metasomatized mantle xenoliths as a record of the ... · france; 2department of applied geology,...

Metasomatized Mantle Xenoliths as a Record of

the Lithospheric Mantle Evolution of the

Northern Edge of the Ahaggar Swell, In Teria

(Algeria)

M.-A. Kaczmarek1,2 *, J.-L. Bodinier1, D. Bosch1, A. Tommasi1,

J.-M. Dautria1 and S. A. Kechid3

1Geosciences Montpellier, Universite de Montpellier and CNRS, Cc 60, Place E. Bataillon, 34095 Montpellier,

France; 2Department of Applied Geology, Institute for Geoscience Research, Curtin University, GPO Box U1987,

Perth, WA 6845, Australia and 3Universite des Sciences et de la Technologie Houari Boumediene, BP 32 El Alia

16111, Bab Ezzouar, Algeria

*Corresponding author. Present address: University of Lausanne, Institute of Earth Sciences, Geopolis,

Mouline, CH-1015 Lausanne, Switzerland.

E-mail: [email protected]

Received March 6, 2013; Accepted February 17, 2016

ABSTRACT

Mantle-derived xenoliths hosted by melilitite lavas from In Teria (Ahaggar, SE Algeria) include gar-

net and spinel peridotites, pyroxenite and phlogopite megacrysts. The spinel and garnet peridotites

record an early deformation event, which formed porphyroclastic microstructures and olivine crys-

tal preferred orientations, followed by static infiltration of hydrous alkaline melts. This metasomaticstage (stage 1) is characterized by the crystallization of phlogopite in the garnet and spinel perido-

tites, amphibole in the spinel peridotites and clinopyroxene in the garnet peridotite, which record

chemical equilibration with an alkaline silicate melt. These early events were largely overprinted by

carbonatitic metasomatism (stage 2), which is observed only in the spinel peridotites. Spinel

peridotite major and trace element compositions, as well as the compositions of newly formed

minerals, are characteristic of interaction with carbonate melt, associated with strong enrichmentin incompatible trace elements in clinopyroxene. This second stage was followed by crystallization

of pyroxenites (stage 3) in vein conduits, probably segregated from alkaline melts. We propose a

scenario in which the different metasomatic imprints record successive stages of interaction be-

tween lithospheric mantle and sublithospheric melts throughout the Cenozoic. In Sr–Nd isotope

space, the host melilitites and several xenoliths are clustered and plot close to the HIMU mantle

end-member. However, some peridotite xenoliths are shifted towards more radiogenic 87Sr/86Sr

values. In 207Pb/204Pb–206Pb/204Pb and 208Pb/204Pb–06Pb/204Pb space the In Teria samples define arelatively large domain characterized by high 206Pb/204Pb and 208Pb/204Pb, consistent with a contri-

bution of an HIMU component, considered to represent a sublithospheric signature. The highest87Sr/86Sr values are comparable with those ascribed to the EM1 mantle end-member, representing

the signature of the lower continental lithosphere, and are probably inherited from the pre-meta-

somatic lithospheric mantle beneath In Teria. Numerical modelling of porous percolation of melt of

sublithospheric origin through an EM1-like lithospheric mantle protolith reproduces the In Teriaperidotite compositions, using moderately sub-chondritic Sr/Nd values for the peridotite (e.g. In

Teria garnet peridotite) and moderately super-chondritic Sr/Nd values in the melt (approximately

ocean island basalt values). A few spinel peridotites require a component characterized by

VC The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected] 1

J O U R N A L O F

P E T R O L O G Y

Journal of Petrology, 2016, 1–38

doi: 10.1093/petrology/egw009

Journal of Petrology Advance Access published April 6, 2016

at University of C

ambridge on A

pril 12, 2016http://petrology.oxfordjournals.org/

Dow

nloaded from

http://www.oxfordjournals.org/

http://petrology.oxfordjournals.org/

a 143Nd/144Nd signature higher than both the EM1 end-member and the local Ahaggar basalts; the208Pb/204Pb compositions of several samples point to a component with a depleted mid-ocean

ridge basalt (MORB) mantle (DMM) signature. Thus the lithospheric mantle beneath In Teria prob-

ably did not have a uniform EM1 signature before the onset of metasomatism; it included a DMM

peridotite component as well as some peridotites with elevated 143Nd/144Nd values recording long-

term LREE depletion.

Key words: Hoggar Swell; mantle xenoliths; metasomatism; North Africa; Sr–Nd–Pb isotopes;subcontinental lithosphere; Tuareg Shield

INTRODUCTION

North Africa is characterized by the existence of several

volcanic swells (e.g. Ahaggar, Tibesti, Darfur) associ-

ated with regional topographic uplift, which suggests

interactions between lithospheric architecture and man-

tle plume activity. However, the relationships between

volcanism and large-scale mantle circulation beneath

North Africa remain poorly understood. Several models

have proposed a plume origin for the recent volcanism

and topography in North Africa, but they differ on the

proposed location of the plumes. Other models high-

light the importance of pre-existing lithosphere archi-

tecture. Ebinger & Sleep (1998) proposed the feeding of

northern African hotspots (Ahaggar, Tibesti and Darfur)

by asthenospheric material rising from the Afar plume

and channelled along zones of thinned lithosphere.

Another view is the development of splash plumes

(Davies & Bunge, 2006); these are buoyancy instabilities

rising from the mantle Transition Zone. The formation

of splash plumes in North Africa might be due to the ac-

cumulation of subducted oceanic slabs during closure

of the Tethys Ocean in the Mesozoic (Lustrino & Wilson,

2007). Another model proposes that the volcanism is

due to local mantle upwellings linked to reactivation of

older lithospheric structures during the early stages of

collision of the African plate with Eurasia (Liegeois

et al., 2005). Edge-driven convection at craton bounda-

ries was also proposed to explain some of the recent

volcanism in North Africa (King & Anderson, 1995,

1998; Missenard & Cadoux, 2012). The absence of hot-

spot tracks and the persistence of a plume signature be-

neath North Africa has been attributed to plume-head

material emplaced during the Cretaceous beneath the

Central Atlantic province and dragged by the African

plate since the Paleocene (Piromallo & Faccenna, 2004).

This study focuses on the Ahaggar (Hoggar) Massif,

which is the most extensively studied volcanic swell in

North Africa, and more specifically on its northeastern

termination. The In Teria district, near Illizi, is located

at the northeastern edge of the Ahaggar volcanic

swell and at the western edge of the Saharan

Metacraton, to the south of the Sahara Basins (Fig. 1a

and b). The lithospheric mantle beneath In Teria

was probably extensively modified during major

lithospheric–asthenospheric interactions that produced

tholeiitic and alkaline magmatism in central Ahaggar

(Aıt-Hamou et al., 2000) (Fig. 1b). The occurrence of gar-

net peridotite among the suite of In Teria mantle xeno-

liths (Dautria et al., 1992) suggests that relics of older

thicker lithosphere may have been preserved beneath

In Teria (Beccaluva et al., 2007). Paradoxically, the In

Teria district belongs to a domain of anomalously high

heat flow (>100 mW m–2) (Takherist & Lesquer, 1989;

Lesquer et al., 1990), and the exhumed mantle xenoliths

display evidence of extensive interaction with silica-

undersaturated silicate and carbonate melts (Dautria

et al., 1992). Both the high heat flow and the metasoma-

tism might be related to marginal effects of Ahaggar

plume activity (Davies & Bunge, 2006), or to more re-

cent mantle flow beneath North Africa (e.g. Ebinger &

Sleep, 1998; Piromallo & Faccenna, 2004; Forte et al.,

2010).

