O U R N A L O F

P E T R O L O G Y

Journal of Petrology, 2015, Vol. 56, No. 9, 1743–1774

doi: 10.1093/petrology/egv052

Advance Access Publication Date: 13 October 2015

Original Article

at Erlangen N

uernberg University on A

ugust 15, 2016http://petrology.oxfordjournals.org/

Dow

nloaded from

http://www.oxfordjournals.org/

http://petrology.oxfordjournals.org/

INTRODUCTION

The complex geology of Central Europe is a result of its

evolution throughout the Phanerozoic, which is domi-

nated by several cycles of plate collision, plate reorgan-ization and rifting. Major parts of the European

lithosphere were accreted during the Variscan Orogeny

and their tectonic fabrics still dominate the structure of

the Central European lithosphere (e.g. Ziegler et al.,

2006; Schulmann et al., 2014). Orogenic collapse fol-

lowed during the Permian and the lithosphere became

stabilized during the Late Cretaceous. However, subse-quent major plate reorganization related to the collision

of Africa and Europe (the Alpine Orogeny), together

with proposed mantle plume activity (e.g. Granet et al.,

1995; Hoernle et al., 1995) led to the formation of the

European Cenozoic Rift System (e.g. Ziegler et al., 2006)

and widespread Tertiary–Quaternary volcanism. ThisCentral European Volcanic Province (CEVP) encom-

passes intraplate and, more commonly, rift-related vol-

canism; for example, in the Massif Central at the

southern tip of the Limagne graben, the Kaiserstuhl in

the Upper Rhine graben, the Vogelsberg linking the

Rhine graben and the Hessian depression, and in the

Ohre (Eger) Rift. Quaternary volcanic activity in CentralEurope is reported from the Massif Central (e.g.

Downes, 1987), the Ohre Rift (e.g. Ulrych et al., 2011,

2013) and the Eifel province, with the eruption of the

Laacher See being the last major eruption in Central

Europe, dated to 12�9 ka (e.g. van den Bogaard, 1995;

Nowell et al., 2006). Although active volcanism inCentral Europe seems to be currently dormant, active

degassing and seismic activity indicate continuing mag-

matic activity in the Ohre Rift (e.g. Weinlich et al., 1999;

Spicak & Horalek, 2001; Brauer et al., 2005).

Detailed studies of these volcanic systems are

required to fully understand the processes driving recent

magmatism in Central Europe. Lavas erupted in contin-ental rifts, for example, provide insights into the compos-

ition of the underlying mantle, in particular the

subcontinental lithospheric mantle (SCLM). More import-

antly, such melts allow constraints on the processes dur-

ing melting, melt extraction and ascent through the

continental lithosphere. Assimilation of crustal materialmay play a major role in their petrogenesis and the

SCLM probably provides a chemically variable but over-

all more enriched source reservoir compared with the

asthenospheric mantle (e.g. Foley, 1992). It is thus crucial

to identify the complex processes that affect volcanism

in continental rifts and to unravel their distinct compos-itional imprints on the erupted volcanic rocks. Insights

into melting and magmatic differentiation, along with

thermobarometric estimates, are important to link past

volcanism with recent signs of volcanic activity.

Nephelinites are volumetrically minor contributors to

global magmatism, but are ubiquitous in continental

intraplate and rift settings. Thus, the understanding ofthe mechanisms that produce these special melt types is

of wide-ranging interest.

Recently, several studies focused on the nature and

composition of the mantle source of alkaline volcanism

related to the Ohre Rift by studying mantle xenoliths

and cumulates entrained in the lavas (Geissler, 2005;

Geissler et al., 2007; Puziewicz et al., 2011; Ackermanet al., 2013, 2014; Spacek et al., 2013). These studies

have shown that the lithospheric mantle underneath the

rift is highly heterogeneous and often overprinted by

metasomatic processes related to alkaline or carbona-

titic melt infiltration. As a result, mantle lithologies are

highly variable, ranging from fertile, clinopyroxene-rich

lithologies [e.g. the ‘pyroxenite suite’ of Puziewicz et al.(2011)] to refractory harzburgites and dunites (e.g.

Puziewicz et al., 2011; Ackerman et al., 2013). Accessory

minerals such as apatite, phlogopite, ilmenite or rutile

have been identified from various locations within the

Ohre Rift (e.g. Puziewicz et al., 2011; Ackerman et al.,

2013, 2014). Direct evidence for carbonatitic melts hasbeen recorded from carbonate minerals and melt pock-

ets in mantle xenoliths from the Oberpfalz (NE Bavaria;

Ackerman et al., 2013; Spacek et al., 2013), Plesny Hill

(Ackerman et al., 2014) and Ksieginki (Puziewicz et al.,

2011), all of which are located west (Oberpfalz) or east

(Plesny Hill and Ksieginki) of the main Ohre Rift struc-ture. The metasomatic processes leading to these modi-

fications of the sub-rift mantle are likely to be multi-

stage events (e.g. cryptic Fe-metasomatism followed by

thermal rejuvenation and melt infiltration) and will thus

differ between distinct volcanic centres related to the

Ohre Rift (e.g. Puziewicz et al., 2011).

However, because xenoliths may sample preferen-tially the cool lithosphere rather than the actual magma

source and are not entrained in every volcanic system,

we demonstrate that the petrology and geochemistry of

volcanic minerals and rocks themselves can be used to

constrain magmatic processes and the mantle source

composition. For this purpose we focused on Zeleznahurka (Eisenbuhl), a young (<300 ka) volcano situated

at the eastern tip of the Ohre Rift that consists predom-

inantly of fresh tephra. We present high-precision geo-

chemical analyses of fresh volcanic glasses and show

that these differ significantly from the compositions of

whole-rock samples from the same locality and other

volcanoes nearby. Mineral textures and compositionsindicate a complex ascent history for the lava, including

the assimilation of crustal material. We determine the

conditions of melting and the magmatic evolution of

the erupted lavas and present further insights into vol-

canic processes and the nature of the mantle beneath

the Ohre Rift by combining major and trace elementdata with radiogenic (Sr–Nd–Hf) and stable O isotope

data.

GEOLOGICAL SETTING

Volcanism related to the Ohre Rift extends for morethan 400 km from northeastern Bavaria through the

Czech Republic into Poland. The rift itself represents an

1744 Journal of Petrology, 2015, Vol. 56, No. 9

at Erlangen N



Dow

nloaded from


ENE–WSW-trending extensional structure, about 25–

30 km wide, that follows Variscan crustal lineaments be-

tween the Saxothuringian terrane in the NW, the

Moldanubian terrane in the SE and the Tepla-

Barrandian terrane between those two terranes(Malkovsky, 1987; Babuska & Plomerova, 2010).

In the region of the Cheb Basin, the crust is thinned to

25–28 km (e.g. Geissler, 2005; Heuer, 2006). The

main phase of rift-related volcanism is dated to about

30–15 Ma with episodic volcanism extending to 0�26 Ma

(Ulrych et al., 2011, and references therein). Quaternary

volcanic rocks are present at Komornı hurka(Kammerbuhl) and the volcanic system of the Mytina

Maar and Zelezna hurka (Fig. 1a), both located roughly

above the locus of crustal thinning (Babuska &

Plomerova, 2010).

Komornı hurka is a 726 6 59 kyr old volcano (Wagner

et al., 2002) that has erupted sodalite-bearing or nephel-ine–olivine-melilitite as scoria and a lava flow (Ulrych

et al., 2013). In contrast, Zelezna hurka is younger and

consists of three eruptive units: volcaniclastic material

formed by a phreatomagmatic eruptive phase at its

base, overlain by highly olivine-phyric lava in the vent,

and tephra layers at the top (Fig. 1b). Recent studies(e.g. Geissler et al., 2004, 2007; Mrlina et al., 2009) that

combined geophysical studies with geochemistry and

information from scientific drilling found the remnants

of a larger eruption diatreme, the Mytina Maar, just

north of Zelezna hurka, which was dated by the Ar–Ar

method at 288 6 17 ka (Mrlina et al., 2007). Additional

evidence for continuing magmatic activity close to the

volcano are active degassing of CO2, mantle-derived He(high 3He/4He) in numerous mofette fields (e.g. at Soos

or Bublak; Weinlich, 2013) and recurrent earthquake

swarms (e.g. Fischer & Horalek, 2003). These earth-

quake swarms have a focal depth of around 6–11 km

and are either linked to the ascent, accumulation or

stagnation of magma in the crust (e.g. Dahm et al.,

2008) or, alternatively, may be explained by fluids as-cending along pre-existing and reactivated fault planes

(e.g. Bankwitz et al., 2003). Interestingly, the 3He/4He in

the gas exhalations increased significantly from 1993 to

2005, reaching values of 6�3 Ra that are similar to the

average of the SCLM (Brauer et al., 2009). This increase

was interpreted as evidence for the ascent of mantle-derived melts into the lithosphere beneath the western

Ohre Rift, with deep dike intrusions in 2006–2008

(Brauer et al., 2005, 2009). Thus, both seismic data and

the active degassing in the western Ohre Rift suggest

continuing magmatic activity at depth.

METHODS

Samples L1 to L3 were collected from a basal, brownish

phreatomagmatic tephra unit (lower tephra; Fig. 1b)

and samples V1 to V4 along a traverse directly

above sample L1 towards the west of the outcrop (vent;

Fig. 1b). Samples U1, U2 (with a fine-grained variety

U2f), and U3 were collected from the base, the middleand upper layer of the upper tephra unit (Fig. 1b). An

additional sample (EG0661) had previously been col-

lected from the upper tephra. For geochemical analyses

of whole-rocks, xenolith- and crystal-poor samples

were selected. Weathered surfaces were removed prior

to crushing and the rocks were rinsed in de-ionized

water. Splits of the crushed materials were further pro-cessed for glass and mineral separation and representa-

tive whole-rock pieces were cut for thin sections. A split

of sample V1 (massive but highly olivine-phyric lava)

was crushed by high-voltage pulse power fragmenta-

tion in the laboratories of Selfrag AG, Kerzers

(Switzerland), to separate olivine crystals for major andtrace element and O isotope analyses.

Major elementsMajor element analyses of glasses and minerals were

performed on a JEOL JXA-8200 electron microprobe atthe GeoZentrum Nordbayern, Friedrich-Alexander-

Universitat Erlangen–Nurnberg. Glasses were analysed

using an acceleration voltage of 15 kV, a beam current

of 15 nA and a defocused beam of 10 mm diameter.

Further details of the analytical conditions have been

given by Brandl et al. (2012). Major element compos-itions of minerals (olivine, clinopyroxene, spinel,

phlogopite and minerals of crustal xenoliths) were

200 km 40

Ch

M

S

DHM

OPF

KH

ZHMLF

WBSZ

Eger Rift

FL

12°E 14°E 16°E

50°N

51°N

49°N

CZ

ATD

PL

CZ

D

Cenozoicvolcanicrocks

Cenozoicsediments

Lowertephra

Lava (vent) Upper tephraL3

L1L2

V1V2V3V4

U3 U2U2f

U1

(a)

(b)

N

appr. 5 m

Fig. 1. (a) Map of the western Ohre Rift at the structural bound-ary between the Variscan Saxothuringian and Moldanubianterranes in the Czech–German border region. OPF, Oberpfalz;KH, Komornı hurka; ZH, Zelezna hurka; DHM, Doupovske horyMountains; M, Mitterteich Basin; Ch, Cheb Basin; S, SokolovBasin. Major structural features: MLF, Marianske Lazne Fault;WBSZ, West Bohemian Shear Zone; FL, Franconian Line. (b)Schematic cross-section of the outcrop at Zelezna hurka, show-ing its lithological structure and sample locations. Lx samplesrepresent the lower tephra unit; Vx samples are from the vent;Ux samples are from the upper tephra unit.

Journal of Petrology, 2015, Vol. 56, No. 9 1745

at Erlangen N



Dow

nloaded from


determined using 20 kV acceleration voltage, 20 nA

beam current and a focused beam.

Major element analyses of whole-rock samples (and

the trace elements Ba, Cr, Ga, Nb, Ni, Rb, Sr, V, Y, Zn

and Zr) were carried out by X-ray fluorescence (XRF) ona Spectro XEPOS plus at the GeoZentrum Nordbayern.

Further details of the analytical technique have been

given by Freund et al. (2013).

Trace elementsThe trace element analyses of whole-rock powders (so-lution inductively coupled plasma mass spectrometry;

ICP-MS) were performed at the Geochemical Analysis

Unit (GAU) at Macquarie University, Sydney. The same

analytical protocol was followed as presented by

Genske et al. (2012). Approximately 100 mg of sample

powder was dissolved using a mix of HNO3–HF–HClacids for digestion. To fully dissolve Fe–Ti oxides, a

mixture of HCl and HClO4 was also used throughout the

digestion procedure. The analytical data for the sam-

ples and international rock standards were obtained on

an Agilent 7500 c/s quadrupole ICP-MS system. All ana-

lytical results (whole-rocks and glasses), including re-

producibility, are reported in Table 1.Laser ablation (LA)-ICP-MS analyses of volcanic

glasses were carried out at the GeoZentrum

Nordbayern on a New Wave Research UP193FX laser

ablation system coupled to an Agilent 7500i quadrupole

ICP-MS system. We averaged our results over at least

four spots from each sample. Laser ablation was carriedout on 25 mm spots with 0�72 GW cm–2 laser energy and

3�6 J cm–2 energy density, measuring the background

for 25 s and the sample for 30 s. Lithium, Si and Mn

were analysed for 10 ms on the maximum peak and

other elements for 25 ms (30 ms for Ta), resulting in a

total of 1�0082 s per mass scan. Silica concentrations

determined by electron microprobe were used for in-ternal calibration and standard glass NIST 612 was

used for external calibration (Pearce et al., 1997).

Accuracy and reproducibility were checked using sec-

ondary standards NIST 614 and BCR-2 g. Precision and

accuracy are generally better than 10%, with slightly

higher values for Cr, Nd and Hf (14�04, 11�93 and11�33% RSD, respectively). Laser ablation conditions for

the trace element analyses of minerals (olivine, clino-

pyroxene, phlogopite, quartz, spinel) were similar to the

conditions for glasses described above, but using a

spot size of 50mm (for small minerals or crystal rims of

25mm; see Table 2 for further details) with 0�66 GW cm–2

laser energy and 3�3 J cm–2. Results of representativemineral analyses are reported in Table 2 (complete data

for mineral analyses can be found in Supplementary

Data Table S1; supplementary data are available for

downloading at http://www.petrology.oxfordjournals.

org).

Detailed comparison and evaluation of trace elementdata determined on rock standards by both LA-ICP-MS

and solution ICP-MS reveal that selected elements and

corresponding element ratios deviate to slightly higher

values than reported in the literature. In particular, the

high field strength element (HFSE) Zr is determined to

be up to 8% higher in BHVO-2 than for high-precision

isotope dilution data presented on the same USGSstandard by Pfander et al. (2007). However, we note that

in comparison with preferred GeoReM data, our data

agree within one standard deviation (i.e. <5%), which is

the commonly achieved precision for the techniques

employed here (Table 1). Nevertheless, taking the max-

imum error into account would still result in relatively

high Zr/Sm and low Zr/Nb for the Zelezna hurka lavascompared with the Ohre Rift (see discussion below).

Radiogenic isotope analysesStrontium and Nd isotopes in whole-rocks and glasses

were analysed at the GAU, Macquarie University,Sydney. The analytical routine applied is the same as

described by Genske et al. (2012). The isotopic analyses

of Sr and Nd were conducted using thermal ionization

mass spectrometry (TIMS) employing a Thermo

Finnigan Triton system. Measured 87Sr/86Sr ratios for

BHVO-2 obtained during this study are listed in Table 1.

The standard NIST SRM 987 was analysed (n¼17) toverify the accuracy of the measurements during the

period of sample analysis. The long-term reproducibil-

ity of NIST SRM 987 is 87Sr/86Sr¼0�710250

(2SD¼0�000034). Ratios were normalized to86Sr/88Sr¼ 0�1194 to correct for mass fractionation. For

the Nd analyses, reference materials BHVO-2 and JMC321 were analysed, yielding 143Nd/144Nd ratios close to

published values (Table 1). The external precision was

determined using JMC 321 (n¼ 15), which yielded143Nd/144Nd¼0�511115 (2SD¼ 0�000047). Ratios were

normalized to 146Nd/144Nd¼ 0�7219 to correct for mass

fractionation.

Determinations of Hf isotope compositions of whole-rocks and combined determinations of Hf isotope com-

positions and Lu and Hf concentrations of glasses by

isotope dilution and spike stripping using a mixed176Lu–180Hf tracer were conducted at the WWU

Munster, Germany. Sample preparation, mass spec-

trometry on a Neptune Plus multicollector (MC)-ICP-MSsystem at the Universitat Munster, and estimation of

measurement and spike-stripping uncertainties fol-

lowed the method of Sprung et al. (2010, 2013), which

makes use of a modified Ln-spec Hf purification scheme

after Munker et al. (2001). To minimize uncertainties

associated with possible isobaric interferences from

spiked Lu, a final Lu removal step using AG 50 W-x8(1 ml) was added in which Hf was immediately eluted

upon loading in 0�5 M HCl–0�05 M HF, leaving Lu ad-

sorbed on the resin. Mass bias was internally corrected

for using the exponential law described by Russell

et al. (1978) and normalizing to 179Hf/177Hf¼0�7325.