To understand the above observations we have con-

ducted a detailed study of the In Teria peridotite and

pyroxenite xenoliths, phlogopite megacrysts, and their

host melilitite. Microstructural analysis based on the

measurement of the crystallographic preferred orienta-

tions (CPO) of constituent minerals allows constraints

to be made on the deformation processes affecting the

lithospheric mantle beneath In Teria and on the relative

timing of deformation and metasomatism, as well as

determination of possible topotaxial relationships dur-

ing melt–rock reaction. New isotopic data (Sr–Nd–Pb)

for the mantle xenoliths and host lavas, together with

major and trace element analyses of whole-rocks and

constituent minerals, allow definition of the compos-

ition of the Ahaggar mantle lithosphere and description

of its evolution in response to different metasomatic

events. The Sr–Nd–Pb isotopic data indicate that the

percolating melt has a mantle affinity with an EM1 sig-

nature, which has not yet been observed from the

Ahaggar, nor in Morocco or Libya (Beccaluva et al.,

2007, 2008; Raffone et al., 2009; Wittig et al., 2010;

Natali et al., 2013). Together these data allow discussion

of the nature and evolution of the lithospheric mantle at

the northern edge of the Ahaggar Swell and the rela-

tionship of this evolution to the different geodynamic

events that affected the region.

GEOLOGICAL SETTING

The main structures of the Ahaggar Massif are inherited

from the Pan-African orogeny, which resulted from

2 Journal of Petrology, 2016, Vol. 0, No. 0

at University of C

ambridge on A


Dow

nloaded from


continental collision between the West African Craton

and an East African block between 580 and 760 Ma (e.g.

Caby et al., 1981; Lesquer et al., 1984; Bertrand et al.,

1986; Liegeois et al., 1994). Most of the uplift occurred

from Late Eocene to Quaternary times (Guiraud &

Bellion, 1995; Guiraud et al., 2005), but fission-track age

data suggest that the regional uplift of the Ahaggar

basement started at least as early as the Cretaceous

(Khaldi et al., 2006). Volcanic activity in the Ahaggar

Massif began during the late Eocene (c. 34 Ma; Remy,

1959; Rossi et al., 1979; Aıt-Hamou et al., 2000), peaked

during the Miocene and continued episodically through

the late Pliocene and into the late Quaternary (Girod,

1971).The In Teria district is situated within an anomal-

ously ‘hot’ east–west-trending zone characterized by

heat flow higher than 100 mW m–2, which has been

attributed to anomalously high temperatures in the

underlying asthenosphere (Fig. 1a; Takherist & Lesquer,

1989; Lesquer et al., 1990). Significantly lower heat flow

is recorded to the south of the study area in the

Ahaggar Massif (53 mW m–2 on average), suggesting a

stable lithosphere �100 km in thickness (Lesquer et al.,

1989). However, free-air gravity data led Crough (1981)

to suggest that the Central Ahaggar is underlain by

anomalously low-density upper mantle. Similar conclu-

sions have been reached based on seismic tomography

models (Ayadi et al., 2000; Sebai et al., 2006; Zhao,

2007; Begg et al., 2009).

The In Teria volcanic district includes about 20 dia-

treme craters, each no larger than 2 km in diameter

(Megartsi, 1972). These cross-cut horizontal

Carboniferous strata and correspond to pipe-like struc-

tures. The ejecta are cemented by calcite and include

lava blocks up to 1 m in size, as well as lapilli, olivine

and phlogopite megacrysts, and various mantle and

crustal xenoliths. Ultramafic xenoliths were collected by

J.-M. Dautria from the ejecta within a 50 km2 area. The

studied samples come from two diatremes, No. 14 and

No. 17 (Fig. 1c); No. 14 is the larger structure (2 km in

diameter), formed of hydromagmatic tuff containing

pyroxenite and mafic granulite xenoliths, phlogopite

megacrysts, phlogopitite xenoliths, spinel peridotite

xenoliths and the only garnet peridotite xenolith

sampled in the In Teria district. The second diatreme

(No. 17) is 500 m in diameter with a 5 m high ejecta rim

characterized by block and ash deposits. The ejecta in-

clude granite, pyroxenite and peridotite xenoliths,

phlogopite megacrysts and pyroclastic breccia. All

xenolith samples analyzed measured between 5 and

12 cm. Pyroxenite xenoliths represent the majority of

the xenolith population. Most pyroxenite xenoliths have

a black coating and may contain centimetre-sized

amphibole crystals. The xenoliths are embedded in a

porphyritic melilitite host lava (Megartsi, 1972), which

formed at 1000–1200�C at a pressure greater than 20

kbar (Girod, 1971; Kechid & Megartsi, 2005). Although

the age of the volcanic activity is not well established,

Mediterranean Sea

Atlan

tic O

cean

West African Craton

Ahaggar

Illizi

Sahara Basins

80

100

80

(a)

N

Algiers

(b)

300km

Saharan Meta-craton

Tamanrasset

Atakor(alk)

Illizi

AdrarN'Ajjer (alk)

Manzaz (alk)

Tahalra (alk)

Taharaq (thol)

Eggere (alk)

26º48'N

Nº14

Nº15

Nº13

Nº1726º50'N

E'04º9E'83º9

(c)

InTeria

8º30'E

24º30'N

22º30'N

26º30'N

6º30'E4º30'E

(c)

Fig. 1. (a) Map of NW Africa and the Ahaggar Shield showingthe major tectonic units and north–south-trending mega-faults.Dashed lines represent heat flow isolines of 80 and 100 mWm–2 values to the north of the Ahaggar Shield (after Takherist& Lesquer, 1989). The black rectangle corresponds to the loca-tion of (b). (b) Map showing the major faults of the AhaggarSwell and the major volcanic fields, modified after Dautria &Lesquer (1989). Dark grey is the Pan-African basement, lightgrey represents the Paleozoic sedimentary cover, black areasare volcanic districts of Late Mesozoic to Miocene–Quaternaryage, and white represents Quaternary sedimentary cover. (c)The In Teria study area (near Illizi). Dotted lines indicate num-bered diatreme craters (image from Google EarthVR ).

Journal of Petrology, 2016, Vol. 0, No. 0 3

at University of C

ambridge on A


Dow

nloaded from


the freshness of some tuff rings suggests a Quaternary

age (Conrad, 1969; Megartsi, 1972).