All 176Hf/177Hf values are given relative to176Hf/177Hf¼ 0�28216 for Ames Hf, which is isotopically

indistinguishable from JMC-475 (Scherer et al., 2000).


at Erlangen N



Dow

nloaded from

http://petrology.oxfordjournals.org/lookup/suppl/doi:10.1093/petrology/egv052/-/DC1



http://www.petrology.oxfordjournals.org

http://www.petrology.oxfordjournals.org


Table 1: Major and trace element and isotope analyses of samples from Zelezna hurka and international reference material

L1 L2 L3 V1 V2 V3 V4 U1 U2lower lower lower vent vent vent vent upper uppertephra tephra tephra lava lava lava lava tephra tephraGlass Glass Glass WR WR WR WR WR GlassEMPA EMPA EMPA XRF XRF XRF XRF XRF EMPA

SiO2 (wt %) 40�68 40�27 39�88 39�78 39�94 40�35 40�48 41�55 39�74TiO2 (wt %) 3�19 2�93 3�08 2�98 2�97 2�99 2�91 2�82 3�08Al2O3 (wt %) 14�34 14�58 14�78 11�37 11�45 11�58 11�31 11�61 14�17FeOt (wt %) 10�68 10�70 10�48 9�89Fe2O3

t (wt %) 12�69 12�70 12�20 12�42 12�23MnO (wt %) 0�25 0�27 0�25 0�215 0�214 0�216 0�211 0�205 0�24MgO (wt %) 5�54 4�80 5�03 12�63 12�50 12�54 12�68 12�12 5�37CaO (wt %) 15�23 14�56 14�64 13�05 12�86 12�90 12�80 12�55 14�08Na2O (wt %) 4�60 5�25 4�57 3�29 3�46 3�08 3�32 2�68 4�43K2O (wt %) 3�75 4�06 4�03 1�99 2�29 2�06 2�12 2�35 3�69P2O5 (wt %) 1�29 1�33 1�38 0�774 0�763 0�791 0�744 0�735 1�20S (ppm) 1294 1375 1210 1124Cl (ppm) 3167 3426 3318 2994LOI (wt %) 0�75 0�41 0�84 0�53 0�68Total (wt %) 100�12 99�36 98�68 99�52 99�55 99�54 99�52 99�52 96�42

LA- LA- sol. sol. sol. XRF XRFppm ICP-MS ICP-MS ICP-MS ICP-MS ICP-MS

Li 13�0 12�3 7�09 6�89 6�67Sc 22�1 6�51 33�3 32�8 28�7V 560 471 308 311 317 294 286Cr 680 583 521 565 549 518Mn 2244 2028Co 63�7 38�5 56�2 55�6 56�2Ni 53�8 10�48 214 222 213 229 208Cu 101 117 88�1 66�4 77�8Zn 214 158 102 99 101 103 97�2Ga 19�2 18�9 18�6 19�6 16�4Rb 97�4 116 60�2 62�8 51�2 65�3 80�1Sr 1528 1769 903 779 736 894 966Y 28�5 30�7 27�1 26�9 25�6 27�5 27Zr 275 255 283 281 280 264 269Nb 161 193 121 120 122 101 99�7Sn 2�98 1�83Cs 1�41 1�41 0�813 0�731 0�604Ba 1241 1515 779 757 823 806 872La 98�9 117 73�5 69�7 72�1Ce 186 218 138 133 136Pr 20�3 23�1 16�2 15�5 16�0Nd 76�8 86�0 60�0 58�7 59�1Sm 13�2 14�0 10�6 10�3 10�5Eu 3�87 4�10 3�06 2�95 3�02Gd 9�41 9�78 8�38 8�06 8�11Tb 1�14 1�26 1�14 1�11 1�11Dy 6�62 6�71 5�26 5�08 5�09Ho 1�06 1�16 0�918 0�897 0�891Er 2�60 2�86 2�27 2�19 2�17Tm 0�338 0�352Yb 2�18 2�20 1�65 1�62 1�60Lu 0�286 0�296 0�231 0�226 0�220Hf 5�61 4�26 5�66 5�64 5�62Ta 8�80 10�4 5�87 5�68 5�73W 2�34 2�52Pb 6�31 7�48 3�17 1�42 2�06 2�10 6�00Th 12�1 14�7 9�17 8�62 8�42 8�00 7�60U 3�61 4�47 2�32 2�34 2�4187Sr/86Sr 0�703891 0�703631 0�703455 0�703479 0�703591 0�703404143Nd/144Nd 0�512812 0�512758 0�512861 0�51286 0�512862 0�512855176Hf/177Hf 0�283002 0�283008 0�282997 0�283003 0�283000d18O 6�25 5�67 5�44 5�57dupl. 6�13 5�65 5�69

(continued)


at Erlangen N



Dow

nloaded from


Table 1: Continued

U2f U3 EG-0661 VG-2 BE-N BR NIST-614 BCR-2G BHVO-2 Absol. dev.upper upper upper n ¼ 6 to GeoReMtephra tephra tephraGlass WR GlassEMPA XRF EMPA EMPA XRF XRF

SiO2 (wt %) 40�52 40�70 40�51 50�53 38�63 38�51TiO2 (wt %) 3�10 2�88 3�11 1�79 2�67 2�66Al2O3 (wt %) 14�75 11�35 14�77 13�83 10�03 10�03FeOt (wt %) 10�60 10�48 11�63Fe2O3

t (wt %) 12�62 12�93 13�14MnO (wt %) 0�26 0�211 0�27 0�21 0�202 0�200MgO (wt %) 5�13 12�52 5�02 6�99 12�96 13�12CaO (wt %) 14�70 12�98 14�52 11�18 14�01 13�62Na2O (wt %) 4�55 2�66 4�52 2�67 3�16 2�99K2O (wt %) 4�04 2�41 4�01 0�19 1�40 1�37P2O5 (wt %) 1�38 0�744 1�35 0�21 1�06 1�04S (ppm) 1067 1227 1427Cl (ppm) 3359 3351 305LOI (wt %) 0�42 2�44 2�81Total (wt %) 99�56 99�49 99�11 99�62 99�48 99�48

LA- XRF LA- LA- LA- sol.ppm ICP-MS ICP-MS ICP-MS ICP-MS ICP-MS

Li 6�79 12�3 1�91 9�81 4�15 0�653Sc 9�35 7�51 1�59 35�1 32�7 0�677V 469 302 435 0�999 439 305 11�9Cr 2�29 545 4�89 1�17 16�9 256 24�1Mn 2017 1851 1�49 1526Co 39�5 37�3 0�792 39�2 43�7 1�27Ni 12�47 225 10�9 0�912 12�4 124 4�65Cu 72�7 95�9 3�55 14�6 132 5�33Zn 154 98�9 146 2�23 157 102 1�31Ga 16�4 20�5 1�51Rb 110 67 102 0�874 49�2 9�53 0�420Sr 1743 1013 1561 44�3 323 389 7�21Y 33�1 24�7 27�7 0�737 33�7 28�4 2�40Zr 281 265 249 0�755 170 180 7�67Nb 190 100 171 0�796 11�9 18�9 0�829Sn 1�73 1�94 1�68 2�11Cs 1�32 1�36 0�682 1�11 0�103 0�003Ba 1491 870 1325 3�15 640 133 1�69La 119 103 0�688 24�2 15�8 0�571Ce 214 195 0�769 50�5 38�1 0�563Pr 23�1 20�7 0�761 6�31 5�51 0�162Nd 87�7 77�3 0�727 27�4 24�8 0�323Sm 14�9 12�7 0�718 6�35 6�25 0�182Eu 4�11 3�78 0�693 1�81 2�00 0�068Gd 11�0 9�14 0�813 6�18 6�35 0�114Tb 1�39 1�18 0�639 0�956 0�960 0�040Dy 7�31 6�32 0�667 6�34 5�33 0�024Ho 1�21 1�05 0�699 1�26 1�01 0�032Er 3�02 2�63 0�713 3�53 2�59 0�054Tm 0�402 0�352 0�680 0�515Yb 2�38 2�16 0�741 3�47 1�98 0�023Lu 0�349 0�290 0�677 0�512 0�28 0�002Hf 5�16 4�98 0�746 4�58 4�26 0�103Ta 10�7 9�04 0�781 0�773 1�08 0�057W 2�41 2�15 0�852 0�526Pb 6�92 4�90 7�04 2�34 10�3 1�56 0�045Th 15�5 4�60 12�75 0�702 5�77 1�28 0�064U 4�32 3�93 0�764 1�67 0�43 0�02587Sr/86Sr 0�703459 60�000012*143Nd/144Nd 0�512978 60�000015*176Hf/177Hf 0�283009 0�283099 60�000004*d18O 5�77 5�48dupl. 5�57

(continued)


at Erlangen N



Dow

nloaded from


International rock standard BHVO-2 was analysed for

its Hf isotope composition alongside the unknowns,

yielding a 176Hf/177Hf composition of 0�283099

(2SD¼0�000004).

Oxygen isotope analysesOxygen isotope analyses were performed on fresh and

visibly inclusion-free olivines and glasses. Single grainswere handpicked and cleaned and then coarsely

crushed in a steel mortar to obtain multiple splits. Small

chips were embedded for electron microprobe analysis

(EMPA) and LA-ICP-MS analyses and the remaining ma-

terial (2–3 mg) was used to determine the O isotope

composition. Oxygen isotope analyses were performedusing the 25 W-Synrad CO2-laser fluorination line at the

GeoZentrum Nordbayern following the methods

described by Haase et al. (2011) and Genske et al.

(2013). The long-term reproducibility of the UWG-2

garnet standard obtained during the course of this

study is 5�84 6 0�07% (1SD, n¼ 29).

RESULTS

Samples from the lower tephra are of lapilli size, with

brownish-weathered surfaces but fresh and glassy in-

teriors. Blocky lavas in the western part of the outcrop

are situated in the position of the former vent and ap-pear unaltered and fresh. These lavas contain abundant

xenoliths and up to centimetre-sized crystals of olivine,

clinopyroxene and phlogopite. The upper tephra

reaches a thickness of about 5–7 m at the eastern side

of the outcrop and is uniformly dark. The crystal assem-

blages in this unit are similar to those found in the vent

lavas. Xenoliths in the upper tephra are of variable sizeranging from millimetre-scale to more than half a metre

and show reddish, oxidized surfaces. Crustal xenoliths

are common in all the lithological units sampled.

Petrography and mineral compositionWe selected representative samples for petrographic

description by optical examination under a binocular

microscope (Fig. 2). Samples L1 to L3 were collected

from the lower brown tephra (Fig. 1b) and have a glassybut phenocryst-rich groundmass. Samples V1 to V4

(massive lava flow in the vent) and U1 to U3 (upper

tephra) have a holo- to cryptocrystalline matrix and

about 30 vol. % vesicles (Fig. 2a). The mineral assem-

blages of these samples are similar and include about

20 vol. % olivine, 10–15 vol. % clinopyroxene and acces-sory spinel and hauyne. Within their volcanic matrix the

lavas host felsic crustal xenoliths (Fig. 2b) and ultra-

mafic cumulates (clinopyroxene and phlogopite and/or

olivine) (Fig. 2a–c).

OlivineOlivine is by far the most abundant mineral in theZelezna hurka lavas. Crystals of olivine range in size

from a few micrometres to 10 mm. Single crystals are

Table 1: Continued

BIR-1 Absol. dev.to GeoReM

SiO2 (wt %)TiO2 (wt %)Al2O3 (wt %)FeOt (wt %)Fe2O3

t (wt %)MnO (wt %)MgO (wt %)CaO (wt %)Na2O (wt %)K2O (wt %)P2O5 (wt %)S (ppm)Cl (ppm)LOI (wt %)Total (wt %)

sol. ICP-MSppm

Li 3�12 0�078Sc 45�0 1�98V 317 2�47CrMnCo 52�5 0�475Ni 179 12�7Cu 120 1�28Zn 67�4 4�65Ga 15�0 0�331Rb 0�191 0�009Sr 105 4�18Y 16�8 1�21Zr 15�3 1�33Nb 0�547 0�003SnCs 0�005 0�002Ba 6�52 0�622La 0�617 0�002Ce 1�90 0�016Pr 0�387 0�017Nd 2�33 0�046Sm 1�10 0�021Eu 0�487 0�043Gd 1�89 0�017Tb 0�353 0�007Dy 2�51 0�004Ho 0�576 0�016Er 1�72 0�065TmYb 1�59 0�055Lu 0�239 0�011Hf 0�566 0�016Ta 0�036 0�001WPb 3�22 0�115Th 0�039 0�007U 0�011 0�00187Sr/86Sr143Nd/144Nd176Hf/177Hfd18Odupl.

*2SD relative to BHVO-2.


at Erlangen N



Dow

nloaded from


Table 2: Representative major and trace element analyses of mineral phases and one interstitial glass from Zelezna hurka

Mineral: Olivine Olivine Olivine Cpx Cpx Phl Hauyne Spinel GlassSample: V4-17 V4-17 U1-Ol11 V1-12 I V1-12 VI V1-12 I V1 H5 V4-5 I V1 glassSpot size:* core rim 25 mm

wt %SiO2 40�56 40�10 40�52 50�18 48�87 38�27 31�99 0�09 68�96TiO2 b.d.l. 0�05 0�01 1�13 1�17 4�21 0�06 1�32 0�71Al2O3 0�06 0�05 0�03 6�74 7�50 17�29 26�44 22�63 6�43Cr2O3 0�05 0�01 0�04 0�27 0�38 0�32 0�01 44�38 b.d.l.FeOt 9�65 13�23 10�56 5�39 5�42 7�06 0�60 18�95 3�50MgO 50�14 46�58 49�97 15�27 14�56 20�06 0�58 15�97 1�39MnO 0�12 0�33 0�16 0�13 0�11 0�04 n.d. 0�40 0�12CaO 0�14 0�79 0�18 21�01 20�39 0�03 8�16 b.d.l. 2�19Na2O 0�01 b.d.l. b.d.l. 0�92 0�99 0�70 11�23 b.d.l. 6�15K2O b.d.l. 0�01 b.d.l. b.d.l. b.d.l. 9�24 7�87 b.d.l. 10�49NiO 0�35 0�05 0�24 0�02 b.d.l. 0�08 n.d. b.d.l. 0�02P2O5 0�01 0�02 0�02 0�03 0�01 0�25 0�17 0�12SO2 10�31Total 101�09 101�22 101�70 101�07 99�42 97�30 97�50 103�91 100�06ppmLi 2�09 4�48 2�03 1�00 0�999 14�2 1�19 27�4Sc 3�10 6�34 3�56 77�5 77�3 9�17 2�68 3�58V 5�92 5�29 5�67 355 375 424 930 58�5Cr 540 340 325 3022 2992 3071 n.d. 4�79Mn 1188 2122 1439 1053 1112 274 1271 728Co 146 163 159 33�6 34�2 78�8 184 8�26Ni 3597 1018 2190 168 169 668 1264 4�12Cu 2�76 1�89 2�93 1�76 1�82 1�86 7�16 53�3Zn 76�2 98�2 79�6 21�9 22�0 46�2 437 77�4Rb b.d.l. b.d.l. b.d.l. 0�024 0�062 332 0�025 153Sr b.d.l. 0�115 b.d.l. 77�7 75�8 115 0�016 216Y 0�023 0�076 0�034 8�88 8�81 0�044 b.d.l. 8�41Zr 0�059 0�289 0�034 41�7 41�8 4�57 0�803 33�9Nb b.d.l. 0�020 b.d.l. 0�436 0�452 8�88 0�446 24�1Cs b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 1�65 b.d.l. 1�63Ba b.d.l. 0�129 b.d.l. 0�098 0�514 2731 b.d.l. 431La b.d.l. 0�010 b.d.l. 3�00 2�77 0�007 b.d.l. 1�90Ce b.d.l. 0�038 b.d.l. 11�3 10�8 0�016 b.d.l. 29�8Pr b.d.l. 0�003 b.d.l. 2�05 1�89 b.d.l. b.d.l. 3�15Nd b.d.l. b.d.l. b.d.l. 10�9 10�6 b.d.l. b.d.l. 12�0Sm b.d.l. b.d.l. b.d.l. 3�17 3�02 b.d.l. b.d.l. 1�81Eu b.d.l. b.d.l. b.d.l. 0�953 1�02 0�046 b.d.l. 0�846Gd b.d.l. b.d.l. b.d.l. 2�77 2�86 b.d.l. b.d.l. 1�91Tb b.d.l. b.d.l. b.d.l. 0�398 0�374 b.d.l. b.d.l. 0�245Dy b.d.l. b.d.l. b.d.l. 2�27 2�14 b.d.l. b.d.l. 1�84Ho b.d.l. 0�005 b.d.l. 0�357 0�367 b.d.l. b.d.l. 0�288Er b.d.l. 0�028 0�010 0�878 0�807 b.d.l. b.d.l. 0�708Tm b.d.l. b.d.l. b.d.l. 0�113 0�11 b.d.l. b.d.l. 0�100Yb b.d.l. 0�051 b.d.l. 0�641 0�65 b.d.l. b.d.l. 0�759Lu b.d.l. 0�006 b.d.l. 0�073 0�085 b.d.l. b.d.l. 0�065Hf b.d.l. 0�028 b.d.l. 2�31 2�17 0�117 0�034 0�637Ta b.d.l. b.d.l. b.d.l. 0�125 0�146 0�687 0�025 0�968Pb b.d.l. 0�082 0�055 0�172 0�129 0�431 0�202 28�3Th b.d.l. b.d.l. b.d.l. 0�063 0�055 b.d.l. b.d.l. 1�74U b.d.l. b.d.l. b.d.l. 0�012 0�012 b.d.l. b.d.l. 0�514d18O 5�24Min. form.Si ¼ 0�99 0�99 0�98 1�82 1�81 5�29Ti ¼ 0�00 0�00 0�00 0�03 0�03 0�44Al ¼ 0�00 0�00 0�00 0�29 0�33 2�82Cr ¼ 0�00 0�00 0�00 0�01 0�01 0�04Fe2þ ¼ 0�20 0�27 0�21 0�16 0�17 0�82Fe3þ ¼ 0�00 0�00 0�00 0�00 0�00 0�00Mg ¼ 1�82 1�71 1�81 0�83 0�80 4�13Mn2þ ¼ 0�00 0�01 0�00 0�00 0�00 0�01Ca ¼ 0�00 0�02 0�00 0�82 0�81 0�00Na ¼ 0�00 0�00 0�00 0�06 0�07 0�19K ¼ 0�00 0�00 0�00 0�00 0�00 1�63Ni ¼ 0�01 0�00 0�00 0�00 0�00 0�01P