ANALYTICAL METHODS

Crystal preferred orientationsOlivine, pyroxene, and amphibole CPO were measured

in 19 samples by indexing of electron back-scattered

diffraction (EBSD) patterns using a JEOL JSM 5600

scanning electron microscope at Geosciences

Montpellier. Data collection, indexing and analysis of

electron backscatter diffraction patterns (EBSP) were

performed with the CHANNEL 5.10 software by Oxford

Instruments. For each sample, we obtained crystallo-

graphic orientation maps covering most of the thin sec-

tion (maps are usually 20 mm� 35 mm) with a regular

grid step ranging from 35 to 55mm depending on the

mean grain size. Average raw indexation rates were

�70%. The EBSD data were noise-reduced using a

‘wildspike’ correction to remove isolated erroneous

points, followed by a six-neighbour zero solution ex-

trapolation following standard procedures (Prior et al.,

2002; Bestmann & Prior, 2003). At each of these steps,

the resulting orientation maps were compared with

band contrast maps to ensure that the data treatment

did not compromise the data. The crystallographic

orientation data are presented as lower hemisphere

pole figures prepared using the Pf_ctf software of D.

Mainprice (http://www.gm.univ-montp2.fr/PERSO/main

price/W_data/CareWare_Unicef_Programs/). Average

Euler angles for each grain (one point per grain) were

used to avoid over-representation of large grains in thin

section. (The complete dataset is available in

Supplementary Data as Fig. A1; the supplementary data

are available for downloading at http://www.petrology.

oxfordjournals.org.) In those samples in which the foli-

ation could not be identified, the CPO data are pre-

sented with the maximum concentration of [100] axes

of olivine parallel to x (east–west) and the maximum

concentration of [010] axes of olivine parallel to z

(north–south) for easy comparison between the

samples.

Whole-rock and mineral compositionsWhole-rock major element compositions were analyzed

at Nancy SARM by inductively coupled plasma optical

emission spectroscopy (ICP-OES). Whole-rock trace

element compositions were determined by inductively

coupled plasma mass spectrometry (ICP-MS) using a

Quadrupole VG-PQ2 system at Montpellier University

(France) following the procedure described by Ionov

et al. (1992). Major elements in minerals were analyzed

by electron microprobe using the CAMECA-SX100 at

the Microsonde Sud facility (Universite de Montpellier

II) equipped with five wavelength-dispersive spectrom-

eters (Supplementary Data Table A2). Operating condi-

tions comprised an acceleration voltage of 20 kV and a

10 nA beam current. K and Na were counted for 20 s

with a 10 s background and the other elements were

counted for 30 s with a 15 s background. Mineral trace

element data were obtained by laser ablation (LA)-ICP-

MS using the facility available at the AETE platform

(Supplementary Data Table A3). The GeoLas Qþ laser

system used at Geosciences Montpellier is an Excimer

(Compex 102) operating in the deep UV (193 nm).

Ablations were performed in a pure He atmosphere

(�0�6 l min–1) using a beam diameter ranging between

30 and 120mm, with an energy density of c. 15� 10–3 J

cm–2. The LA platform is linked to an extended range

Element 2 ICP-MS system operated in low-resolution

mode at 1350 W. The ICP-MS system was daily tuned to

maximum sensitivity while keeping oxide production to

its minimum level (ThO/Th� 1%). The NIST612 glass

was used as external standard (Pearce et al., 1997) and

SiO2 or CaO contents determined by electron probe for

each mineral were used as an internal standard. Data

were processed using the GLITTER software package

(Van Achterbergh et al., 2001).

Sr–Nd–Pb isotopesSr, Nd and Pb isotopic compositions were obtained

from 400 mg of powdered whole-rock samples. After a

step of leaching with 6N HCl at 80�C for 30 min and

three cycles of rinsing with purified milli-Q H2O, sam-

ples were then dissolved for 48 h on a hot plate in a mix-

ture of HF 48% (1:2) and 13 N HNO3 (1:2). After

evaporation to dryness (120�C), 2�5 ml of 13 N HNO3

was added to the residue and kept at about 100�C for

24–48 h and evaporated. A last cycle of dissolution and

evaporation was performed with 2 ml of 13 N HNO3

(1:2). For Pb separation, after complete evaporation,

�100ml of 8 N HBr was added to the sample and kept at

90�C for 5 h before another complete evaporation. The

chemical separation of Pb was carried out using 50 ml of

anion exchange resin (AG1X8, 200–400 mesh) using

0�5 N HBr and 6 N HCl as eluants. Strontium isotopes

were separated using Sr Eichrom resin (Pin et al., 1994).

For nine samples we conducted a stronger leaching to

compare Sr isotope compositions. The first step of

leaching was with 2�5 N HCl at 95�C for 1 h 30 min fol-

lowed by three cycles of rinsing with purified Milli-Q

H2O; samples were then leached with 6 N HCl at 95�C

for 1 h 30 min followed by three cycles of rinsing with

purified Milli-Q H2O. After leaching an identical dissol-

ution procedure was applied. Neodymium isotopes

were separated during three steps: first a rare earth

element (REE) separation using AG50WX12 cation ex-

change resin, followed by two steps of Nd purification

using HDEHP columns. Total blank contents for Pb, Sr

and Nd were less than 30, 60 and 30 pg, respectively.

Lead and neodymium isotopic compositions were

measured on a VG Plasma 54 and a Nu 500 multicollec-

tor (MC)-ICP-MS system at the Ecole Normale

Superieure de Lyon. Pb isotopic compositions were

measured with an external precision of c. 100–150 ppm

for 206,207,208Pb/204Pb using the Tl normalization method


at University of C

ambridge on A


Dow

nloaded from

http://www.gm.univ-montp2.fr/PERSO/mainprice/W_data/CareWare_Unicef_Programs/

http://www.gm.univ-montp2.fr/PERSO/mainprice/W_data/CareWare_Unicef_Programs/

http://petrology.oxfordjournals.org/lookup/suppl/doi:10.1093/petrology/egw009/-/DC1


http://www.petrology.oxfordjournals.org

http://www.petrology.oxfordjournals.org




described by White et al. (2000). Every two samples

were bracketed between the NIST 981 standard. For Nd

isotopic measurements, every two samples were brack-

eted between the Lyon ‘in-house’ Nd standard (a

500 ppb dilution of the JMC commercial solution of ICP-

MS Nd standard, batch 801149A; Luais et al., 1997),

with an average 143Nd/144Nd¼ 0�512133 6 22 (2r)

(n¼ 20). Strontium isotopic compositions were meas-

ured on a Finnigan Triton TI mass spectrometer at the

‘Laboratoire de Geochimie GIS’ of Nımes. Results on

NBS 987 Sr standards yielded a mean value of87Sr/86Sr¼ 0�710254 6 11 (2r) (n¼ 18). Results on NBS

981 Pb standards yielded a mean value of208Pb/204Pb¼36�682 6 1 (2r), 207Pb/204Pb¼ 15�4874 6 4

(2r), 206Pb/204Pb¼ 16�9331 6 4 (2r) (n¼ 14). Given the

inferred Quaternary age of the In Teria magmatic dis-

trict (Conrad, 1969; Megartsi, 1972), no age correction

was performed on the measured isotopic ratios.

PETROGRAPHY AND TEXTURE OF THE

XENOLITHS

Garnet peridotiteSample Int-14-7 is a phlogopite-bearing garnet lherzo-

lite with a weakly foliated porphyroclastic texture

(Table 1, Fig. 2a). Although the petrography of this

sample was previously described by Dautria et al.

(1992), we summarize here the key characteristics of

this rock. It has a centimeter-scale compositional layer-

ing marked by alternate olivine and pyroxene enrich-

ment. Garnet has a spheroidal shape with a diameter

between 4 and 8 mm. Brown Al-spinel locally forms cor-

onas around garnet (Fig. 2a). Primary clinopyroxene

(Cpx-I) is located around garnet or is associated with

orthopyroxene; it has a grain size up to 1 mm.

Orthopyroxene porphyroclasts (Opx-I) have grain sizes

smaller than 1 mm. Olivine (Ol-I) has rounded or elon-

gated shapes (aspect ratios 4:1) with a grain size up to

1 mm. Elongated olivine grains show low-angle boun-

daries or undulose extinction perpendicular to the grain

elongation, which is in general parallel to the layering.

The primary assemblage composed of olivine, ortho-

pyroxene, clinopyroxene and garnet is overprinted by

metasomatic phlogopite (Phl-II). The occurrence of Phl-

II is irregular; it occurs either around garnet or as an

interstitial phase with a poikilitic texture (up to 2�5 mm

in size) (Fig. 2b).

Spinel peridotitesLherzolite, harzburgite and Ol-websteriteIn Teria spinel peridotites are characterized by a por-

phyroclastic texture formed by primary minerals that

Table 1: Modal composition of studied samples from In Teria

Primary phases (I) Secondary Tertiary phases (III) Melt-P

phases (II) (%)

Grt Ol Opx Cpx Sp Amph Phl Ol Cpx Sp–Chr F-K

Garnet peridotiteInt-14-7 6 49 29 10 1�5 — 4 — — 0�5 — —Spinel peridotiteLherzolite and harzburgiteErem-2 — 62 9 5 1 8 tr 6 8 1 — 15Il-17-4 — 76 6 — 1 2 tr 4�5 10 0�5 — 15Il-17-5 — 51�5 — — 0�5 8 5 15 19�5 0�5 — 35Il-17-10 — 57 7 — 1 6 4 9 15 1 — 25Il-17-13 — 68 6�5 — 0�5 5 2 7 10 1 — 18Il-17-14 — 58�5 5 — 0�5 2 9 10�5 14 0�5 — 25Il-17-18 — 67 14�5 — 0�5 3 tr 6 7 2 — 15Int-15-1 — 56 19 — 1 6 tr 6 8 2 — 16Int-17-1 — 58�5 — — 0�5 1 tr 8 30 1 1 40Int-17-6 — 79�5 — — 0�5 6 1 5 7 0�5 0�5 13Int-17-7 — 69 10 — 1 15 1 1 2 0�5 0�5 4Int-17-12 — 64�5 2 — 0�5 13 tr 5 14 0�5 0�5 20Int-17-13 — 73 2 — 1 14 tr 2 7 0�5 0�5 10Ol-websteriteInt-14-6 — 34�5 50 — 0�5 tr tr 4 7 1 3 15WehrliteIl-17-21 — 20 — — 5 1 24�5 49 0�5 — 74Int-17-14 — 7 — — 4 tr 38 50 1 — 89Ol-clinopyroxeniteIl-17-16 — — — 6 16 tr 29 48 1 — 78Pyroxenite Cpx Amph Phl SpIl-17-22 59 24 15 2Il-17-24 51 25 19 5Il-17-27 43 34 22 1Int-17-100 62 19 15 4

Modal proportions were estimated visually. Grt, garnet; Ol, olivine; Opx, orthopyroxene; Cpx, clinopyroxene; Amph, amphibole;Phl, phlogopite; Sp–Chr, spinel and chromite; F-K, alkali-feldspar; Melt-P, proportion of melt pockets; tr, trace.


at University of C

ambridge on A


Dow

nloaded from


are overprinted by two metasomatic assemblages. The

primary assemblage (I) comprises olivine, orthopyrox-

ene and spinel 6 clinopyroxene. Erem-2 is the only

spinel-bearing sample containing primary clinopyrox-

ene (Cpx-I) (Table 1, Fig. 2c). Cpx-I grains from

sample Erem-2 are anhedral, up to 0�5 cm wide, and

contain thin exsolution lamellae of orthopyroxene.

The orthopyroxenes (Opx-I) in all peridotites are

2–5 mm in size and have irregular shapes (Fig. 2d and

e). Olivine (Ol-I) occurs as elongated crystals (up to

8 mm long) that define the foliation (Fig. 2f). Olivine

grains display undulose extinction and subgrains ori-

ented perpendicular to the grain elongation (Fig. 2f).

Spinel, commonly replaced by chromite, is up to

Fig. 2. Photomicrographs in plane-polarized light (PPL) and cross-polarized light (XPL) and backscattered electron (BSE) imagesillustrating the typical microtextures of the In Teria peridotites. (a) Garnet peridotite with a layer rich in orthopyroxene; PPL. (b)Garnet with a corona essentially composed of spinel and clinopyroxene; PPL. Phlogopite (Phl-II) is poikilitic. (c) Xenolith Erem-2with primary clinopyroxene (Cpx-I), orthopyroxene (Opx-I) and olivine (Ol-I), showing reaction zones with former melt pockets; PPL.(d) Primary orthopyroxene (Opx-I) in IL-17-8 is partially replaced by amphibole-II and phlogopite-II; PPL. (e) Opx-I in the IL-17-4 is‘dismembered’ and partially replaced by crystallized melt pockets; PPL. (f) Spinel peridotite IL-17-20 characterized by tabular olivinecrystals and subgrain boundaries; XPL. Dashed line indicates foliation. (g) Coarse granular spinel peridotite IL-17-10 displays foli-ation marked by spinel and amphiboleþphlogopite. The green areas correspond to former melt pockets with clinopyroxene (Cpx-III) enrichment; PPL. Dashed line indicates foliation. (h) BSE image of IL-17-10 showing a melt pocket composed of clinopyroxene(Cpx-III), olivine (Ol-III), interstitial glass and holes (black areas) generated during polishing. (i) Melt pocket with partially orientedCpx-III in IL-17-11; PPL. It should be noted that amphibole is replaced by Cpx-III. (j) BSE image of an amphibole–clinopyroxene sym-plectite in IL-17-16. grt, garnet; cpx, clinopyroxene, opx, orthopyroxene, ol, olivine, amph, amphibole, gl, glass.


at University of C

ambridge on A


Dow

nloaded from


400mm in size and has a skeletal shape, but with align-

ment parallel to the foliation when the latter is observed

(Fig. 2g). The Ol-websterite, Int-14-6, has a texture and

grain size similar to those of the lherzolites and harz-

burgites, but is characterized by a higher proportion of

orthopyroxene (�50%, Table 1).The second mineral assemblage (II), previously

referred to as ‘metasomatic hydrous mineral phases’ by

Dautria et al. (1992), corresponds to the occurrence of

phlogopite (Phl-II) and amphibole (Amph-II) in variable

proportions (Table 1, Fig. 2b–d). Amphibole and phlogo-

pite often replace Opx-I or spinel (Fig. 2d and g). The

spinel–phlogopite–amphibole aggregates are aligned

with the foliation, but the elongation of single amphi-

bole and phlogopite grains (up to 1 mm) is not always

parallel to the foliation (Fig. 2g) and these minerals do

not display intracrystalline deformation features.