Cat. ¼ 3�01 3�01 3�02 4�03 4�03 15�64Fo/Mg# 90�3 86�3 89�4 83�5 82�7 83�5Wollast. 45�2 45�4Enstatite 45�7 45�1Ferrosilite 9�1 9�4

The full data table is given in Supplementary Data Table S1. b.d.l., below detection limit. n.d., not determined.*If not indicated otherwise, size of LA-ICP-MS spot is 50mm.


at Erlangen N



Dow

nloaded from



generally idiomorphic to hypidiomorphic. Larger crys-

tals are normally zoned (e.g. Fig. 3a) with well-defined

cores and overgrowth rims and sometimes dissolutiontextures at their grain boundaries (Fig. 2d). Small clino-

pyroxene phenocrysts often crystallize around these

olivine grains (Fig. 3e). Smaller groundmass crystals

lack chemical zoning. In terms of composition, olivine

displays a bimodal distribution in forsterite content be-

tween crystal cores (�Fo90) and overgrowth rims and

groundmass crystals (�Fo86; inset Fig. 4). These twogroups are also distinct in their Ni (Fig. 4a), Ca (Fig. 4c)

and Mn (not shown) concentrations. Cores have up to

3000 ppm Ni, between 800 and 1500 ppm Mn and 900–

1300 ppm Ca. The olivine rims and smaller olivine crys-

tals from the matrix have lower Ni contents of between

400 and 1300 ppm, but higher Mn (1300–2500 ppm) andCa concentrations (1100–4000 ppm; some up to

8300 ppm). Representative mineral analyses and data

for olivine separated from sample V2 (Ol-1 to Ol-18) are

presented in Table 2.

The olivine separates were also analysed for their Oisotope composition. Grains with compositional hetero-

geneity (i.e. zoning) where the variability in forsterite

content exceeds 60�5 were excluded from the O isotope

analysis. The isotopic composition of O in these olivines

ranges from d18O¼þ5�0 to þ5�7% V-SMOW, with the

full range of variability present in homogeneous high-

forsterite olivines (Fo88�7–90�2). Three separated olivinegrains have lower forsterite contents (Fo�83�6) than

those reported from thin-section EMPA work, but have

a corresponding d18O of about þ5�2%. Applying a typ-

ical fractionation factor between olivine and silicate

melt of �0�4% (e.g. Eiler, 2001), we find that melts in

equilibrium with the olivines should have d18O values ofþ5�4 to þ6�1%, in good agreement with the O isotope

data obtained for the glass samples (see below).

(a) (b)

(c) (d)

2 mm 2 mm

500 µm 500 µm

ZH-V1 ZH-V1

ZH-V1 ZH-L3

cpx+phlqtz

UCC xeno

olcpx+cpx+phlogphlog

qtz

UCCxeno

ol

phl

di

Ti-aug

ol

UCC xeno

cpx

qtz

Fig. 2. Representative thin-section photographs. (a) Sample V1 plane-polarized light and (b) cross-polarized light. The clinopyrox-enes–phlogopite cumulate is shown enlarged in (c) (plane-polarized light). (d) Olivine showing a dissolution texture and adjacentcrustal xenolith fragments; cross-polarized light.


at Erlangen N



Dow

nloaded from


ClinopyroxeneClinopyroxenes generally occur as microphenocrysts,

as glomerocrysts that show zonation (Fig. 3b) or around

olivine grains (Fig. 3e), or as crystal cumulates inter-

grown with phlogopite (e.g. in sample V1, Figs 2a–c and

3d) or olivine (e.g. sample U3, Fig. 3f). One cumulate ofclinopyroxene (size of single crystals about 2–3 mm)

and phlogopite found in sample V1 is about 7–8 mm in

diameter, but cumulates and megacrysts in volcanic

bombs of the nearby Mytina Maar can reach several

centimetres in size (Geissler, 2005; Geissler et al., 2007).

Spongy reaction textures are especially visible within

and around the rims of the Ti-augite rich clinopyrox-

enes (see spongy ‘fractures’ in Fig. 3d). Diopside occurs

as zones around Ti-augite and shows largely idiomor-

phic overgrowth textures (Fig. 3d).

(a) (b)

(c) (d)

(e) (f)

phl

ol

sp

qtz

Ti-aug

digl

ol

Ti-aug

cpxlaths

spongyreaction

zone

Fig. 3. Electron microprobe backscattered electron (BSE) images of key petrological features of the Zelezna hurka lavas. (a) Olivinecrystals with forsterite-rich cores (Fo�90) and slightly more Fe-rich margins (Fo�86). (b) Zoned glomerophyric clinopyroxene. (c)Upper crustal xenolith composed of quartz (dark grey), potassic feldspar (grey) and accessory muscovite. (d) Contact between asingle (disaggregated) quartz crystal and a cumulate composed of intergrown phlogopite (not shown) and clinopyroxene (Ti-augitewith diopsidic overgrowth). The presence of mingled interstitial glass (dotted area) and the spongy reaction zone in the clinopyrox-ene cumulate should be noted. (e) Rounded phlogopite hosted in olivine. The clinopyroxene laths oriented along the edge of theolivine crystal should be noted. (f) Cumulate of clinopyroxene (Ti-augite with diopsidic overgrowth rim) and olivine.


at Erlangen N



Dow

nloaded from


Cores of clinopyroxenes in cumulates are augitic andhave high Mg# (82–90), low TiO2 (<1�8 wt %) and Al con-

tents (<0�33 a.p.f.u.; Fig. 5a) but high Cr# (up to 18) and

high Na2O (>0�7 wt %; Fig. 5b). However, green-core

clinopyroxenes, as reported from the Eifel (Duda &

Schmincke, 1985) and basanites from Slovakia (Dobosi

& Fodor, 1992) have not been observed. Clinopyroxene

overgrowth rims as well as microphenocrysts in thematrix have a greater wollastonite and ferrosilite com-

ponent (Fig. 5c). Moreover, overgrowth rims and

phenocrysts form trends towards compositions con-

trasting with those of clinopyroxene cores in cumulates

(lower Na2O and Mg#, higher TiO2 and wollastonite

contents). Aluminium substitutes for Si, leading to anegative correlation of Al (a.p.f.u.) and Si contents. The

fine laths of clinopyroxene (of a few hundred

micrometres) show rhythmic zonation on backscatteredelectron images. However, a clear evolution towards a

defined mineral composition is not observed and the

mineral composition ranges from augite with variable

contents of Ti to diopside with up to 5�7 wt % TiO2. In

terms of trace elements, cores show normal convex

rare earth element (REE) patterns with the bulge cen-

tred at the middle REE (MREE; Supplementary Data(SD) Fig. S1a), similar to mantle clinopyroxenes in

xenoliths from the Oberpfalz (Ackerman et al., 2013).

Clinopyroxene grains in the clinopyroxene–phlogopite

cumulate of sample V1 show a simple zonation from

core (augite: Wo43En48Fs9) to rim (diopside:

Wo50En42Fs8) but diopside overgrowth rims and thespongy contact zones between clinopyroxene and

phlogopite show a less weak bulge mainly owing to

Fig. 4. (a) Forsterite vs Ni (ppm) content in olivine (grouped into analyses of cores, ‘intermediate’, rim and groundmass) analysed inthis study compared with (b) literature data [literature data also plotted in (a) as grey symbols]. The ‘intermediate’ group definesspot analyses between the clearly defined core and overgrowth rim to test for any hidden chemical zonation. (c) and (d) show for-sterite content vs CaO (wt %) concentration in olivines from this study and literature data, respectively. The inset in (c) shows ahistogram of olivine compositions analysed in this study. Blue lines indicate the evolution of olivine composition during fractionalcrystallization. We used partition coefficients from Beattie (1994) for Fe (0�51–1�55), Mg (1�96–4�40) and Ca (0�0192–0�0375), andfrom Seifert et al. (1988) for Ni (3�8–6�0). We chose the high end of the range for all partition coefficients except for Fe, for which weused a partition coefficient of 1�32 to match an Mg–Fe exchange coefficient of 0�3, typical for a wide range of basaltic liquids(Roeder & Emslie, 1970). We selected a mafic dyke from the Ohre Rift as the starting liquid composition (olivine melanephelinitefrom the Spojil Dyke; Vaneckova et al., 1993), adopted to fit the composition of early crystallizing olivine. We added 37 ppm Ni(þ9�2%) but halved the concentration of CaO (–50%) to match the starting composition with the composition of the most primitiveolivine crystallized. The contents of relevant elements or oxides in the starting liquid are 10�35 wt % FeOt, 16�78 wt % MgO, 5�04 wt% CaO and 440 ppm Ni. The composition of the first olivine crystallizing in our model is marked by a blue star. The evolutionarystages [1, 2 and 3 in (a) and (b)], as discussed in the text, should be noted. Literature data include OPF (Oberpfalz) peridotites fromAckerman et al. (2013) and various types of olivine recovered from the Mytina Maar and Zelezna hurka (Geissler, 2005). Greenshaded field indicates the range of mantle olivine.


at Erlangen N



Dow

nloaded from



enrichment of the light REE (LREE) relative to MREE and

heavy REE (HREE; SD Fig. S1b). Spongy reaction tex-tures between Ti-augite and diopside (e.g. Fig. 3d) show

a chemical composition intermediate between the two

end-members (SD Fig. S1).

Accessory phasesPhlogopite is the main accessory mineral with an over-

all volume of the order of 1–2%; single crystals reach

sizes of up to 10 mm. Phlogopite crystals intergrownwith clinopyroxene (cumulates) are less rounded than

those that occur as inclusions in olivines (samples V2

and U3; Fig. 3b). These latter phlogopites are relatively

small (less than 300–400mm) and slightly more magnes-

ian (Mg# 87–88) than the large phlogopites in cumulates

(Mg# 83–84). In terms of BaO and TiO2 concentrations,

phlogopites in the Zelezna hurka lavas are relativelyprimitive, similar to mantle phlogopites and intermedi-

ate between those found in lamproites and carbonatites

(SD Fig. S2; Dunworth & Wilson, 1998).

Additional accessory minerals include hauyne and

opaque crystals of the spinel group. These spinels

sensu lato are common and occur predominantly as in-

clusions in olivine or in direct contact with olivine. Theyare generally Fe- and Ti-poor and display variable Cr

and Al concentrations, resulting in variable Cr#, ranging

from 37 to 58. Only very few spinels (often in contact

with clinopyroxenes and lower in Mg#) have Fe- and Ti-

rich compositions and are solid solutions of Ti-magnet-

ite. However, the majority of spinels belong to thegroup of Mg–Al-chromites (solid solutions of spinel

sensu stricto, hercynite, chromite and Mg–Al-chromite).

Hauyne is present as small (<100mm), idomorphic crys-

tals, typically associated with clinopyroxene

phenocrysts.

Crustal xenolithsMicroscopically, we distinguish two xenolith groups: a

felsic variety containing quartz and feldspar and a

brownish, porous variety with an undefined mineral as-

semblage. In addition to quartz and potassic feldspar,

the felsic xenoliths (Fig. 3c) also contain albite and mus-covite. Modal mineral contents cover a broad range

from almost pure quartz to almost pure potassic feld-

spar. These felsic xenoliths are interpreted as frag-

mented parts of the regional host-rocks, which are

dominated by phyllites, quartzites and mica shists

(Geissler et al., 2007). Some of these xenolithic rockfragments seem to have disintegrated almost com-

pletely leading to isolated, subrounded quartz crystals

in the volcanic matrix. In sample V1, the assimilation–

melting reaction between a quartz crystal and a clino-

pyroxene–phlogopite cumulate has been preserved in

the form of an interstitial, mingled glass (Fig. 3d). The

mineral assemblage of the brownish xenoliths couldnot be resolved with certainty but probably includes

amphibole and may represent altered parts of the phyl-

litic host-rocks.

Geochemistry of volcanic rocks and glassesMajor element compositionThe compositions of the Zelezna hurka samples arerelatively uniform (Fig. 6). However, the major element

contents of glasses from the lower and the upper tephra

differ significantly from those of the whole-rock sam-

ples. In particular, the glasses have much lower MgO

contents at a given SiO2 compared with the whole-rock

samples. Furthermore, the FeOt (Fig. 6c) contents of theglasses are slightly lower, whereas the concentrations

of the other elements are higher than those of the

Enstatite

Ferrosillite

Wol

last

onite

Diopside

Hedenbergite

30

40

50

60

100

90 80 70 60

403020100

Augite

65 70 75 80 85 90 95

Mg#

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Al [

a.p

.f.u

.]

CoresZonationRimsMatrixCumulates

65 70 75 80 85 90 95

Mg#

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Na 2

O [

wt.

%]

(c)

(a)

(b)

cumulates(main group)

Fig. 5. (a) Aluminium (a.p.f.u.) and (b) Na2O (wt %) vs Mg# inclinopyroxene. A noteworthy feature is the difference betweencumulates (Ti-augite) and diopsidic phenocrysts and over-growth rims, also obvious in the proportional changes in min-eral components (c).


at Erlangen N



Dow

nloaded from





whole-rock samples (Fig. 6). Whole-rock samples strad-

dle the boundary between basanites and foidites in a

volatile-free total alkalis (Na2OþK2O) versus silica(SiO2) diagram (Fig. 7a, TAS diagram; Le Bas &

Streckeisen, 1991). However, the volcanic glasses

show significantly higher concentrations of alkalis

(>8 wt %) than the whole-rock samples. Thus, glasses

from Zelezna hurka can be classified as strongly

SiO2-undersaturated foidites. Melilite or nepheline have

not been observed in thin section and based on a

SiO2þAl2O3 (�54–55 wt %) versus CaOþNa2OþK2O(23�0–23�5 wt %) discriminant diagram (Le Bas, 1989) we

conclude that the lavas are part of the nephelinite

rock series, even though glass samples plot on the

boundary between the nephelinite and melilitite rock

series (Fig. 7b). We further note that the glasses contain

Fig. 6. Variation of MgO vs (a) TiO2, (b) Al2O3, (c) FeOt, (d) CaO, (e) Na2O and (f) K2O (all in wt %) ZH, Zelezna hurka samples, fromthis study and Ulrych et al. (2013). Data sources: central Ohre Rift, rift shoulder, Oberpfalz and Komornı hurka from Ulrych et al.(2013) and Haase & Renno (2008) and references therein; Mytina Maar from Geissler et al. (2007); Ohre Rift melilitites from Ulrychet al. (2008).


at Erlangen N



Dow

nloaded from


high concentrations of volatile elements such as S andCl (Table 1). The Cl/Nb ratios of about 17–20 are higher

than in mid-ocean ridge basalt (MORB) and ocean is-

land basalt (OIB) but similar to those of other continen-

tal rift regions (e.g. Rowe et al., 2015).

Trace element geochemistryThe whole-rock samples show primitive mantle-normal-

ized (Lyubetskaya & Korenaga, 2007) incompatibleelement patterns that are enriched in highly incompat-

ible trace elements and the LREE relative to HREE [e.g.

(La/Yb)N¼ 28�5–33�5, (Gd/Yb)N¼ 4�0–4�1 and (La/

Sm)N¼ 4�2–5�0] (Fig. 8a)]. The most prominent anomaly

in the multi-element pattern is a negative Pb anomaly

and there are slightly negative K and Ti anomalies. TheHFSE Nb and Ta are slightly enriched (Fig. 8a) whereas

Zr and Hf are slightly depleted. Overall, the whole-rock

samples are well within the range reported for the Ohre

Rift (e.g. Haase & Renno, 2008; Ulrych et al., 2013) but

with relatively high Rb, Ba, Th, U, Nb and Ta and low

MREE to HREE resulting in an overall slightly steeper

slope in a multielement diagram (Fig. 8a) comparedwith most other samples from the Ohre Rift. In contrast,

the trace element compositions of glasses extend the

range of published Ohre Rift data (Fig. 8a). Slight but

significant differences in trace element composition

exist between the glasses and whole-rocks (e.g. more

pronounced negative anomalies of Zr and Hf), resulting

in a strong contrast in some trace element ratios, suchas Nb/Zr (Fig. 8b). Generally, the whole-rock and glass

samples from Zelezna hurka have highly enriched trace

element ratios relative to the primitive mantle of

Lyubetskaya & Korenaga (2007); for example, Nb/Zr is

6–13�, Ta/Hf is 8–18�, La/Yb is 29–36� (Fig. 8c) and Nb/

U is 1�5–2�0� higher relative to the respective ratios ofthe primitive mantle.