The third assemblage (III) consists of reaction aggre-

gates or patches (possible former melt-pockets), which

are present in variable proportions in the spinel perido-

tites (10–40% of the sample volume; Table 1). These re-

action aggregates are composed of microgranular

clinopyroxene, olivine, chromite, Al-spinel, sulphides,

K-feldspar and glass (Fig. 2c and e–i) and were inter-

preted by Dautria et al. (1992) as the products of carbo-

natitic melt percolation through the peridotite. Dark

green clinopyroxene (Cpx-III) and associated minerals

partially replace primary orthopyroxene (Fig. 2e) and

secondary amphibole (Fig. 2i); Cpx-III is euhedral and

has a grain size between 50 and 200 mm (Fig. 2h and i); it

may show elongated shapes parallel to orthopyroxene

or amphibole cleavages (Fig. 2i). Ol-III crystallizes as

neoblasts with subhedral shapes and displays a grain

size between 50 and 150mm (Fig. 2h).

Wehrlite and Ol-clinopyroxeniteWehrlite and Ol-clinopyroxenite form the Cpx-III-rich

peridotite xenoliths that are common at In Teria (5–10%

of collected xenoliths). Three samples were studied:

Il-17-21, Int-17-14 and Il-17-16. Wehrlites Int-17-14 and

Il-17-21 contain primary olivine (7 and 20%, respect-

ively; Table 1) with a grain size up to 1 mm and anhedral

shapes. The clinopyroxene is similar to Cpx-III observed

in the spinel peridotites. It has euhedral shapes with a

grain size between 50 and 200mm. Ol-III grain size varies

between 50 and 150mm. Amphibole and phlogopite are

occasionally found and have anhedral shapes with

grain sizes of up to 0�3 mm.The Ol-clinopyroxenite Il-17-16 is a peculiar sample,

characterized by the lack of primary olivine (Ol-I). This

sample contains clinopyroxene megacrysts with nu-

merous thin orthopyroxene exsolution lamellae and

clinopyroxene–amphibole symplectites (Fig. 2j). The

clinopyroxene megacrysts are embedded in a fine-

grained reaction matrix, similar to that which composes

Fig. 2. Continued.


at University of C

ambridge on A


Dow

nloaded from


the reaction patches in the spinel peridotites, character-

ized by a large proportion of Cpx-III and Ol-III (Table 1).

PyroxenitesPyroxenite xenoliths are more abundant (�70%) than

peridotite xenoliths at In Teria (�20%; Kechid &

Megartsi, 2005). They are characterized by a cumulate

texture, the absence of olivine and garnet, and no reac-

tion aggregates or patches like those observed in the

spinel peridotite xenoliths. In some samples subtle

layering is defined by variations in grain size. They are

mainly clinopyroxenite, composed of clinopyroxene

and phlogopite, with or without amphibole; websterites

are rare. Clinopyroxenes are zoned xenomorphic crys-

tals, measuring between 3 and 4 mm in size. Phlogopite

has a grain size of 1–3 mm. It is generally interstitial and

associated with clinopyroxene. When present, the

amphibole is millimeter-sized and is interstitial between

clinopyroxene and phlogopite.

CRYSTAL PREFERRED ORIENTATIONS

Crystal preferred orientations (CPO) of olivine, ortho-

pyroxene, clinopyroxene and amphibole were analyzed

in the garnet peridotite and 18 spinel peridotite xeno-

liths. Results for the garnet peridotite and six spinel

peridotites are shown in Fig. 3. The complete dataset is

available as Supplementary Material Fig. A1.

The garnet peridotite (Int-14-7) has an olivine CPO

characterized by a strong maximum orientation of

[010] axes close to the pole of the foliation (z) (max-

imum density MD¼ 7�86, Fig. 3a). The [001] and [100]

axes form a girdle roughly parallel to the foliation

plane (x). Both axes have weak maxima subparallel to

each other, but the [001] maximum is stronger than

the [100] maximum. This distribution characterizes a

fiber-[010] olivine CPO. Orthopyroxene and clinopyr-

oxene have similar CPO with a strong alignment of

[001] axes in the foliation plane and subparallel to the

olivine [100] maxima. The [100] and [010] axes of both

pyroxenes form incomplete girdles in the foliation

plane. The [010] axes form a maximum at high angle

to the foliation.

Most spinel peridotites show a weak preferred orien-

tation of olivine grains with a point concentration of

[100] axes parallel to the foliation plane, probably mark-

ing the lineation. The olivine [010] axes form either a

maximum perpendicular to the foliation plane or a gir-

dle normal to the [100] maximum (Fig. 3b–d and

Supplementary Material Fig. A1). In all spinel perido-

tites olivine [001] axes are more dispersed than the

other two, but they form a weak maximum in the foli-

ation generally perpendicular to the [100] maximum.

In spinel peridotites, orthopyroxene has a weak CPO,

which is characterized nevertheless by a weak [001]

maximum subparallel to the olivine [100] maximum

(e.g. Il-17-17 and Int-17-10; Fig. 3b). The multiple max-

ima present in the orthopyroxene CPO are probably

due to over-representation of some former orthopyrox-

ene porphyroclasts, which were corroded and dislo-

cated during subsequent metasomatism, now being

present as separate grains. This effect was corrected for

as much as possible during data reduction.Clinopyroxene and especially amphibole CPO in spi-

nel peridotites are strongly dispersed, but distributions

of the [100], [010] and [001] axes are consistently paral-

lel between these minerals. They display two types of

microstructural relationships with olivine and orthopyr-

oxene from the primary mineral assemblage: (1) Type

A, where despite the much more dispersed CPO, the

clinopyroxene and amphibole [001] axes maxima are

parallel to the [001] maximum of orthopyroxene and to

the [100] axes maximum of olivine (Fig. 3b); (2) Type B,

where clinopyroxene and amphibole display similar

CPO orientations, which are discordant to orthopyrox-

ene CPO (Fig. 3d). Six samples show an intermediate or-

ganization between Type A and B (Fig. 3c and

Supplementary Data Fig. A1). The high dispersion of

the CPO, as well as the Type B and intermediate textural

relationships are consistent with petrographic observa-

tions, which imply that all clinopyroxene and amphibole

in the spinel peridotites are products of metasomatism.

WHOLE-ROCK CHEMISTRY

The previous study of Dautria et al. (1992) on In Teria

xenoliths reported major and trace element data for

peridotite and pyroxenite xenoliths, as well as host

melilitite samples. Major elements were analyzed by

atomic absorption and trace elements (Hf, Ta, Th, U,

and REE) were determined by neutron activation (sam-

ples marked with asterisks in Tables 2 and 3). In this

study we report major element analyses for 10 new

samples by ICP-OES and new trace element analyses

for all samples by solution ICP-MS. Overall, 17 perido-

tite and 11 pyroxenite xenoliths, one phlogopite mega-

cryst and five host melilitites were analysed in this

study for major and trace elements. Analytical results

are given in Tables 2 and 3 and illustrated in Figs 4

and 5.