Isotope geochemistryIn general, samples from the Ohre Rift and the CEVPspan a large range in their Nd–Sr isotopic composition,

with the Ohre Rift samples showing higher 87Sr/86Sr

at a given 143Nd/144Nd relative to other CEVP samples

(Fig. 9a). Whole-rock samples from Zelezna hurka are

relatively homogeneous in their Sr–Nd–Hf isotope com-

positions but slight differences are evident in Sr–Nd iso-

tope space amongst glass samples from the lowertephra unit (Fig. 9a). Their compositions overlap in eNd–

eHf (Fig. 9b) and fall within the broader mantle array of

Vervoort et al. (1999), consistent with data from the

CEVP. Two glass samples have significantly lower143Nd/144Nd and higher 87Sr/86Sr compared with other

Quaternary volcanic rocks from the Ohre Rift (e.g.Haase & Renno, 2008; Ulrych et al., 2013).

The O isotope compositions of glasses are between

d18O þ5�4 and þ5�8% V-SMOW, with the exception of

sample L1, which has a value of þ6�2% (þ6�13 and

þ6�25%; one duplicate analysis). The high d18O of sam-

ple L1 is associated with higher MgO, CaO, Cl/K, Ce/Pb

and 87Sr/86Sr, but lower Nb/Zr, La/Sm, K/Ti and177Hf/176Hf values.

DISCUSSION

Insights from mineral phasesOlivine antecrystsThe compositions of olivine crystals reveal insights into

the plumbing system of Zelezna hurka. Cores of olivinecrystals show normal mineral zonation (i.e. decrease in

forsterite content towards the rim) and plot within the

field of olivines in equilibrium with melts from ‘ordin-

ary’ mantle peridotite rather than melts from pyroxe-

nitic lithologies (Straub et al., 2011). However, olivine

cores are distinct from primary mantle olivines found inmantle peridotite xenoliths of the same region (Fig. 4c;

lower Fo and Ni, higher CaO). Primary mantle olivines

Fig. 7. (a) Volatile-free total alkalis, Na2OþK2O (wt %) vs SiO2

(wt %) after Le Bas & Streckeisen (1991). Whole-rock samplesfrom Zelezna hurka overlap with the range of basanitic lavasreported from the Oberpfalz, but glasses are much higher intheir alkali content at a given SiO2 content. (b) Combined vola-tile-free oxide diagram, SiO2þAl2O3 (wt %) vsCaOþNa2OþK2O (wt %), for the discrimination between bas-anite, nephelinite and melilitite after Le Bas (1989). Data sour-ces as in Fig. 6.


at Erlangen N



Dow

nloaded from


usually have CaO contents of less than 0�1 wt % and

characteristic Ni concentrations of 2600–3200 ppm,whereas magmatic olivines have CaO concentrations

>0�18 wt % (Stamper et al., 2014). Olivines entrained in

the Zelezna hurka lavas are thus of magmatic origin.

However, they are in disequilibrium with the host lava,

which is evident both petrographically (common dissol-

ution textures) and chemically (Mg# of melt much lower

than expected if assuming olivine–melt equilibrium).These olivines are thus antecrysts and their evolu-

tion can be differentiated into three generations:

(re-)crystallization (stage 1) in equilibrium with mantle

peridotite (Fo> 89, Ni> 1700 ppm; Straub et al., 2011),

as demonstrated by the modelled crystallization trend

shown in Fig. 4 (see figure caption for model details)and SD Table S2. Minor differences between the calcu-

lated fractionation trend and observed mineral

compositions during stage 1 (Fig. 4a) may be explained

by minor changes in the olivine–melt partition coeffi-cient for Ni as a result of changes in liquid composition

(e.g. concomitant crystallization of magnetite), pressure

and temperature (e.g. Matzen et al., 2013). Further crys-

tallization of olivine along the predicted crystallization

path from forsterite contents of 89 to 84 (Stage 2 in

Fig. 4a) is accompanied by a drop in Ni content from

>2000 ppm to <1000 ppm (Fig. 4a) but a minor increasein CaO (Fig. 4c). In Stage 3, groundmass crystals and

rims of olivine crystals show a significant drop in Ni

at constant forsterite content (500 ppm Ni and below;

Fig. 4a) but a strong enrichment in CaO (up to >1�0 wt

%; Fig. 4c) and MnO. These low-Ni–high-CaO rims of

olivine antecrysts have not been reported from Zeleznahurka in previous studies (e.g. Geissler, 2005). A very

similar pattern in mineral evolution (although at a

Fig. 8. (a) Multi-element plot for whole-rock and glass samples normalized to the primitive mantle values of Lyubetskaya &Korenaga (2007). Red, glass data; orange, whole-rock data. The grey field corresponds to literature data with two typical patternsshown as grey lines. (b) Nb/Zr vs MgO (wt %) and (c) chondrite-normalized (Palme & O’Neill, 2003) (La/Sm)N vs La (ppm). Data sour-ces as in Fig. 6.


at Erlangen N



Dow

nloaded from



smaller magnitude and at overall more primitive com-

positions) is recorded in fresh olivines of the

Udachnaya East kimberlite in Yakutia (Kamenetsky

et al., 2008). Furthermore, the composition of the olivineantecrysts cannot be explained by crystallization from a

single parental liquid, but instead requires at least one

other liquid with a very different composition (e.g. high

Ca and Mn, low Ni).

Compositional profiles from the rim of olivine crys-

tals towards the core (SD Fig. S3) allow tracking of thedifferent steps of crystal evolution (e.g. composition of

parental melt, solid-state crystal diffusion, presence or

Fig. 9. (a) 87Sr/86Sr vs 143Nd/144Nd for volcanic rocks from the CEVP (Hocheifel: Jung et al., 2006; Rohn: Jung et al., 2013;Vogelsberg: Jung & Masberg, 1998) compared with data from the Ohre Rift region (Ohre Rift: Haase & Renno, 2008, and referencestherein; Ulrych et al., 2013; Ohre Rift melilitites: Ulrych et al., 2008; Oberpfalz mantle xenoliths: Ackerman et al., 2013; Komornıhurka: Haase & Renno, 2008; Zelezna hurka glasses and whole-rocks: this study). Also shown are the approximate positions ofmantle endmembers PREMA, DMM, EM1 and EM2 (Stracke, 2012), LVC (Hoernle et al., 1995) and EAR (Cebria & Wilson, 1995; asdefined by Lustrino & Wilson, 2007). (b) (eNd)i vs (eHf)i of volcanic rocks from the CEVP [Vogelsberg (17 Ma), Rhon (24 Ma),Hocheifel (40 Ma): Jung & Masberg, 1998; Jung et al., 2011; Pfander et al., 2012], Zelezna hurka glasses and whole-rocks (this study)and the Udachnaya East kimberlite (Kamenetsky et al., 2009b). Fields for MORB, HIMU, EM1 and EM2 (after Pfander et al., 2007) areshown for comparison. The mantle array is after Vervoort et al. (1999; eHf¼1�33 eNdþ3�19) and data for CHUR are from Bouvieret al. (2008). Decay constants for 147Sm and 176Lu are from Begemann et al. (2001) and Scherer et al. (2001), respectively.


at Erlangen N



Dow

nloaded from


absence of crystal–melt equilibrium). Diffusion of cal-

cium in olivine is significantly slower than that of Fe or

Mg, whereas Ni shows similar diffusion rates to Fe and

Mg (e.g. Jurewicz & Watson, 1988; Petry et al., 2004).

The forsterite content changes in a narrow zone be-tween roughly 50–100mm from the rim from about 90 in

the core to 86�0–86�5 at the rim, but the overgrowth rim

itself displays a constant forsterite content. The Ni con-

centration, in contrast, shows the most significant de-

crease, concurrent with the change in forsterite, but

also decreases across the overgrowth rim where the

forsterite content is constant (SD Fig. S3a and b). TheCa profile (SD Fig. S3c) shows a different pattern, with

almost constant concentrations over a wide range, even

over most of the overgrowth rim and the zone of for-

sterite change. We therefore conclude that the high-Fo

olivine cores represent olivines crystallized in the (litho-

spheric) mantle, whereas their rims were diffusivelymodified after entrainment in the parental melt of the

Zelezna hurka nephelinite.

To constrain further the composition of the parental

magma, we empirically calculated the (theoretical) par-

tition coefficient of Ca between olivine and host-

magma, assuming that a diffusion profile reflects anattempt to reach a chemical equilibirium at the crystal–

melt interface. Applying the method of Libourel (1999),

we see that the partitioning for CaO between olivine

and melt, if calculated based on olivine composition

only [equation 14 of Libourel (1999)], is much lower

than if calculated from the molar fraction of CaO in

olivine and a*meltCaO [equations 12 and 13 of Libourel(1999); note that this partition coefficient is of a hypo-

thetical nature as olivines are in fact in disequilibrium

with the host lava]. This implies that, to explain the

high concentration of CaO in the olivine rims, a liquid

is required that contained much higher concentrations

of CaO. Based on experimental data (e.g. Libourel,1999, fig. 5), we can estimate a minimum concentra-

tion of �20 wt % CaO in the parental liquid to explain

the high enrichment of Ca (5000 ppm or more; SD Fig.

S3c) in the olivine rims at intermediate alkali contents

(4–8 wt %). It is important to note that the total alkali

content (Na2OþK2O) has a much stronger influence

on the partitioning of Ca into olivine than oxygen fuga-city or temperature (Jurewicz & Watson, 1988;

Libourel, 1999).

Clinopyroxene cumulates and phenocrystsClinopyroxenes can be separated into an augitic cumu-

late domain (cores) and a more evolved and LREE-en-riched phenocryst (and overgrowth rim) domain

(diopside). These clinopyroxene overgrowth rims and

phenocrysts formed at a later stage of magmatic differ-

entiation compared with the olivines, as indicated by (1)

the lower Mg# (68–84) compared with olivines (86–90�5)

and (2) the growth of clinopyroxene phenocrysts at themargins of pre-existing olivines (e.g. Fig. 3e). These

clinopyroxenes show an increase in wollastonite and

ferrosilite components with increasing differentiation

(Fig. 5c), implying that Mg–Fe exchange follows a nor-

mal differentiation trend whereas the Ca contents do

not. Consistent with our previous observation of Ca en-

richment in olivine overgrowth rims, the Ca-rich com-position of clinopyroxene overgrowth rims and

phenocrysts may be seen as another hint of a highly

Ca-enriched parental melt. The Ti enrichment in clino-

pyroxene is equal to the enrichment observed in SW

German melilitites by Dunworth & Wilson (1998) and

has been considered to be close to the limits of Ti4þ

substitution.Clinopyroxenes are an important carrier for trace

elements and their distinctive trace element patterns

allow us to infer magmatic processes during crystalliza-

tion. We previously described the similarity in REE pat-

terns between clinopyroxenes in the lavas of Zelezna

hurka and mantle xenoliths of the Oberpfalz (SD Fig.S1). Even more interesting are their similar trace elem-

ent patterns compared with clinopyroxenes hosted as

inclusions in fresh olivines from the Udachnaya East

kimberlite (Fig. 10). Both clinopyroxene domains show

a relative enrichment in the LREE to MREE and a nega-

tive Zr anomaly. However, there are also some differ-ences compared with the Udachnaya clinopyroxenes,

which are more enriched especially in the LREE but

more depleted in the HREE relative to Zelezna hurka

(Fig. 10). This HREE depletion could result from the pref-

erential partitioning of these elements into garnet at

pressures >2�5 GPa (in the garnet stability field), which

is also supported by their high Na and Cr concentra-tions, indicating even higher pressures of >4�5 GPa

(Kamenetsky et al., 2009a). The most prominent con-

trast, however, is the strong negative Ti anomaly in the

kimberlitic clinopyroxene inclusions, which could result

from direct substitution of Ti for Si in the olivine host

(Hermann et al., 2005) or the preferential partitioning ofTi into ilmenite, present in the kimberlite but absent in

Zelezna hurka lavas.

Accessory phasesZelezna hurka phlogopites probably formed in equilib-

rium with mantle peridotite, reflected in their high Mg#,low Ba and intermediate Ti concentrations. The phlogo-

pite inclusions in olivine may have formed by reaction

of pre-existing melt (in inclusions in olivine) during the

entrainment of olivines in a Ba-enriched residual melt

(Seifert & Kampf, 1994) or by reaction of olivine with

CO2 to form phlogopite and carbonate melt (Mysen &

Virgo, 1980). In contrast, phlogopite in cumulates hasslightly lower Mg# and higher Ba and Ti contents. The

overall enrichement in these elements may be an effect

of shallow magmatic processes (Seifert & Kampf, 1994).

All phlogopites from Zelezna hurka are intermediate

between the distinct melilitite and leucite–nephelinite

trends of Dunworth & Wilson (1998) but similarto phlogopite in the mantle and kimberlites (see SD

Fig. S2).


at Erlangen N



Dow

nloaded from











Spinel sensu lato shows a broad negative correlation

between Mg# and stoichiometrically calculated Fe3þ

content, indicating a trend towards more oxidizing con-

ditions. The corresponding mineralogical change from

Mg–Al-chromite towards Ti-magnetite is commonly

observed in carbonate-rich magmas (e.g. Jones &Wyllie, 1985). Further evidence for highly oxidizing con-

ditions in the melt is the presence of hauyne, a sodalite-

group mineral that contains oxidized sulphur as a major

phase (�5 wt % S). The association of titanomagnetite

and hauyne with clinopyroxene phenocrysts indicates

that high oxygen fugacities were reached late during

magmatic differentiation.

Crustal assimilationPhysical evidence for interaction between the ascending

magmas and the continental crust is provided by the

presence of quartz crystals in the matrix and crustalxenoliths. This interaction has a high potential to alter

the chemical composition of the erupted lavas and thus

to obscure the nature of the parental melt. A spectacular

snapshot of crustal assimilation is preserved in sample

V1 (Fig. 3d). Here, the composition of interstitial glass

that formed at the contact of a (crustal) quartz crystal and

a clinopyroxene–phlogopite-cumulate shows a strongchemical contrast to the otherwise homogeneous matrix

glass. The interstitial glass is mingled (Fig. 3d) and very

SiO2-rich (c. 69 wt %; Table 2) as opposed to the strongly

silica-undersaturated (�40 wt % SiO2) matrix glass

(Table 1). With respect to the low solidus temperature of

quartz in the presence of phlogopite and H2O–CO2 va-pour (�700–800�C at crustal depths; e.g. Wones &

Dodge, 1977; Bohlen et al., 1983) and the presence of

numerous dispersed single quartz crystals (likely to rep-

resent disintegrated crustal xenoliths) in the lavas, as-

similation of crustal material may play an important role

in the petrogenesis of the Zelezna hurka lavas. However,

because sample V1 was taken from a blocky lava, in

which temperatures may remain at higher levels for alonger period of time relative to the explosively erupted

tephra, we need to further constrain the possible role of

crustal assimilation using geochemistry.

Assimilation of crustal material may effectively

change elemental concentrations and ratios as well as

isotope ratios [e.g. decreasing Ce/Pb or Nb/U in con-

junction with increasing SiO2, 87Sr/86Sr or d18O ofwhole-rock or glass (subsequently noted as d18OWR; e.g.

Taylor, 1980; Jung & Hoernes, 2000; Jung et al., 2013)]

and has been found to play a key role in the magmatic

evolution of several suites of the CEVP. Lavas from the

Rhon, for example, show a broad positive correlation

between d18OWR and SiO2 (Fig. 11a), interpreted as theresult of combined assimilation and fractional crystal-

lization processes (Jung et al., 2013). Similarly, one of

our new samples from Zelezna hurka shows higher

d18OWR, associated with higher 87Sr/86Sr (0�7039),

whereas all other samples show mantle-like d18OWR

[mantle range of Taylor (1980) and Eiler et al. (2000)].

Quantification of the influence of continental crustalassimilation in the petrogenesis of the Zelezna hurka

lavas is difficult, in particular with respect to their iso-

tope characteristics. Further insights can be obtained

from the major element compositions of the crustal

xenocrysts and the general trace element charac-

teristics of the continental crust. We identified quartz,K-feldspar, albite and muscovite as the main phases in

the crustal xenoliths; amphibole may also be present.

LuRb

BaTh

UNb

TaLa

CePb

PrSr

NdZr

HfSm

EuTi

GdTb

DyY

HoEr

TmYb

100

10

1

0.001

0.01

0.1

Cpx phenocr. & rims

Cpx

/ Pr

imiti

ve M

antle

Cpx cores

Kimberlite cpx inclusions(Udachnaya East)

Fig. 10. Primitive mantle-normalized [values of Lyubetskaya & Korenaga (2007)] trace element patterns of clinopyroxenes fromZelezna hurka (cores of cumulates, and phenocrysts and overgrowth rims) and inclusions hosted in fresh olivines of theUdachnaya East kimberlite (Kamenetsky et al., 2009a). The similarity in the trace element patterns should be noted. However, theUdachnaya East kimberlite shows a more pronounced difference between the light (more enriched) and heavy (depleted relative tothe primitive mantle) REE.


at Erlangen N



Dow

nloaded from


Figure 12 shows the chemical composition of pheno-

crysts [clinopyroxene, phlogopite, hauyne and poten-

tially amphibole; found in cumulates by Geissler et al.(2007)], along with olivine antecrysts and crustal xeno-

crysts (quartz, potassic feldspar, albite, muscovite).

The first phase of melting in metapelitic crustal xeno-

liths (such as phyllite) involves albite-rich plagioclase

(oligoclase), controlled by the H2O and alkali releaseduring the breakdown of muscovite (Grapes, 1986).

This pattern is well reflected by the trend of the Ohre

Fig. 11. Stable oxygen isotope composition in per mil relative to V-SMOW. Grey shaded bands indicate the d18O range (in per milrelative to V-SMOW) of normal (N)-MORB glasses (Eiler et al., 2000) and the range of d18OWR for primitive melts containing 4–5 wt% Na2O (Eiler, 2001). (a) d18O of Zelezna hurka (ZH) glasses and whole-rocks (Rhon: Jung et al., 2013; Garrotxa, NE Spain: Cebriaet al., 2000) vs SiO2 (wt %). Most of the Zelezna hurka glasses (except for L1) plot in the d18OWR range of N-MORB (Eiler et al., 2000).Rhon samples have been affected by assimilation of crustal material and subsequent fractional crystallization (Jung et al., 2013);this is also visible in the broad correlation between d18OWR and SiO2 (trend encompassed by the dashed lines). (b) d18OWR vs CaO/Al2O3 of whole-rocks from the Rhon (Jung et al., 2013), Garrotxa (NE Spain; Cebria et al., 2000) and Zelezna hurka glasses. The aver-age compositions of upper (UCC) and lower continental crust (LCC) are shown for comparison (Rudnick & Gao, 2003). Rhon lavasare consistent with assimilation of continental crust whereas glasses from Zelezna hurka show a positive correlation between O iso-tope composition and CaO/Al2O3.


at Erlangen N



Dow

nloaded from


Fig. 12. Major oxide composition of Ohre Rift lavas (data sources as in Fig. 6; this study) compared with single magmatic andxenocrystic mineral phases and experimental melt compositions for KLB-1 [dry peridotite of Hirose & Kushiro (1993); run details inFig. 15], PERC-3 [carbonated peridotite of Dasgupta et al. (2007); run details in Fig. 15] and KC2 [carbonatite melt of Sweeney(1994)]. (a) CaO and (b) Na2O vs SiO2 and (c) K2O vs Al2O3 (all in wt %). The compositions of the Quaternary volcanic rocks of theOhre Rift are controlled by variable proportions of the primary magmatic minerals [Cpx, clinopyroxene; Hbl, hornblende (Geissleret al., 2007); Hy, hauyne; Phl, phlogopite; Ol, olivine] rather than by assimilation of crustal components (Ab, albite; Kfsp, K-feldsparMs, muscovite; Qz, quartz). It should be noted that fractional crystallization would result in trajectories away from the crystallizingmineral compositions, whereas assimilation trajectories (grey arrows) would point towards the assimilated mineral. Albite-richplagioclase and potassic feldspar exhibit a range of compositions (grey and brown crosses, respectively).


at Erlangen N



Dow

nloaded from


Rift lavas in Fig. 12, extending from a parental magmacomposition, probably similar to the PERC-3 compos-

ition of Dasgupta et al. (2007), towards variable mix-

tures of albite and muscovite (plus K-feldspar).

However, the composition of the Zelezna hurka lavas is

controlled by the accumulation of olivine (whole-rock)

and fractional crystallization of clinopyroxene (glasses)as evident from the trajectories between whole-rocks,

glasses, olivine antecrysts, clinopyroxene and other

magmatic phases [hauyne as phenocryst and phlogo-

pite (þ hornblende; Geissler et al., 2007) as cumulate

phases; Fig. 12]. All these samples consistently fall at

the low-silica end of the range of most Ohre Rift lavas,

with only melilitites (Ulrych et al., 2008) being moreundersaturated. The predominant crustal assimilation

of albite and mica (6 K-feldspar) could also explain the

presence of residual single quartz crystals. The trends

in Fig. 12 point to a parental melt composition of the

Zelezna hurka lavas with even lower SiO2 and Al2O3,

but higher CaO and total alkalis, if we assume that thehighly silica-undersaturated nephelinites of Zelezna

hurka have already assimilated significant amounts of

silica-rich crustal lithologies. To produce such parental

melt compositions from a peridotite source, significant

amounts of carbonate (i.e. CO2) must be involved (see

PERC-3 and KC2 in Fig. 12 and discussion below).

In contrast to major elements, where large amountsof assimilated material are necessary to significantly

change the whole-rock chemical composition, trace

elements (and their ratios) are more sensitive to assimi-

lation. Trace element ratios such as Nb/U or Ce/Pb have

proven to be powerful tracers, with respective values of

47 6 10 and 25 6 5 in oceanic basalts (Hofmann et al.,1986) and 4.4 and 3.7 in the continental crust (Rudnick &

Gao, 2003; Fig. 13). Zelezna hurka lavas have Nb/U

similar to the range observed in oceanic basalts but

with Ce/Pb one to four times higher (Fig. 13).

Furthermore, O isotopes show a narrow range between

5�4 and 6�3% V-SMOW and are positively correlated

with CaO/Al2O3 in the glasses (Fig. 11b), contrary towhat is expected for the assimilation of upper continen-

tal (granitoid) crustal material as observed in the Rhon

province [Fig. 11b; tephrites, phonolites and trachytes

of Jung et al. (2013)].

Precise information on the composition of the paren-

tal melt (which is likely to have higher Ce/Pb) remains

obscure and thus assimilation of crustal material can beneither excluded nor confirmed and quantified with cer-

tainty. However, the Zelezna hurka lavas have much

higher trace element abundances than average upper

continental crust (Rudnick & Gao, 2003). The assimila-

tion of crustal material would thus lead to a relative de-

pletion of trace elements in the melt. Assuming amaximum of 10 vol. % of upper continental crust being

assimilated in the lavas erupted would not change their

overall incompatible trace element patterns signifi-

cantly. It is not sufficient to explain the high enrichment

in Nb, Ta, LREE to MREE and Sr along with a relative de-

pletion in Pb, Zr and Hf (Fig. 8a). These trace elementcharacteristics are thus assumed to be of primary mag-

matic origin (melting and/or source). Before we further

constrain these, we will first try to reconstruct the

plumbing system of Zelezna hurka.

Constraints on the magma plumbing systemThermobarometryThe composition of clinopyroxene phenocrysts can be

used to determine the thermobarometric conditions of

crystallization using equations 32c, 33 and 34 of Putirka

(2008; Excel spreadheets are available for download

from the website of K. Putirka: http://www.fresnostate.

edu/csm/ees/faculty-staff/putirka.html). Successful P–Testimates were assumed if (1) the calculated Fe–Mg ex-

change coefficient is within the range of experimental

observations (0�28 6 0�08; Putirka, 2008) and (2) the

clinopyroxene components [diopside–hedenbergite

and enstatite–ferrosilite, calculated using the normative

procedure of Putirka et al. (2003)] from measured min-eral compositions are in agreement (6 0�01) with the ex-

pected clinopyroxene components calculated from the

host-rock composition. This resulted in 38 successful P–

T estimates (Fig. 14; SD Table S3) that fall into two

groups. The first group consists of Na- and Cr-rich aug-

ites that occur as cores of cumulate crystals (Fig. 3d)

and indicate pressures of around 1�0–1�3 GPa at tem-peratures of 1250–1300�C (Fig. 14). The second clinopyr-

oxene group (diopsidic rims) displays lower pressures

and temperatures at around 0�8 GPa and 1150–1200�C,

respectively. These two groups are distinct from each

other even given the method’s relatively large uncer-

tainty of 60�15 GPa and 650�C (Putirka, 2008). We con-clude that cumulates formed in the upper SCLM,

whereas phenocrysts and overgrowth rims crystallized

47±10

25±5

UCC

0 20 40 60 80 100 120

Nb/U

0

20

40

60

80

100

120C

e/P

b

Assimilation of upper continental crust

Oh e (Eger) RiftOh e (Eger) Rift: melilititesOberpfalzKomorní h rkaM tina MaarZH whole rocksZH matrix glassesSWG melilitites

Fig. 13. Nb/U vs Ce/Pb for Ohre Rift lavas (data sources as inFig. 6; this study) and the SW German (SWG) melilitites ofHegner et al. (1995) and Dunworth & Wilson (1998). All theQuaternary lavas studied here fall within the oceanic array forNb/U (47 6 10; Hofmann et al., 1986) but are more enriched inCe/Pb. The average upper continental crust (UCC) has Nb/Uand Ce/Pb of 4�4 and 3�7, respectively (Rudnick & Gao, 2003).Assimilation trajectories are indicated.


at Erlangen N



Dow

nloaded from

http://www.fresnostate.edu/csm/ees/faculty-staff/putirka.html

http://www.fresnostate.edu/csm/ees/faculty-staff/putirka.html



in the lower continental crust (Fig. 14; Moho <28 km;

Babuska & Plomerova, 2010).

The temperature estimates of Geissler et al. (2007)

yield lower values for their xenolith suite (including

hornblendite, pyroxenite, wehrlite) compared with our

cpx–melt temperatures at similar pressures (Fig. 14).Geissler et al. (2007), however, speculated also about

the possible intrusive origin of these rocks or existing

mineral disequilibria in these rocks. In contrast, a dis-

tinct suite of cumulates (e.g. ol–cpx–spl cumulates)

studied by Geissler et al. (2007) (red open diamonds in

Fig. 14) points towards the magmatic temperatures re-

corded in our samples. This could possibly indicatelater reheating of those cumulates after their entrain-

ment in the Zelezna hurka host nephelinite (our data);

further studies are required to resolve the complex ther-

mobarometric history.

Volcanism in the West-Eifel and Hocheifel regions of

Germany shows close similarities to the volcanismin the Ohre Rift region (e.g. SiO2-undersaturated, high-

alkali volcanic rocks such as foidites, basanites and

nephelinites; e.g. Duda & Schmincke, 1985; Jung et al.,

2006) and is controlled by a combination of intraplate

magmatic activity and lithospheric extension (propaga-

tion of pre-rift volcanism linked to the Upper Rhine

Graben; Fekiacova et al., 2007). Depths of crystallizationare similar to our estimates for the nephelinitic rocks of

Zelezna hurka with respect to crystallization below the

Moho and within the lower crust (e.g. Duda &

Schmincke, 1985). However, Sachs & Hansteen (2000)

calculated the depth of a possible magma chamber at

0�6–0�7 GPa, which is at the low end of the pressures

calculated for the crystallization of clinopyroxene in theZelezna hurka lavas. The range of pressures recorded

by clinopyroxene overgrowth rims and phenocrysts at

Zelezna hurka (0�7–1�0 GPa; Fig. 14) argues against the

presence of a lower crustal magma chamber under-

neath this volcano but instead for continuous crystal-

lization during melt ascent.

Constraints on the parental melt compositionSo far, we have demonstrated that there is strong min-

eralogical evidence (in terms of mineral assemblage,

composition and evolution) for a genetic link betweenthe Quaternary nephelinites of Zelezna hurka and other

silica-undersaturated magmas, such as melilitites and

even kimberlites. Olivine antecrysts show very similar

chemical evolution, clinopyroxenes have similar trace

element patterns, phlogopites are intermediate be-

tween those found in carbonatites and lamproites, andhauyne and titanomagnetite indicate a shallow depth of

oxidation of the melt (e.g. during magma ascent).

Further evidence for the major role of carbonate in the

genesis of the parental magmas comes from the major

element composition of these rocks, which makes them

less sensitive to secondary processes such as crustal

assimilation.We use a simple Rayleigh fractionation model

involving olivine and an olivineþ clinopyroxene assem-

blage (Fig. 15) and compiled literature data from melt-

ing experiments of various source lithologies (e.g. dry

peridotite versus carbonated peridotite) close to our

thermobarometric estimates. For dry peridotite we usedthe melt composition generated by melting KLB-1 at

1�5 GPa and 1350�C (Hirose & Kushiro, 1993).

Experiments on carbonated peridotite are generally per-

formed at much higher pressures than those on dry

peridotite; thus we selected the melt composition gen-

erated by melting a peridotite carbonated with 1�0 wt %

CO2 (PERC-3) at 3�0 GPa and 1350�C [run A509 ofDasgupta et al. (2007)]. For each of these starting pri-

mary magma compositions we then calculated fraction-

ation paths for two scenarios: (1) simultaneous

crystallization of clinopyroxene (composition of L3-12

clinopyroxene phenocrysts) and olivine (composition of

L3-12 antecryst) in the relative proportion 4:1; (2) a two-step model involving first 20 vol. % fractionation of oliv-

ine only, followed by simultaneous crystallization of a

1SD

Focal depthof recent

earthquakeswarms

MOHO

SCLM

CC

Eifelmagma

chamber

incipient melting

major m

eltingalkali basalt

geotherm

Xenolithre-equilibrium

Re-

heat

ing/

entr

ainm

ent?

800 1000 1200 1400

Temperature [°C]

0

0.5

1.0

1.5

2.0

Pre

ssu

re [

GP

a]

10

20

30

40

50

60

0

CoresZonationRimsMatrixCumulates

Geissler et al. (2007)

CumulatesXenoliths

Dep

th [km

]

Fig. 14. Thermobarometric estimates for clinopyroxene crystal-lization using the method of Putirka (2008). (For full details seethe main text and SD Table S2.) The onset of clinopyroxenecrystallization is at about 1�2 GPa in the SCLM. However, mostclinopyroxenes indicate pressures of crystallization between0�7 and 1�0 GPa, close to the Moho (Babuska & Plomerova,2010) or within the continental crust. Focal depths of recentearthquake swarms (e.g. Horalek et al., 2000; http://www.ig.-cas.cz/en/structure/observatories/west-bohemia-seismic-net-work-webnet) and the inferred depth of a crustal magmachamber beneath the Eifel (Sachs & Hansteen, 2000) areshown for comparison. Geotherms, melting paths and add-itional P–T data are from Geissler et al. (2007).


at Erlangen N



Dow

nloaded from


http://www.ig.cas.cz/en/structure/observatories/west-bohemia-seismic-network-webnet




combined olivine–clinopyroxene crystal assemblage

similar to the first scenario (cpx:ol¼ 4:1).Fractionation of plagioclase or an olivineþplagio-

clase assemblage would result in a slight decrease in

CaO/Al2O3 and is not evident from compositional or

petrographic observations. Exclusive fractionation of

olivine does not change CaO/Al2O3 significantly but is

effective in explaining the relative compositional differ-

ence between glasses and whole-rocks (involving about20 vol. % olivine accumulation). However, with the

onset of fractional crystallization of clinopyroxene, CaO/

Al2O3 changes dramatically. One of the major differ-

ences between melts of carbonate-bearing and carbon-

ate-free mineral assemblages is the large contrast in

CaO/Al2O3. In carbonate-free peridotite, CaO/Al2O3 in-creases with increasing degree of partial melting but is

always less than 1�0 owing to the excess of Al2O3 over

CaO (Fig. 15). However, during melting of carbonate-

bearing peridotite and in experiments performed at

constant pressure (PERC and PERC-3 at 3�0 GPa;

Dasgupta et al., 2007), CaO/Al2O3 values are controlled

by the changing CaO contents of the partial melts ratherthan by changes in Al2O3 concentrations. At low de-

grees of partial melting, CaO concentrations are very

high because of non-modal melting dominated by clino-

pyroxene (high CaO/Al2O3). Further melting at increas-

ing temperatures consumes clinopyroxene (‘cpx-out’)

with the result of decreasing CaO and increasing Al2O3

(melting of garnet) content in the melt (Dasgupta et al.,

2007). The experimental evidence for various factors

exerting control on CaO/Al2O3 values explored above is

thus consistent with production of the parental magmafor Zelezna hurka lavas as low-degree partial melts of

carbonated peridotite.