Garnet peridotiteThe garnet peridotite has low MgO (35�7 wt %) and high

Al2O3 and Fe2O3 contents (3�3 and 11�4 wt %, respect-

ively; Fig. 4, Table 2). Such a composition overlaps the

field of off-cratonic mantle xenoliths (Fig. 4a; Canil,

2004). However, comparison with the mantle melting

array shows that this garnet peridotite does not repre-

sent a melting residue (Fig. 4b and c). This sample has a

relatively flat REE pattern with (La/Lu)N¼2�81 [the sub-

script N indicates Primitive Mantle normalized values

from McDonough & Sun (1995)], a positive Ti anomaly,

and Rb–Ba enrichment relative to the REE (Fig. 5a and

b). Except for Ti, the other high field strength elements

(HFSE: Nb, Ta, Zr and Hf) do not form clear anomalies

relative to neighbouring elements.


at University of C

ambridge on A


Dow

nloaded from





(a)

Ga

rne

t p

erid

otite

MD

= 2

.91

PfJ

= 1

.60

MD

= 7

.86

PfJ

= 3

.37

MD

= 4

.24

PfJ

= 1

.66

N =

115

8

MD

= 4

.75

PfJ

= 1

.44

MD

= 5

.39

PfJ

= 1

.77

MD

= 6

.62

PfJ

= 2

.20

MD

= 3

.15

PfJ

= 1

.34

MD

= 4

.37

PfJ

= 1

.68

MD

= 5

.83

PfJ

= 2

.35

N =

965

N =

408

ZZ

ZZ

ZZ

ZZ

Z

[100

] [0

01]

[010

]

(b)

Sp

ine

l p

erid

otite

(Typ

e A

)

Clin

opyr

oxen

eelobihp

mA

enexorypohtrO

Oliv

ine

[100

] [0

01]

[010

] [1

00]

[001

] ]100[

]001[ ]010[

[010

]

Il-17

-17

N =

405

N =

192

N =

110

5N

= 3

00

MD

= 2

.86

PfJ

= 1

.40

MD

= 3

.36

PfJ

= 1

.59

MD

= 2

.17

PfJ

= 1

.12

MD

= 3

.09

PfJ

= 1

.19

MD

= 3

.76

PfJ

= 1

.27

MD

= 2

.80

PfJ

= 1

.31

MD

= 4

.74

PfJ

= 1

.61

MD

= 5

.22

PfJ

= 1

.59

MD

= 4

.87

PfJ

= 1

.79

MD

= 2

.71

PfJ

= 1

.24

MD

= 2

.86

PfJ

= 1

.25

MD

= 2

.48

PfJ

= 1

.19

X

N =

85

N =

936

N =

178

N =

541

PfJ

= 2

.34

MD

= 2

.81

PfJ

= 1

.59

MD

= 2

.69

PfJ

= 1

.26

MD

= 2

.99

PfJ

= 1

.30

MD

= 2

.78

PfJ

= 1

.26

MD

= 3

.42

PfJ

= 1

.44

PfJ

= 1

.31

MD

= 5

.98

PfJ

= 2

.02

MD

= 9

.89

PfJ

= 2

.70

MD

= 4

.86

PfJ

= 1

.60

MD

= 5

.93

PfJ

= 1

.87

MD

= 4

.40

PfJ

= 1

.91

MD

= 5

.78

MD

= 6

.23

Int-1

7-10

X

X

ZZ

Z

Int-1

4-7

Fig

.3.

Cry

sta

lp

refe

rre

do

rie

nta

tio

ns

(CP

O)

of

oliv

ine

,o

rth

op

yro

xe

ne

,cl

ino

py

rox

en

ea

nd

am

ph

ibo

lefr

om

InT

eri

ag

arn

et

pe

rid

oti

tex

en

olith

(a)

an

dsp

ine

lp

eri

do

tite

xe

no

lith

s(b

–d).

Sp

ine

lp

eri

do

tite

sa

reo

rga

niz

ed

inth

ree

gro

up

sd

efi

ne

db

yth

ere

lati

on

sb

etw

ee

nth

eo

rth

op

yro

xe

ne

,cl

ino

py

rox

en

ea

nd

am

ph

ibo

leC

PO

.(b

)T

yp

eA

,w

he

reo

liv

ine

,o

rth

op

yro

xe

ne

,cl

ino

py

rox

en

ea

nd

am

ph

ibo

lea

reco

he

ren

t;(c

)in

term

ed

iate

typ

eb

etw

ee

nT

yp

es

Aa

nd

B,

cha

ract

eri

zed

by

ap

art

ial

corr

ela

tio

nb

etw

ee

nth

eo

rth

op

yro

xe

ne

an

dth

ecl

ino

py

rox

en

ea

nd

am

ph

ibo

leC

PO

;(d

)T

yp

eB

,w

he

reo

liv

ine

an

do

rth

op

yro

xe

ne

CP

Oa

reco

he

ren

tb

ut

un

corr

ela

ted

toth

ecl

ino

py

rox

en

ea

nd

am

ph

ibo

leC

PO

.T

wo

sam

ple

so

fe

ach

typ

ea

reillu

s-tr

ate

d.

(Th

eco

mp

lete

da

tase

tis

giv

en

inS

up

ple

me

nta

ryD

ata

,F

ig.

A1

.)E

qu

al-

are

alo

we

rh

em

isp

he

rest

ere

og

rap

hic

pro

ject

ion

sin

the

stru

ctu

ral

refe

ren

cefr

am

e.

Nis

the

nu

mb

er

of

gra

ins

an

aly

zed

.C

on

tou

rsa

rea

t0�5

mu

ltip

les

of

un

ifo

rmd

istr

ibu

tio

nin

terv

als

.O

rth

op

yro

xe

ne

an

da

mp

hib

ole

ste

reo

plo

tsw

ere

con

tou

red

wh

en

mo

reth

an

10

0g

rain

sco

uld

be

me

as-

ure

din

the

thin

sect

ion

.M

Dis

ma

xim

um

de

nsi

ty;P

fJis

asc

ala

rm

ea

sure

of

the

stre

ng

tho

fth

ea

xis

ori

en

tati

on

.


at University of C

ambridge on A


Dow

nloaded from



(c)

Sp

ine

l p

erid

otite

(In

term

ed

iate

)