Mantle sources and meltingIsotopic constraintsA distinct mantle reservoir has been proposed as acommon source component of the Cenozoic mafic alka-

line magmatism in Europe [i.e. ‘component A’ of Wilson

& Downes (1992); ‘Low Velocity Component’ of Hoernle

et al. (1995); ‘European Asthenospheric Reservoir’ of

Cebria & Wilson (1995)]. However, an alternative view

explains this common reservoir by variable degrees ofmixing between at least three distinct mantle sources

(e.g. Haase & Renno, 2008). Notably, basalts from

Lower Silesia extend to more radiogenic 143Nd/144Nd

values (Blusztajn & Hart, 1989; Fig. 9a) than the pro-

posed European mantle reservoir, close to the prevalent

mantle (PREMA) composition of Stracke (2012), thus

putting into question whether there is a uniqueEuropean mantle reservoir (Fig. 9a). For Zelezna hurka,

Sr and Nd isotope compositions may also point to-

wards three-component mixing (e.g. PREMA, EM1,

EM2), as suggested by Haase & Renno (2008), but could

also be explained by variable amounts of crustal con-

tamination. Slightly elevated 87Sr/86Sr ratios would beconsistent with a limited amount of crustal assimilation,

as discussed above. However, the observed range in

0 5 10 15 20 25

MgO [wt.%]

0.0

0.5

1.0

1.5

2.0

CaO

/Al 2

O3

PERC@1350°CCaO/Al2O3

= 2.62PERC-3

50

20

10

52

20% Ol

Cpx:Ol (4

:1)

Cpx

:Ol (

4:1)

1 GPa, 1250°C

3 GPa, 1500°C

20% Ol

5050

50

increasing F

1

Oh e (Eger) Rift

OberpfalzKomorní h rkaM tina MaarZH whole rocksZH matrix glassesthis studyliterature data

KLB-1 1350°CHK-66Mix-1GPERC-3 1350°PERC-3PERC

OR melilitites

Fig. 15. Variation of MgO (wt %) vs CaO/Al2O3. Blue diamonds, melting experiments of a carbonated peridotite (PERC and PERC-3with 2�5 and 1�0 wt % CO2, respectively) performed by Dasgupta et al. (2007) at 3�0 GPa pressure and temperatures from 1300 to1600�C. Melting trends are indicated by blue arrows. Orange diamonds, experimental melt compositions (1�0–3�0 GPa, 1250–1500�C) from Hirose & Kushiro (1993) of a natural spinel peridotite (HK-66); green diamonds, results from melting experiments(1�0–2�5 GPa, 1375–1500�C) on a garnet pyroxenite (Mix-1 G) by Hirschman et al. (2003). Hexagons, starting melt compositions[blue, PERC-3 at 1350�C and 3�0 GPa; orange, KLB-1 spinel lherzolite at 1350�C and 1�5 GPa of Hirose & Kushiro (1993)] for a simpleRayleigh fractionation model involving olivine and an olivine–clinopyroxene assemblage. Mineral/liquid partition coefficients usedfor MgO, CaO and Al2O3 are mean values of the experimentally determined range for olivine (Beattie, 1994) and clinopyroxene(Adam & Green, 2006). Tick marks correspond to 1%, 2%, 5%, 10%, 20% and 50% crystallization. Data for the Ohre Rift andOberpfalz are shown for comparison (see Fig. 6 for references).


at Erlangen N



Dow

nloaded from


Sr–Nd isotope space could also be consistent with a

parental melt linked to carbonatite; the Zelezna hurka

samples plot at the overlap between the HIMU mantle

end-member and oceanic and continental carbonatites

(Hoernle et al., 2002), again indicating a minor influenceof upper continental crust assimilation.

Our new data include the first 176Hf/177Hf data from

the Ohre Rift. Samples from Zelezna hurka plot close to

the mantle array of Vervoort et al. (1999) in the overlap

of the MORB, EM2 and HIMU fields (Pfander et al., 2012;

Fig. 9b). The kimberlites of Udachnaya East

(Kamenetsky et al., 2009b) plot at the low end of theHIMU field in (eNd)i–(eHf)i space (Fig. 9b). This observa-

tion is consistent with constraints on the nature of the

HIMU, EM1 and EM2 mantle reservoirs by Jackson &

Dasgupta (2008). HIMU is characterized by silica defi-

ciency (�43 wt % SiO2) and high CaO/Al2O3 (�1�1), and

is likely to have evolved from carbonation of peridotite.In contrast, EM1 and EM2 have higher silica and lower

CaO/Al2O3 (>1�0) and may represent mantle peridotite

enriched with sediments and by low-degree partial melt

metasomatism, respectively (Jackson & Dasgupta,

2008).

The geochemical challenge to be addressed here isto identify and quantify the distinct mantle reservoirs

that contribute to magma generation below Zelezna

hurka, especially in terms of their asthenospheric or

lithospheric (metasomatic) origin. Melting owing to up-

welling of asthenospheric mantle in response to exten-

sional tectonics is indicated by the abundant

occurrence of basaltic lavas in the Ohre Rift where thelithosphere was thinned to about 80 km thickness (e.g.

Babuska & Plomerova, 2010). Widespread mixing of an

isotopically depleted (asthenospheric) component and

an isotopically more enriched (lithospheric) component

may be an efficient way to explain the isotopic compos-

itions of most CEVP samples (including Zelezna hurka),as originally proposed by Wilson & Downes (1991).

Further insights from trace element modellingWe have shown above that Sr–Nd–Hf isotope data indi-

cate the presence of a mixed mantle source for the

CEVP magmatism. Before we further constrain thenature of the metasomatic agents, we demonstrate

below the need for a heterogeneous and enriched man-

tle source by melt modelling.

Pfander et al. (2012) developed a model in which

they demonstrated the effect of melt metasomatism on

the composition of the subcontinental lithospheric man-

tle. The SCLM is characterized by an enrichment in traceelements reflecting its ability to freeze infiltrating melts

and preserve their components in the form of volatile-

bearing minerals (e.g. amphibole, phlogopite). This

model reproduces the geochemical variability observed

in the CEVP and Ohre Rift lavas by variable proportions

of mixing between an asthenospheric melt componentand a lithospheric melt component, each of which

may be the results of variable degrees of partial melting

(Fig. 16). At the low degrees of melting assumed for the

metasomatizing melt, Nb can be fractionated efficiently

from other elements (Pfander et al., 2007) leading to ele-

vated Nb/Ta at low Zr/Nb (Fig. 16a) or high La/Yb at high

Nb/La (Fig. 16b). The results of melting spinel- or gar-net-peridotite and the modelled metasomatized SCLM

(‘lithospheric melts’) are shown in Fig. 16. In this model,

the Quaternary volcanic rocks from the Ohre Rift to-

gether with the SW German melilitites show the stron-

gest signature of a contribution from a metasomatized

mantle source.

Composition and evolution of the mantle sourceSeveral studies of mantle xenoliths have demonstrated

that the lithospheric mantle beneath the Ohre Rift is sig-

nificantly altered by metasomatic processes (e.g.

Geissler et al., 2007; Puziewicz et al., 2011; Ackermanet al., 2013, 2014). As a result, the lithology of mantle

xenoliths is bimodal with a refractory (mainly harzbur-

gitic) peridotite suite and a metasomatic pyroxenite

suite (e.g. Geissler et al., 2007; Puziewicz et al., 2011).

The style of metasomatism is variable and ranges from

carbonatitic melt infiltration to ‘Fe-metasomatism’ as a

result of alkaline silicate melt infiltration (Puziewiczet al., 2011). Physical evidence for the presence of car-

bonatitic melts is recorded in silicate and silicate–car-

bonate melt pockets in xenoliths of the Oberpfalz (Zinst,

Hirschentanz and Teichelberg), hosting subhedral

phenocrysts of olivine, clinopyroxene and combin-

ations of the two, as well as carbonate minerals and il-menite (e.g. Ackerman et al., 2013). However, if we

assume that fertile lithologies (e.g. metasomatic clino-

pyroxene-rich veins) melt preferentially (e.g. Foley,

1992; Phipps Morgan & Morgan, 1999), then these

xenolith suites would probably represent fragments of

the more refractory residual mantle, rather than the ac-

tual magma source.Our new petrological and geochemical data provide

some additional evidence for the role of carbonatitic

melt infiltration and reaction with mantle peridotite.

Several studies of the CEVP have shown O isotopes to

be a powerful tracer of mantle metasomatism (e.g.

Kempton et al., 1988). High d18O in combination withhigh CaO/Al2O3 should thus provide strong evidence for

the interaction of carbonatite and peridotitic mantle.

Indeed, olivines entrained in Zelezna hurka lavas have

d18O significantly higher (up to þ5�6% V-SMOW) than

normal mantle olivine (þ5�2 6 0�2% V-SMOW) and ex-

tend into the field of fresh olivines entrained in the

Udachnaya East kimberlite (Fig. 17). Similarly, glasses(‘d18OWR’; Fig. 11) have a heavier O isotope signature,

even though these may have been altered to lighter val-

ues by degassing, as indicated by the high vesicularity

of the Zelezna hurka scoria (up to 30 vol. %) and the

high concentration of S and Cl, two elements with a

lower volatility than H2O or CO2 (likely to be degassed).The corresponding decrease in d18OWR could be of the

order of 1–2% (up to –0�4% per 10 wt % volatile loss;


at Erlangen N



Dow

nloaded from


Eiler, 2001) and initial d18O would then be similar to the

range of CEVP lavas (up to þ7�5% V-SMOW) reported

by Mayer et al. (2014). In combination with CaO/Al2O3

we can further show that there is a positive correlation

between heavier O isotope compositions and Ca excess

and that this trend is contrary to crustal assimilation

trends (Fig. 11b).

Additional details on the nature of the metasomaticagent may be resolved by considering the distinct trace

element composition of the Zelezna hurka lavas

(a)

(b)

H-group meltsL-group melts

GrtSpl

10

5

3

2

10

5

2

1

0.7

1010

5

52

23

31

1 0.7 0.5

0.7 0.5

30

asthenospheric

melts

lithosphericmelts

1

1 2 3 4 5 6 7 8

Zr/Nb

12

14

16

18

20

22

24

Nb

/Ta

0.7%

30%10%

5%

1%

2%

0.7%

3%

3%

2%

1% 0.7%

1%

2%

0.5%

Spl-Peridotite

Grt-Peridotite

metasomatisedSpl-Peridotite

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Nb/La

0

10

20

30

40

50

60

70

80

90

La/

Yb

222

NephelinitesOther lavas

VogelsbergEifelRhönSWG melilititesOR melilitites

Komorní hůrkaZH whole rock

Mýtina Maar

this studyliterature data

Ohře Rift

ZH glasses

Fig. 16. (a) Nb/Ta vs Zr/Nb for Quaternary and older volcanic rocks of the Ohre Rift and the CEVP (data sources as in Fig. 6). Coloreddashed curves indicate melting trends for garnet peridotite and spinel peridotite and a metasomatized, refractory spinel peridotite.[For full details see Pfander et al. (2012).] (b) La/Yb vs Nb/La for the same samples as in Fig. 13 (data sources as in Fig. 6) with melt-ing trends for asthenospheric melts and lithospheric melts according to Pfander et al. (2012). L- and H-group melts (moderately andhighly trace element enriched melts, respectively) are derived from melting an amphibole- and phlogopite-bearing spinel peridotitewith an enriched composition that has been constrained from natural mantle xenoliths from the Hessian depression [see Pfanderet al. (2012) for details]. Numbers adjacent to the model curves indicate the per cent melting.


at Erlangen N



Dow

nloaded from


[especially the transitional composition of glasses be-

tween the nephelinite and melilitite rock series (Fig. 7b)]

compared with other CEVP volcanic rocks. Carbonatiticmelt metasomatism is an effective process by which to

fractionate certain trace element ratios such as Ba/Th,

K/La, Zr/Sm or Ti/Eu (e.g. Sweeney et al., 1995; Yaxley

et al., 1998). More specifically, large ion lithophile elem-

ents (LILE), LREE and HREE, and also Th, Nb, Ta and Sr

become moderately to highly enriched, and decoupled

from Ti abundances (e.g. Green & Wallace, 1988;Yaxley et al., 1991, 1998). These characteristic enrich-

ments are easily recognizable in multi-element patterns

(Fig. 8a) with high concentrations of the LILE, a positive

Nb and Ta anomaly, and slight negative anomalies of

Zr, Hf and Ti. As a result, the Quaternary volcanic rocks

from the Ohre Rift (along with lavas from the Oberpfalz)plot at the high end of the CEVP range in CaO/Al2O3 vs

(La/Yb)N (Fig. 18a). This range is even further extended

by the melilitites of the Ohre Rift and those from SW

Germany. For the latter, an origin by partial melting of a

carbonated peridotite source has been proposed by

Dunworth & Wilson (1998). It has to be noted, however,

that La can also be fractionated from Yb by melting inthe stability field of garnet, as HREE are retained as

compatible elements in residual garnet. More unambi-

gous is the fractionation of Zr from Sm, a process that

has been associated with carbonatitic melt metasoma-

tism (e.g. Pfander et al., 2012). Samples from the

Oberpfalz, the Massif Central, the SW German melili-tites and Quaternary volcanic rocks of the Ohre Rift

show the strongest fractionation of (Ti/Eu)N relative to

(Zr/Sm)N (Fig. 18b). Whereas the change in (Ti/Eu)N

could also be attributed to assimilation of material of

the continental crust (UCC and LCC in Fig. 18b), this pro-

cess is insufficient to explain the subchondritic Zr/Sm

ratios, which provides strong evidence for carbonatitemetasomatism in the SCLM (Pfander et al., 2012).

The genetic link between kimberlites, melilititesand nephelinitesForsterite-rich, high-Ni olivine cores may be interpreted

either as the earliest crystallization products of a meltthat infiltrates the lithospheric mantle (e.g. Dunworth &

Wilson, 1998) or as direct reaction products of carbona-

titic melt infiltration into the SCLM. This reaction of en-

statite and dolomite to forsterite, diopside and melt

(e.g. Yaxley et al., 1991; Dalton & Wood, 1993) would be

a plausible explanation for the close similarities in min-eralogy and chemical composition between the Zelezna

hurka nephelinites and kimberlites. Olivines entrained

in the lava show the same compositional trends as

fresh olivines in the Udachnaya East kimberlite

(Kamenetsky et al., 2008), clinopyroxenes have similar

trace element patterns, and accessory phases such as

phlogopite and spinel sensu lato suggest a genetic link.Further evidence for the major role of a carbonate

phase during the petrogenesis of the Zelezna hurka

magmas comes from radiogenic (enriched mantle sig-

natures) and stable isotopes (high d18O of olivines and

glasses) and major (e.g. high CaO/Al2O3, high Cl and S

concentrations) and trace elements (e.g. low Ti/Eu andZr/Sm, Fig. 18b). Shallow oxidation owing to pressure

release indicated by titanomagnetite and hauyne

phenocrysts in the Zelezna hurka lavas is also observed

in kimberlites (Yaxley et al., 2012).

However, there are also important differences be-

tween ‘high-carbonate’ lavas (such as kimberlites) and

the nephelinites of Zelezna hurka. First of all, there isevidence for much shallower depths of melting and

melt segregation for the nephelinites and a general dif-

ference in their eruption style. Kimberlites form major

diatremes that indicate explosive eruptions, whereas

the Quaternary nephelinites in the Ohre Rift may erupt

extrusively (Komornı Hurka, Zelezna hurka vent) or ex-plosively when the ascending melt comes into contact

with aquifers (Zelezna hurka tephra, Mytina Maar). The

elemental and isotopic characteristics of the nephelin-

ites are also less extreme than those of kimberlites,

which, at least in the case of the fresh Udachnaya East

kimberlite, show a HIMU-like isotope signature and a

much higher degree of silica-undersaturation (�32 wt %SiO2; Kamenetsky et al., 2009b). This is also reflected in

the abundance of clinopyroxene, which is a major

phase in nephelinites but rarely present in kimberlites

(e.g. hosted as inclusions in olivine; Kamenetsky et al.,

2009a).

However, our new data provide direct magmaticevidence for a genetic link between kimberlites, melili-

tites and nephelinites as suggested previously by

Fig. 17. d18O of olivine vs forsterite content for Zelezna hurkacompared with data from South African melilitites (Day et al.,2014) and the Azores (Genske et al., 2013), where the oxygenisotope composition of olivine is related to assimilation–frac-tional crystallization processes (AFC; assimilation of alteredoceanic crust; blue arrow). An assimilation trend expected forcontinental crustal material is shown in orange [bulk continen-tal crust Mg#¼55 (Rudnick & Gao, 2003), d18O�6�5%], assum-ing a melt–olivine fractionation of �0�5% and a d18O of 7–14%for granitoid crust (Eiler, 2001). It should be noted that the Fo-rich olivines with d18O higher than the mantle array (Matteyet al., 1994) extend into the field (dark grey) of fresh olivinesfrom the Udachnaya East kimberlite (Kamenetsky et al., 2008).


at Erlangen N



Dow

nloaded from


experimental studies (e.g. Lee & Wyllie, 1997). We sug-

gest that the dominant controls on the parental melt

composition are the amount of carbonate involved, the

depth of melting and melt segregation, and the amount

of melt–rock reaction (assimilation of peridotite wall-rock during melt ascent). Kimberlites would represent

one endmember with high amounts of carbonate, a

great depth of melt segregation (erupting in thick cra-

tonic lithosphere) and minor interaction with the re-

sidual peridotite mantle. Rifting and lithosphere

thinning may produce preferential pathways for alkali-

carbonate melts or fluids to migrate towards the sur-face, with the potential to infiltrate and metasomatically

enrich the SCLM (e.g. Dalton & Wood, 1993; Giuliani

0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2

(Zr/Sm)N

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

(Ti/E

u) N

0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2

CaO/Al2O3

0

10

20

30

40

50

60

70

80

(La/

Yb

) N

PM

UCC

LCC

(a)

(b)

Ohře Rift

ZH glasses

NephelinitesOther lavas

Oberpfalz

VogelsbergEifelRhönSWG melilitites

OR melilitites

Komorní hůrkaZH whole rocks

Mýtina Maar

Fig. 18. (a) Chondrite-normalized (Palme & O’Neill, 2003) La/Yb vs CaO/Al2O3. Quaternary rocks from the Ohre Rift and SW Germanmelilitites point towards a mantle source metasomatized by carbonatitic melts (e.g. characteristic trace element enrichment andhigh CaO/Al2O3). (b) (Ti/Eu)N vs (Zr/Sm)N [normalized to Cl-values of Palme & O’Neill (2003)]. Compositions of upper (UCC) andlower continental crust (LCC) according to Rudnick & Gao (2003) and the value for the Primitive Mantle (PM; Palme & O’Neill, 2003)are shown for comparison. Assimilation of continental crust or partial (batch) melting of peridotite may effectively lower (Ti/Eu)Nbut cannot explain the fractionation of (Zr/Sm)N. These trace element ratios point clearly towards a carbonated mantle source. Datasources as in Fig. 13.


at Erlangen N



Dow

nloaded from


et al., 2012). Nephelinites could then form either directly

from the carbonatitic melt percolating and assimilating

residual peridotite or by earlier mantle metasomatism

and later reactivation of these veins, or a combination

of the two. Pfander et al. (2012) proposed an age ofmantle metasomatism in Central Europe of �100 Ma,

which would argue for mantle metasomatism

decoupled from recent volcanism. However, there is

also evidence for the presence of crustal fluids that

could be linked to recent melt infiltration into the SCLM.

Even though the depths of earthquake swarms in the

area of Novy Kostel (linked to migrating fluids) are shal-lower (8�5–9�5 km depth; Horalek et al., 2000) than

our estimates of the depth of melt migration (>20 km;

Fig. 14), near-surface fluids carry a strong mantle iso-

tope signal (e.g. high 3He/4He; Weinlich, 2013).

CONCLUDING REMARKS

Previous studies have shown that the CEVP parental

magmas formed by melting variable proportions of

metasomatized (enriched) subcontinental lithospheric

mantle and depleted asthenospheric mantle. The neph-elinites of the Quaternary Zelezna hurka volcano in the

Ohre Rift provide strong mineralogical and chemical

evidence for the nature of this mantle metasomatism,

involving alkaline-carbonate melts or fluids, as follows.

1. Olivine antecrysts entrained in the nephelinite lava

show chemical evidence for crystallization in the

mantle, subsequent ‘normal’ crystallization and a

later overprint reflecting solid-state diffusion; this

evolutionary pattern is similar to that observed in

fresh olivines of the Udachnaya East kimberlite.2. The trace element patterns of clinopyroxenes also

resemble those of kimberlitic clinopyroxenes and

their crystallization conditions argue for a continu-

ous process in the upper SCLM and lower crust ra-

ther than a long period of melt stagnation at a

distinct level in the lithosphere.3. The compositions of accessory phlogopite are inter-

mediate between those in lamproites and carbona-

tites; the presence of spinels sensu lato and hauyne

argues for shallow oxidation of the melt, similar to

observations in carbonate-rich melts.

4. Elevated O isotope ratios (relative to mantle values)

and distinct trace element enrichment (e.g. LILE,LREE, Nb and Ta) and trace element ratios (e.g. Ti/

Eu, Zr/Sm) provide further evidence for a contribu-

tion of carbonatite to the final melt composition.

Radiogenic Sr–Nd–Hf isotope data support this view,

although their signatures may also be explained by

processes other than carbonate melt metasomatism.

Furthermore, we have shown that crustal assimila-

tion may play a role in the petrogenesis of the Zelezna

hurka nephelinite, but is insufficient to account for allthe mineralogical and chemical evidence we have

found for carbonate melt–peridotite interaction. The

genetic link between kimberlites, melilitites and nephel-

inites has been previously suggested based on experi-

mental and xenolith studies, but this study provides

direct magmatic evidence. The depth of alkali-carbonate

melt–fluid segregation, its total volume and the propor-tion of peridotite assimilation during melt ascent

through the mantle may control the final magma type.

ACKNOWLEDGEMENTS

We collected our samples without using mechanical

tools to avoid any damage on the protected outcrop ofZelezna hurka and we would like to encourage every

visitor to this location to help to preserve it in its current

condition. We thank H. Bratz and M. Hertel at

GeoZentrum Nordbayern and N. Pearson at GEMOC for

their analytical help. We also acknowledge the co-oper-

ation and support of A. Weh and the Selfrag AG(Kerzers, Switzerland) for their help with high-voltage

pulse power fragmentation of olivine-phyric rocks. L.

Ackerman, S. Jung, J. Pfander and editor M. Wilson are

acknowledged for comments that significantly im-

proved the quality and clarity of this paper. P.A.B.

thanks G. Yaxley and O. Nebel for constructive com-

ments on an earlier version of this paper.

FUNDING

This work was supported by a grant of the

‘Sonderfonds fur wissenschaftliche Arbeiten an der

Universitat Erlangen–Nurnberg’ to P.A.B. and F.S.G.and by funding through grant WI 3675/1-1 from the

Deutsche Forschungsgemeinschaft. P.A.B. benefited

from a Feodor Lynen Research Fellowship of the

Alexander von Humboldt Foundation.

SUPPLEMENTARY DATA

Supplementary data for this paper are available at

Journal of Petrology online.

REFERENCES

Ackerman, L., Spacek, P., Magna, T., Ulrych, J., Svojtka, M.,Hegner, E. & Balogh, K. (2013). Alkaline and carbonate-richmelt metasomatism and melting of subcontinental litho-spheric mantle: evidence from mantle xenoliths,NE Bavaria, Bohemian Massif. Journal of Petrology 54,2597–2633.

Ackerman, L., Medaris, G., Jr, Spacek, P. & Ulrych, J. (2014).Geochemical and petrological constraints on mantle com-position of the Ohre (Eger) rift, Bohemian Massif: peridotitexenoliths from the Ceske Stredohorı Volcanic complex andnorthern Bohemia. International Journal of Earth Sciencesdoi:10.1007/s005.31-014-1054-1.

Adam, J. & Green, T. (2006). Trace element partitioning be-tween mica- and amphibole-bearing garnet lherzolite andhydrous basanitic melt: 1. Experimental results and the in-vestigation of controls on partitioning behaviour.Contributions to Mineralogy and Petrology 152, 1–17.


at Erlangen N



Dow

nloaded from



Babuska, V. & Plomerova, J. (2010). Mantle lithosphere controlof crustal tectonics and magmatism of the western Ohre(Eger) Rift. Journal of GEOsciences 55, 171–186.

Bankwitz, P., Schneider, G., Kampf, H. & Bankwitz, E. (2003).Structural characteristics of epicentral areas in CentralEurope: study case Cheb Basin (Czech Republic). Journal ofGeodynamics 35, 5–32.

Beattie, P. (1994). Systematics and energetics of trace-elementpartitioning between olivine and silicate melts: implicationsfor the nature of mineral/melt partitioning. ChemicalGeology 117, 57–71.

Begemann, F., Ludwig, K. R., Lugmair, G. W., Min, K., Nyquist,L. E., Patchett, P. J., Renne, P. R., Shih, C. Y., Villa, I. M. &Walker, R. J. (2001). Call for an improved set of decay con-stants for geochronological use. Geochimica etCosmochimica Acta 65, 111–121.

Blusztajn, J. & Hart, S. R. (1989). Sr, Nd, and Pb isotopic charac-ter of Tertiary basalts from southwest Poland. Geochimicaet Cosmochimica Acta 53, 2689–2696.

Bohlen, S. R., Boettcher, A. L., Wall, V. J. & Clemens, J. D.(1983). Stability of phlogopite–quartz and sanidine–quartz: Amodel for melting in the lower crust. Contributions toMineralogy and Petrology 83, 270–277.

Bouvier, A., Vervoort, J. D. & Patchett, P. J. (2008). The Lu–Hfand Sm–Nd isotopic composition of CHUR: Constraintsfrom unequilibrated chondrites and implications for the bulkcomposition of terrestrial planets. Earth and PlanetaryScience Letters 273, 48–57.

Brandl, P. A., Beier, C., Regelous, M., Abouchami, W., Haase, K.M., Garbe-Schonberg, D. & Galer, S. J. G. (2012). Volcanismon the flanks of the East Pacific Rise: Quantitative con-straints on mantle heterogeneity and melting processes.Chemical Geology 298–299, 41–56.

Brauer, K., Kampf, H., Niedermann, S., & Strauch, G. (2005).Evidence for ascending upper mantle-derived melt beneaththe Cheb basin, central Europe. Geophysical ResearchLetters 32, L08303.

Brauer, K., Kampf, H. & Strauch, G. (2009). Earthquake swarmsin non-volcanic regions: What fluids have to say.Geophysical Research Letters 36, L17309.

Cebria, J. M. & Wilson, M. (1995). Cenozoic mafic magmatismin Western/Central Europe: a common European astheno-spheric reservoir. Terra Abstracts 7, 162.

Cebria, J. M., Lopez-Ruiz, J., Doblas, M., Oyarzun, R., Hertogen,J. & Benito, R. (2000). Geochemistry of the Quaternary alkalibasalts of Garrotxa (NE Volcanic Province, Spain): a case ofdouble enrichment of the mantle lithosphere. Journal ofVolcanology and Geothermal Research 102, 217–235.

Dahm, T., Fischer, T. & Hainzl, S. (2008). Mechanical intrusionmodels and their implications for the possibility of magma-driven swarms in NW Bohemia Region. Studia Geophysicaet Geodaetica 52, 529–548.

Dalton, J. A. & Wood, B. J. (1993). The compositions of primarycarbonate melts and their evolution through wallrock reac-tion in the mantle. Earth and Planetary Science Letters 119,511–525.

Dasgupta, R., Hischmann, M. & Smith, N. (2007). Partial meltingexperiments of peridotiteþCO2 at 3 GPa and genesis ofalkalic ocean island basalts. Journal of Petrology 48,2093–2124.

Day, J. M. D., Peters, B. J. & Janney, P. E. (2014). Oxygen isotopesystematics of South African olivine melilitites and implica-tions for HIMU mantle reservoirs. Lithos 202–203, 76–84.

Dobosi, G. & Fodor, R. V. (1992). Magma fractionation, replen-ishment, and mixing as inferred from green-core clinopyr-oxenes in Pliocene basanite, southern Slovakia. Lithos 28,133–150.

Downes, H. (1987). Tertiary and Quaternary volcanism in theMassif Central, France. In: Fitton, J. G. & Upton, B. G. J. (eds)Alkaline Igneous Rocks. Geological Society, London,Special Publications 30, 517–530.

Duda, A. & Schmincke, H.-U. (1985). Polybaric differentiation ofalkali basaltic magmas: evidence from green-core clinopyr-oxenes (Eifel, FRG). Contributions to Mineralogy andPetrology 91, 340–353.

Dunworth, E. A. & Wilson, M. (1998). Olivine melilitites of theSW German Tertiary Volcanic Province: mineralogy andpetrogenesis. Journal of Petrology 39, 1805–1836.

Eiler, J. M. (2001). Oxygen isotope variations of basaltic lavasand upper mantle rocks. In: Valley, J. W. & Cole, D. (eds)Stable Isotope Geochemistry. Mineralogical Society ofAmerica and Geochemical Society, Reviews in Mineralogyand Geochemistry 43, 319–364.

Eiler, J. M., Schiano, P., Kitchen, N. & Stolper, E. M. (2000).Oxygen isotope evidence for recycled crust in the sources ofmid ocean ridge basalts. Nature 403, 530–534.

Fekiacova, Z., Mertz, D. F. & Renne, P. R. (2007). Geodynamicsetting of the Tertiary Hocheifel volcanism (Germany), PartI: 40Ar/39Ar geochronology. In: Ritter, R.R., & Christensen,U.R. (eds) Mantle Plumes. Berlin: Springer, pp. 185–206.

Fischer, T. & Horalek, J. (2003). Space–time distribution ofearthquake swarms in the principal focal zone of the NWBohemia/Vogtland seismoactive region: period 1985–2001.Journal of Geodynamics 35, 125–144.

Foley, S. F. (1992). Petrological characterization of the sourcecomponents of potassic magmas: geochemical and experi-mental constraints. Lithos 28, 187–204.

Freund, S., Beier, C., Krumm, S. & Haase, K. M. (2013). Oxygenisotope evidence for the formation of andesitic–dacitic mag-mas from the fast-spreading Pacific–Antarctic Rise by as-similation–fractional crystallisation. Chemical Geology 347,271–283.

Geissler, W. (2005). Seismic and petrological investigations ofthe lithosphere in the swarm-earthquake and CO2-degass-ing region Vogtland/NW-Bohemia. GeoForschungsZentrumPotsdam, Scientific Technical Report STR05/06, 1–169.

Geissler, W., Kampf, H., Seifert, W. & Dulski, P. (2007).Petrological and seismic studies of the lithosphere in theearthquake swarm region Vogtland/NW Bohemia, centralEurope. Journal of Volcanology and Geothermal Research159, 33–69.

Geissler, W. H., Kampf, H., Bankwitz, P. & Bankwitz, E. (2004).Das quartare Tephra-Tuff-Vorkommen von Mytina (Sudranddes westlichen Eger-Grabens/Tschechische Republik):Indikationen fur die Ausbruchs- und Deformationsprozesse.Zeitschrift fur Geologische Wissenschaften 32, 31–54.

Genske, F. S., Turner, S. P., Beier, C. & Schaefer, B. F. (2012).The petrology and geochemistry of lavas from the westernAzores islands of Flores and Corvo. Journal of Petrology 53,1673–1708.

Genske, F. S., Beier, C., Haase, K. M., Turner, S. P., Krumm, S. &Brandl, P. A. (2013). Oxygen isotopes in the Azoresislands: Crustal assimilation recorded in olivine. Geology 41,491–494.

Giuliani, A., Kamenetsky, V. S., Phillips, D., Kendrick, M. A.,Wyatt, B. A. & Goemann, K. (2012). Nature of alkali-carbon-ate fluids in the sub-continental lithospheric mantle.Geology 40, 967–970.

Granet, M., Wilson, M. & Achauer, U. (1995). Imaging a mantleplume beneath the French Massif Central. Earth andPlanetary Science Letters 136, 281–296.

Grapes, R. H. (1986). Melting and thermal reconstitution of pel-itic xenoliths, Wehr Volcano, East Eifel, West Germany.Journal of Petrology 27, 343–396.


at Erlangen N



Dow

nloaded from


Green, D. H. & Wallace, M. E. (1988). Mantle metasomatism byephemeral carbonatite melts. Nature 336, 459–462.

Haase, K. & Renno, A. (2008). Variation of magma generationand mantle sources during continental rifting observed inCenozoic lavas from the Eger Rift, Central Europe. ChemicalGeology 257, 195–205.

Haase, K. M., Krumm, S., Regelous, M. & Joachimski, M. (2011).Oxygen isotope evidence for the formation of silicicKermadec island arc and Havre–Lau backarc magmas byfractional crystallisation. Earth and Planetary ScienceLetters 309, 348–355.

Hegner, E., Walter, H. J. & Satir, M. (1995). Pb–Sr–Nd isotopiccompositions and trace element geochemistry of mega-crysts and melilitites from the Tertiary Urach volcanic field:source composition of small volume melts under SWGermany. Contributions to Mineralogy and Petrology 122,322–335.

Hermann, J., O’Neill, H. S. C. & Berry, A. J. (2005). Titaniumsolubility in olivine in the system TiO2–MgO–SiO2: no evi-dence for an ultra-deep origin of Ti-bearing olivine.Contributions to Mineralogy and Petrology 148, 746–760.

Heuer, B. (2006). Lithospheric and upper mantle structure be-neath the Bohemian Massif obtained from teleseimsmic Pand S receiver functions. GeoForschungsZentrum Potsdam,Scientific Technical Report STR06/12, 1–161.

Hirose, K. & Kushiro, I. (1993). Partial melting of dry peridotitesat high pressures: Determination of compositions of meltssegregated from peridotite using aggregates of diamond.Earth and Planetary Science Letters 114, 477–489.

Hirschmann, M. M., Kogiso, T., Baker, M. B. & Stolper, E. M.(2003). Alkalic magmas generated by partial melting of gar-net pyroxenite. Geology 31, 481–484.

Hoernle, K., Zhang, Y.-S. & Graham, D. (1995). Seismic and geo-chemical evidence for large-scale mantle upwelling beneaththe eastern Atlantic and western and central Europe. Nature374, 34–39.

Hoernle, K., Tilton, G., Le Bas, M. J., Duggen, S. & Garbe-Schonberg, D. (2002). Geochemistry of oceanic carbonatitescompared with continental carbonatites: mantle recycling ofoceanic crustal carbonate. Contributions to Mineralogy andPetrology 142, 520–542.

Hofmann, A. W., Jochum, K. P., Seufert, M. & White, W. M.(1986). Nb and Pb in oceanic basalts: new constraints onmantle evolution. Earth and Planetary Science Letters 79,33–45.

Horalek, J., Sıleny, J., Fischer, T., Slancova, A. & Bouskova, A.(2000). Scenario of the January 1997 West BohemiaEarthquake Swarm. Studia Geophysica et Geodaetica 44,491–521.

Jackson, M. G. & Dasgupta, R. (2008). Compositions of HIMU,EM1, and EM2 from global trends between radiogenic iso-topes and major elements in ocean island basalts. Earth andPlanetary Science Letters 276, 175–186.

Jones, A. P. & Wyllie, P. J. (1985). Paragenetic trends of oxideminerals in carbonate-rich kimberlites, with new analysesfrom the Benfontein Sill, South Africa. Journal of Petrology26, 210–222.

Jung, C., Jung, S., Hoffer, E. & Berndt, J. (2006). Petrogenesis ofTertiary mafic alkaline magmas in the Hocheifel, Germany.Journal of Petrology 47, 1637–1671.

Jung, S. & Hoernes, S. (2000). The major- and trace-elementand isotope (Sr, Nd, O) geochemistry of Cenozoic alkalinerift-type volcanic rocks from the Rhon area (centralGermany): petrology, mantle source characteristics andimplications for asthenosphere–lithosphere interactions.Journal of Volcanology and Geothermal Research 99,27–53.

Jung, S. & Masberg, P. (1998). Major- and trace-element sys-tematics and isotope geochemistry of Cenozoic mafic vol-canic rocks from the Vogelsberg (central Germany). Journalof Volcanology and Geothermal Research 86, 151–177.

Jung, S., Pfander, J. A., Brauns, M. & Maas, R. (2011). Crustalcontamination and mantle source characteristics in contin-ental intra-plate volcanic rocks: Pb, Hf and Os isotopes fromcentral European volcanic province basalts. Geochimica etCosmochimica Acta 75, 2664–2683.

Jung, S., Mezger, K., Hauff, F., Pack, A. & Hoernes, S. (2013).Petrogenesis of rift-related tephrites, phonolites and trach-ytes (Central European Volcanic Province, Rhon, FRG):Constraints from Sr, Nd, Pb and O isotopes. ChemicalGeology 354, 203–215.

Jurewicz, A. J. G. & Watson, E. B. (1988). Cations in olivine, Part1: Calcium partitioning and calcium–magnesium distribu-tion between olivines and coexisting melts, with petrologicapplications. Contributions to Mineralogy and Petrology 99,176–185.

Kamenetsky, V. S., Kamenetsky, M. B., Sobolev, A. V., Golovin,A. V., Demouchy, S., Faure, K., Sharygin, V. V. & Kuzmin, D.V. (2008). Olivine in the Udachnaya-East kimberlite (Yakutia,Russia): types, compositions and origins. Journal ofPetrology 49, 823–839.

Kamenetsky, V. S., Kamenetsky, M. B., Sobolev, A. V., Golovin,A. V., Sharygin, V. V., Pokhilenko, N. P. & Sobolev, N. V.(2009a). Can pyroxenes be liquidus minerals in the kimber-lite magma? Lithos 112, 213–222.

Kamenetsky, V. S., Maas, R., Kamenetsky, M. B., Paton, C.,Phillips, D., Golovin, A. V. & Gornova, M. A. (2009b).Chlorine from the mantle: Magmatic halides in theUdachnaya-East kimberlite, Siberia. Earth and PlanetaryScience Letters 285, 96–104.

Kempton, P. D., Harmon, R. S., Stosch, H. G. & Hoefs, J. (1988).Open-system O-isotope behaviour and trace element en-richment in the sub-Eifel mantle. Earth and PlanetaryScience Letters 89, 273–287.

Le Bas, M. J. (1987). Nephelinites and carbonatites. In: Fitton, J.G. & Upton, B. G. J. (eds) Alkaline Igneous Rocks. GeologicalSociety, London, Special Publications 30, 53–83.

Le Bas, M. J. (1989). Nephelinitic and Basanitic Rocks. Journalof Petrology 30, 1299–1312.

Le Bas, M. J. & Streckeisen, A. L. (1991). The IUGS systematicsof igneous rocks. Journal of the Geological Society, London148, 825–833.

Lee, W.-J. & Wyllie, P. J. (1997). Liquid immiscibility betweennephelinite and carbonatite from 1�0 to 2�5 GPa comparedwith mantle melt compositions. Contributions toMineralogy and Petrology 127, 1–16.

Libourel, G. (1999). Systematics of calcium partitioning be-tween olivine and silicate melt: implications for melt struc-ture and calcium content of magmatic olivines.Contributions to Mineralogy and Petrology 136, 63–80.

Lustrino, M. & Wilson, M. (2007). The circum-Mediterraneananorogenic Cenozoic igneous province. Earth-ScienceReviews 81, 1–65.

Lyubetskaya, T. & Korenaga, J. (2007). Chemical compositionof Earth’s primitive mantle and its variance: 1. Method andresults. Journal of Geophysical Research 112, B03211.

Malkovsky, M. (1987). The Mesozoic and Tertiary basins of theBohemian Massif and their evolution. Tectonophysics 137,31–42.

Mattey, D., Lowry, D. & Macpherson, C. (1994). Oxygen isotopecomposition of mantle peridotite. Earth and PlanetaryScience Letters 128, 231–241.

Matzen, A. K., Baker, M. B., Beckett, J. R. & Stolper, E. M. (2013).The temperature and pressure dependence of nickel


at Erlangen N



Dow

nloaded from


partitioning between olivine and silicate melt. Journal ofPetrology 54, 2521–2545.

Mayer, B., Jung, S., Romer, R. L., Pfander, J. A., Klugel, A., Pack,A. & Groner, E. (2014). Amphibole in alkaline basalts fromintraplate settings: implications for the petrogenesis of alka-line lavas from the metasomatised lithospheric mantle.Contributions to Mineralogy and Petrology 167, 1–22.

Mrlina, J., Kampf, H., Geissler, W. & Van den Bogaard, P.(2007). Assumed Quaternary maar structure at the Czech/German border between Mytina and Neualbenreuth (west-ern Eger rift, Central Europe): geophysical, petrochemicaland geochronological indications. Zeitschrift furGeologische Wissenschaften 35, 213–230.

Mrlina, J., Kampf, H., Kroner, C., Mingram, J., Stebich, M.,Brauer, A., Geissler, W. H., Kallmeyer, J., Matthes, H. &Seidl, M. (2009). Discovery of the first Quaternary maar inthe Bohemian Massif, Central Europe, based on combinedgeophysical and geological surveys. Journal of Volcanologyand Geothermal Research 182, 97–112.

Munker, C., Weyer, S., Scherer, E. & Mezger, K. (2001).Separation of high field strength elements (Nb, Ta, Zr, Hf)and Lu from rock samples for MC-ICPMS measurements.Geochemistry, Geophysics, Geosystems 2, doi:10.1029/2001GC000183.

Mysen, B. O. & Virgo, D. (1980). Solubility mechanisms of CO2

in silicate melts: a Raman spectroscopic study. AmericanMineralogist 65, 885–899.

Nowell, D. A. G., Jones, M. C. & Pyle, D. M. (2006). EpisodicQuaternary volcanism in France and Germany. Journal ofQuaternary Science 21, 645–675.

Palme, H. & O’Neill, H. (2003). Cosmochemical estimates ofmantle composition. In: Carlson, R.W. The Mantle and Core2, Amsterdam: Elsevier, pp. 1–38.

Pearce, N. J. G., Perkins, W. T., Westgate, J. A., Gorton, M. P.,Jackson, S. E., Neal, C. R. & Chenery, S. P. (1997). A compil-ation of new and published major and trace element data forNIST SRM 610 and NIST SRM 612 glass reference materials.Geostandards and Geoanalytical Research 21, 115–144.

Petry, C., Chakraborty, S. & Palme, H. (2004). Experimental de-termination of Ni diffusion coefficients in olivine and theirdependence on temperature, composition, oxygen fugacity,and crystallographic orientation. Geochimica etCosmochimica Acta 68, 4179–4188.

Pfander, J., Munker, C., Stracke, A. & Mezger, K. (2007). Nb/Taand Zr/Hf in ocean island basalts—Implications for crust–mantle differentiation and the fate of niobium. Earth andPlanetary Science Letters 254, 158–172.

Pfander, J. A., Jung, S., Munker, C., Stracke, A. & Mezger, K.(2012). A possible high Nb/Ta reservoir in the continentallithospheric mantle and consequences on the global Nbbudget—Evidence from continental basalts from CentralGermany. Geochimica et Cosmochimica Acta 77, 232–251.

Phipps Morgan, J. & Morgan, W. (1999). Two-stage melting andthe geochemical evolution of the mantle: a recipe for mantleplum-pudding. Earth and Planetary Science Letters 170,215–239.

Putirka, K. D. (2008). Thermometers and barometers for vol-canic systems. In: Putirka, K. D. & Tepley, F. J., III (eds)Minerals, Inclusions and Volcanic Processes. MineralogicalSociety of America and Geochemical Society, Reviews inMineralogy and Geochemistry 69, 61–120.

Putirka, K., Mikaelian, H., Ryerson, F. & Shaw, H. (2003). Newclinopyroxene–liquid thermobarometers for mafic, evolvedand volatile-bearing lava compositions, with applications tolavas from Tibet and the Snake River Plain, Idaho. AmericanMineralogist 10, 1542–1554.

Puziewicz, J., Koepke, J., Gregoire, M., Ntaflos, T. & Matusiak-Malek, M. (2011). Lithospheric mantle modification duringCenozoic rifting in Central Europe: evidence from theKsieginki nephelinite (SW Poland) xenolith suite. Journal ofPetrology 52, 2107–2145.

Roeder, P. & Emslie, R. (1970). Olivine–liquid equilibrium.Contributions to Mineralogy and Petrology 29, 275–289.

Rowe, M. C., Lassiter, J. C. & Goff, K. (2015). Basalt volatile fluc-tuations during continental rifting: An example from the RioGrande Rift, USA. Geochemistry, Geophysics, Geosystems16, doi:10.1002/2014GC005649.

Rudnick, R. L. & Gao, S. (2003). Composition of the continentalcrust. In: Rudnick, R.L. The Crust 3, Amsterdam: Elsevier,pp. 1–64.

Russell, W. A., Papanastassiou, D. A. & Tombrello, T. A. (1978).Ca isotope fractionation on the Earth and other solar systemmaterials. Geochimica et Cosmochimica Acta 42, 1075–1090.

Sachs, P. M. & Hansteen, T. H. (2000). Pleistocene underplatingand metasomatism of the lower continental crust: a xenolithstudy. Journal of Petrology 41, 331–356.

Scherer, E., Munker, C. & Mezger, K. (2001). Calibration of theLutetium–Hafnium Clock. Science 293, 683–687.

Scherer, E. E., Cameron, K. L. & Blichert-Toft, J. (2000). Lu–Hfgarnet geochronology: closure temperature relative to theSm–Nd system and the effects of trace mineral inclusions.Geochimica et Cosmochimica Acta 64, 3413–3432.

Schulmann, K., Lexa, O., Janousek, V. & Lardeaux, J. M. (2014).Anatomy of a diffuse cryptic suture zone: An example fromthe Bohemian Massif, European Variscides. Geology 42,275–278.

Seifert, S., O’Neill, H. S. C. & Brey, G. (1988). The partitioning ofFe, Ni and Co between olivine, metal, and basaltic liquid: Anexperimental and thermodynamic investigation, with appli-cation to the composition of the lunar core. Geochimica etCosmochimica Acta 52, 603–616.

Seifert, W. & Kampf, H. (1994). Ba-enrichment in phlogopite of anephelinite from Bohemia. European Journal of Mineralogy6, 497–502.

Spacek, P., Ackerman, L., Habler, G., Abart, R. & Ulrych, J.(2013). Garnet breakdown, symplectite formation and melt-ing in basanite-hosted peridotite xenoliths from Zinst(Bavaria, Bohemian Massif). Journal of Petrology 54, 1691–1723.

Spicak, A. & Horalek, J. (2001). Possible role of fluids in theprocess of earthquake swarm generation in the WestBohemia/Vogtland seismoactive region. Tectonophysics336, 151–161.

Sprung, P., Scherer, E. E., Upadhyay, D., Leya, I. & Mezger, K.(2010). Non-nucleosynthetic heterogeneity in non-radio-genic stable Hf isotopes: Implications for early solarsystem chronology. Earth and Planetary Science Letters295, 1–11.

Sprung, P., Kleine, T. & Scherer, E. E. (2013). Isotopic evidencefor chondritic Lu/Hf and Sm/Nd of the Moon. Earth andPlanetary Science Letters 380, 77–87.

Stamper, C. C., Blundy, J. D., Arculus, R. J. & Melekhova, E.(2014). Petrology of plutonic xenoliths and volcanic rocksfrom Grenada, Lesser Antilles. Journal of Petrology 55,1353–1387.

Stracke, A. (2012). Earth’s heterogeneous mantle: A product ofconvection-driven interaction between crust and mantle.Earth and Planetary Science Letters 330–331, 274–299.

Straub, S. M., Homez-Tuena, A., Stuart, F. M., Zellmer, G. F.,Espinasa-Perena, R., Cai, Y. & Iizuka, Y. (2011). Formation ofhybrid arc andesites beneath thick continental crust. Earthand Planetary Science Letters 303, 337–347.


at Erlangen N



Dow

nloaded from


Sweeney, R. J. (1994). Carbonatite melt compositions in theEarth’s mantle. Earth and Planetary Science Letters 128,259–270.

Sweeney, R. J., Prozesky, V. & Przybylowicz, W. (1995).Selected trace and minor element partitioning between peri-dotite minerals and carbonatite melts at 18–46 kb pressure.Geochimica et Cosmochimica Acta 59, 3671–3683.

Taylor, H. P., Jr. (1980). The effects of assimilation of countryrocks by magmas on 18O/16O and 87Sr/86Sr systematics in ig-neous rocks. Earth and Planetary Science Letters 47, 243–254.

Ulrych, J., Dostal, J., Hegner, E., Balogh, K. & Ackerman, L. (2008).Late Cretaceous to Paleocene melilitic rocks of the Ohre/EgerRift in northern Bohemia, Czech Republic: Insights into the ini-tial stages of continental rifting. Lithos 101, 141–161.

Ulrych, J., Dostal, J., Adamovic, J., Jelınek, E., Spacek, P.,Hegner, E. & Balogh, K. (2011). Recurrent Cenozoic volcanicactivity in the Bohemian Massif (Czech Republic). Lithos123, 133–144.

Ulrych, J., Ackerman, L., Balogh, K., Hegner, E., Jelınek, E.,Pecskay, Z., Prichystal, A., Upton, B. G. J., Zimak, J. &Foltynova, R. (2013). Plio-Pleistocene basanitic and melilititicseries of the Bohemian Massif: K–Ar ages, major/trace elem-ent and Sr–Nd isotopic data. Chemie der Erde 73, 429–450.

van den Bogaard, P. (1995). 40Ar/39Ar ages of sanidinephenocrysts from Laacher See Tephra (12,900 yr BP):Chronostratigraphic and petrological significance. Earth andPlanetary Science Letters 133, 163–174.

Vaneckova, M., Holub, F. V., Soucek, J. & Bowes, D. R. (1993).Geochemistry and petrogenesis of the Tertiary alkaline vol-canic suite of the Labe tectonovolcanic zone, CzechRepublic. Mineralogy and Petrology 48, 17–34.

Vervoort, J. D., Patchett, P. J., Blichert-Toft, J. & Albarede, F.(1999). Relationships between Lu–Hf and Sm–Nd isotopicsystems in the global sedimentary system. Earth andPlanetary Science Letters 168, 79–99.

Wagner, G. A., Gogen, K., Jonckheere, R., Wagner, I. & Woda,C. (2002). Dating of Quaternary volcanoes Komorni Hurka(Kammerbuhl) and Zelezna Hurka (Eisenbuhl), CzechRepublic, by TL, ESR, alpha recoil and fission track

chronometry. Zeitschrift fur GeologischeWissenschaftlichen 30, 191–200.

Weinlich, F. H. (2013). Carbon dioxide controlled earthquake dis-tribution pattern in the NW Bohemian swarm earthquake re-gion, western Eger Rift, Czech Republic—gas migration in thecrystalline basement. Geofluids 1–17, doi:10.1111/gfl.12058.

Weinlich, F. H., Brauer, K., Kampf, H., Strauch, G., Tesar, J. &Weise, S. M. (1999). An active subcontinental mantle volatilesystem in the western Eger rift, Central Europe: Gas flux, iso-topic (He, C, and N) and compositional fingerprints.Geochimica et Cosmochimica Acta 63, 3653–3671.

Wilson, M. & Downes, H. (1991). Tertiary–Quaternary exten-sion-related alkaline magmatism in western and centralEurope. Journal of Petrology 32, 811–849.

Wilson, M. & Downes, H. (1992). Mafic alkaline magmatismassociated with the European Cenozoic rift system.Tectonophysics 208, 173–182.

Wones, D. R. & Dodge, F. C. W. (1977). The stability of phlogopitein the presence of quartz and diopside. In: Fraser, D.G.Thermodynamics in Geology. Dordrecht: Springer, pp. 229–247.

Yaxley, G. M., Crawford, A. J. & Green, D. H. (1991). Evidencefor carbonatite metasomatism in spinel peridotite xenolithsfrom western Victoria, Australia. Earth and PlanetaryScience Letters 107, 305–317.

Yaxley, G. M., Green, D. H. & Kamenetsky, V. (1998).Carbonatite metasomatism in the southeastern Australianlithosphere. Journal of Petrology 39, 1917–1930.

Yaxley, G. M., Berry, A. J., Kamenetsky, V. S., Woodland, A. B.& Golovin, A. V. (2012). An oxygen fugacity profile throughthe Siberian Craton—Fe K-edge XANES determinations ofFe3þ/

PFe in garnets in peridotite xenoliths from the

Udachnaya East kimberlite. Lithos 140–141, 142–151.Ziegler, P. A., Schumacher, M. E., Dezes, P., Van Wees, J.-D. &

Cloetingh, S. (2006). Post-Variscan evolution of the litho-sphere in the area of the European Cenozoic Rift System. In:Gee, D. G. & Stephenson, R. A. (eds) European LithosphereDynamics. Geological Society, London, Memoirs 32,97–112.


at Erlangen N



Dow

nloaded from


magmatic evidence for carbonate metasomatism in the ... · magmatic evidence for carbonate...

Documents