Il-17

-9

MD

= 4

.25

PfJ

= 2

.00

MD

= 7

.06

PfJ

= 2

.90

MD

= 4

.04

PfJ

= 1

.54

MD

= 3

.14

PfJ

= 1

.30

MD

= 2

.95

PfJ

= 1

.49

MD

= 2

.88

PfJ

= 1

.34

MD

= 5

.21

PfJ

= 1

.78

MD

= 3

.66

PfJ

= 1

.40

MD

= 5

.89

PfJ

= 1

.82

MD

= 2

.32

PfJ

= 1

.18

MD

= 2

.18

PfJ

= 1

.18

MD

= 1

.90

PfJ

= 1

.11

N =

791

N =

122

0N

= 3

60N

= 3

04

MD

= 4

.73

PfJ

= 1

.77

MD

= 3

.88

PfJ

= 1

.34

MD

= 4

.99

PfJ

= 1

.75

N =

305

Il-17

-11

MD

= 2

.83

PfJ

= 1

.49

MD

= 4

.68

PfJ

= 2

.01

MD

= 3

.61

PfJ

= 1

.42

MD

= 3

.01

PfJ

= 1

.35

MD

= 3

.01

PfJ

= 1

.34

MD

= 3

.67

PfJ

= 1

.40

N =

401

N =

104

3N

= 2

17

MD

= 3

.86

PfJ

= 1

.32

MD

= 3

.61

PfJ

= 1

.57

MD

= 4

.12

PfJ

= 1

.46

X X

(d)

Sp

ine

l p

erid

otite

(Typ

e B

)

Il-17

-12

Il-17

-13

MD

= 3

.89

PfJ

= 1

.75

MD

= 4

.71

PfJ

= 1

.96

MD

= 2

.82

PfJ

= 1

.29

MD

= 4

.93

PfJ

= 1

.69

MD

= 3

.61

PfJ

= 1

.32

MD

= 4

.50

PfJ

= 1

.73

MD

= 3

.01

PfJ

= 1

.51

MD

= 3

.86

PfJ

= 1

.75

MD

= 2

.87

PfJ

= 1

.18

MD

= 4

.55

PfJ

= 1

.59

MD

= 4

.99

PfJ

= 1

.41

MD

= 3

.48

PfJ

= 1

.47

MD

= 2

.53

PfJ

= 1

.25

MD

= 3

.19

PfJ

= 1

.27

MD

= 3

.04

PfJ

= 1

.34

MD

= 2

.67

PfJ

= 1

.26

MD

= 3

.21

PfJ

= 1

.34

MD

= 3

.05

PfJ

= 1

.53

MD

= 2

.64

PfJ

= 1

.30

MD

= 2

.19

PfJ

= 1

.13

MD

= 1

.98

PfJ

= 1

.16

MD

= 3

.36

PfJ

= 1

.45

MD

= 3

.42

PfJ

= 1

.47

MD

= 3

.24

PfJ

= 1

.48

N =

162

N =

404

N =

277

N =

226

N =

327

N =

135

7

N =

211

N =

952

X X

[100

] [0

01]

[010

] C

linop

yrox

ene

elobihpm

Aenexorypohtr

OO

livin

e[1

00]

[001

] [0

10]

[100

] [0

01]

]100[ ]001[

]010[[0

10]

ZZ

ZZ

ZZ

ZZ

ZZ

ZZ

Fig

.3.

Co

nti

nu

ed

.


at University of C

ambridge on A


Dow

nloaded from


Tab

le2:

Ma

jor

(wt

%)

wh

ole

-ro

cka

na

lyse

so

fp

eri

do

tite

,p

yro

xe

nit

ex

en

olith

sa

nd

me

liliti

tes

Grt

Sp

ine

lp

eri

do

tite

s

pe

rid

ote

lhe

rzo

lite

sa

nd

ha

rzb

urg

ite

s

Ol-

we

bst

eri

tew

eh

rlit

eO

l-cl

ino

-p

yro

xe

nit

eS

am

ple

:In

t-1

4-7

*E

rem

-2*

Il-1

7-4

Il-1

7-5

Il-1

7-1

3Il-1

7-1

4Il-1

7-1

8In

t-1

5-1

*In

t-1

7-1

*In

t-1

7-6

*In

t-1

7-7

*In

t-1

7-1

2*

Int-

17

-13

*In

t-1

4-6

*ll-1

7-2

1In

t-1

7-1

4*

Il-1

7-1

6(-

2-)

(-9

-)(-

3-)

(-4

-)(-

5-)

(-6

-)(-

7-)

(-8

-)(-

1-)

SiO

24

6�1

44�7

43�0

43�0

44�9

44�9

44�0

44�2

41�3

42�6

44�4

43�5

45�1

34

9�8

49�3

50�4

48�1

Al 2

O3

3�2

82�1

51�7

72�1

32�4

02�0

22�0

41�9

01�3

70�6

02�1

82�0

82�5

40�9

83�1

93�5

55�0

1F

eO

10�3

7�5

78�3

88�2

17�7

07�4

57�6

07�6

08�5

18�0

17�4

77�9

67�9

88�6

04�8

84�8

26�6

1M

nO

0�1

50�1

20�1

50�1

60�1

60�1

30�1

30�1

20�1

50�1

40�1

20�1

30�1

20�1

70�1

10�1

10�1

3M

gO

35�8

41�1

40�6

40�4

38�9

39�4

41�1

42�4

44�2

45�7

42�2

40�4

39�4

37�3

28�1

25�5

21�2

Ca

O2�2

32�2

62�7

42�6

92�6

32�8

92�1

81�7

31�8

21�3

71�8

52�9

32�7

91�0

28�5

09�4

61

2�3

Na

2O

0�2

70�3

31�0

81�2

11�1

01�2

10�6

90�3

90�9

00�4

90�7

31�1

30�8

60�7

22�7

63�1

42�7

2K

2O

0�3

50�0

30�1

30�1

20�3

10�1

00�0

60�0

20�1

40�0

90�1

00�3

50�1

00�2

40�1

90�2

30�2

8T

iO2

0�5

30�0

70�0

70�0

60�0

70�0

40�0

60�0

70�0

60�0

30�0

70�1

10�1

10�0

70�1

40�1

40�2

6P

2O

50�0

30�0

30�0

40�0

40�0

40�0

70�0

30�0

30�0

40�0

40�0

40�0

30�0

30�0

30�0

30�0

80�0

6T

ota

l1

00�2

99�1

99�1

98�9

99�1

99�0

98�9

99�2

99�4

10

0�0

99�9

99�2

99�1

99�8

98�8

99�0

98�8

Mg

#.

0�8

60�9

10�9

00�9

00�9

00�9

10�9

10�9

10�9

00�9

10�9

10�9

00�9

50�8

90�9

10�9

10�8

5

Py

rox

en

ite

sM

eliliti

tes

Sa

mp

le:

Il-1

7-2

2Il-1

7-2

4Il-1

7-2

7In

t-1

7-1

00

*In

t-1

4-1

00

*In

t-1

4-1

01

*In

t-1

7-1

01

*In

t-1

7-1

02

*In

t-1

7-1

03

*In

t-1

7-1

04

*In

t-1

7-1

05

*8

81

4*

88

15

*8

81

6*

88

17

*8

81

8*

SiO

24

1�0

40�0

39�0

40�7

36�6

40�3

45�1

39�9

41�9

43�0

41�0

33�8

33�4

32�3

33�4

34�8

Al 2

O3

11�6

9�9

01

0�5

8�9

06�9

59�4

09�3

31

0�0

01

0�9

59�5

31

0�0

87�3

47�0

77�0

07�3

47�2

2F

eO

10�6

12�5

13�7

12�8

21�8

13�8

7�9

13�1

10�5

11�9

12�0

12�4

12�1

12�3

11�0

11�1

Mn

O0�1

10�1

30�1

20�1

20�1

50�1

10�1

00�1

10�1

10�1

40�1

10�2

30�2

30�2

30�2

20�2

3M

gO

12�2

14�2

12�9

12�7

11�5

16�9

14�3

13�2

13�6

12�3

12�7

14�4

14�2

14�5

15�2

15�3

Ca

O1

4�5

10�6

12�0

13�1

14�0

6�0

14�9

10�7

12�5

14�4

13�2

16�4

16�1

16�1

15�5

16�3

Na

2O

1�6

71�3

81�3

91�3

0�9

40�9

30�9

71�5

21�2

21�4

71�7

02�7

42�5

02�5

52�3

21�5

5K

2O

2�2

43�2

42�3

51�6

70�8

75�6

52�4

52�4

63�0

01�2

71�5

21�7

92�1

02�1

51�3

11�4

5T

iO2

3�7

75�7

65�3

86�6

25�0

53�9

73�2

07�1

83�5

44�4

75�5

53�7

33�6

03�8

03�0

03�0

0P

2O

50�2

60�0

40�0

50�0

20�0

80�0

30�0

40�0

20�0

20�0

20�0

31�3

51�3

91�3

51�1

51�0

7L

OI

1�5

01�1

41�1

00�0

20�0

0�8

0�0

0�4

1�1

0�2

60�5

44�5

85�8

55�7

77�5

16�1

2T

ota

l1

00�6

10

0�5

10

0�0

99�5

10

0�3

99�4

99�1

10

0�7

99�6

10

0�1

99�7

10

0�2

99�8

99�3

99�1

99�4

Mg

#.

0�6

70�6

70�6

30�6

40�4

80�6

80�7

60�6

40�7

00�6

50�6

50�6

70�6

80�6

80�7

10�7

1

*M

ajo

re

lem

en

tw

ho

le-r

ock

an

aly

ses

fro

mD

au

tria

et

al.

(19

92

)a

nd

un

pu

blish

ed

da

tafr

om

J.-

M.D

au

tria

.(-

1-)

corr

esp

on

ds

tosa

mp

len

um

be

rin

gfr

om

Da

utr

iae

ta

l.(1

99

2).


at University of C

ambridge on A


Dow

nloaded from


Table 3: Trace element (ppm) whole-rock analyses of peridotite, pyroxenite xenoliths and melilitites by solution ICP-MS

Garnet Spinel peridotiteperidotite Lherzolite and harzburgite

Sample: Int-14-7 Erem-2 Il-17-4 Il-17-5 Il-17-13 Il-17-14 Il-17-18 Int-15-1 Int-17-1 Int-17-6 Int-17-7 Int-17-12 Int-17-13(-2-) (-9-) (-3-) (-4-) (-5-) (-6-) (-7-) (-8-)

Rb 6�47 0�17 1�58 0�96 5�03 0�68 0�76 0�23 1�21 0�67 0�81 4�67 0�82Sr 23�2 45�9 143 107 107 128 70�2 32�6 82�4 44�7 92�0 83�7 110Y 2�50 2�50 3�77 2�95 4�60 4�96 2�50 2�50 2�50 2�50 2�68 3�72 2�88Zr 12�9 1�77 6�33 13�5 6�48 2�51 3�54 1�66 8�19 16�0 4�69 6�74 1�86Nb 1�35 0�37 22�4 2�80 21�2 27�0 4�09 0�26 19�7 13�1 10�2 21�1 7�38Cs 0�06 0�01 0�03 0�02 0�22 0�02 0�03 0�01 0�02 0�03 0�03 0�21 0�02Ba 65 33 64 32 336 97 63 98 38 37 108 255 31La 1�21 2�37 7�87 4�57 7�97 10�2 4�27 1�86 3�08 1�91 5�93 5�53 5�97Ce 2�72 3�03 17�9 10�1 20�4 26�7 7�92 2�03 7�90 4�83 12�8 14�6 11�3Pr 0�35 0�25 1�95 1�24 2�38 3�12 0�81 0�14 1�05 0�62 1�47 1�85 1�08Nd 1�63 0�85 7�40 5�32 9�61 12�6 2�99 0�48 4�54 2�62 5�68 7�57 3�63Sm 0�38 0�17 1�19 1�03 1�60 2�00 0�45 0�12 0�83 0�48 0�87 1�29 0�52Eu 0�14 0�07 0�38 0�35 0�52 0�66 0�16 0�05 0�27 0�15 0�28 0�41 0�18Gd 0�46 0�25 0�98 0�96 1�30 1�55 0�42 0�19 0�70 0�39 0�71 1�04 0�50Tb 0�08 0�05 0�14 0�13 0�18 0�21 0�06 0�04 0�09 0�05 0�09 0�14 0�08Dy 0�53 0�35 0�79 0�75 1�08 1�14 0�42 0�27 0�53 0�27 0�56 0�82 0�54Ho 0�11 0�08 0�14 0�13 0�19 0�20 0�09 0�06 0�09 0�04 0�11 0�15 0�11Er 0�30 0�23 0�37 0�31 0�50 0�50 0�24 0�17 0�21 0�10 0�28 0�37 0�32Tm 0�04 0�03 0�05 0�04 0�07 0�07 0�03 0�03 0�03 0�01 0�04 0�05 0�05Yb 0�27 0�23 0�33 0�24 0�43 0�43 0�23 0�17 0�18 0�08 0�26 0�32 0�32Lu 0�05 0�04 0�05 0�04 0�07 0�07 0�04 0�03 0�03 0�01 0�04 0�05 0�05Hf 0�42 0�07 0�13 0�14 0�11 0�03 0�08 0�06 0�11 0�18 0�09 0�15 0�07Ta 0�10 0�01 0�40 0�88 0�79 0�46 0�12 0�01 0�79 0�98 0�25 0�76 0�10Pb 0�31 1�20 0�67 0�37 0�47 0�53 0�65 0�96 0�52 0�6 3�12 0�62 0�80Th 0�15 0�73 9�04 0�36 1�36 0�78 0�79 0�57 0�34 0�11 1�07 1�10 0�57U 0�04 0�23 0�23 0�07 0�58 0�18 0�20 0�16 0�08 0�03 0�34 0�18 0�23

Spinel peridotite Pyroxenite Melilitite

Ol- Ol-clino- wehrlitewebsterite pyroxenite

Sample: Int-14-6 Il-17-16 ll-17-21 Il-17-22 Il-17-24 Il-17-27 Int-17-100 8814 8815 8816(-1-)

Rb 2�76 1�60 0�56 35�0 78�0 46�0 45�0 50�0 56�0 41�0Sr 68�9 306 333 283 142 142 174 1292 1209 1106Y 2�50 10�30 7�73 8�96 5�15 7�40 6�03 24�0 23�0 22�8Zr 1�99 18�6 8�10 186 218 429 107 359 333 317Nb 21�0 41�1 34�5 23�0 26�3 28�1 20�2 117 113 107Cs 0�10 0�02 0�01 0�32 0�80 0�76 0�48 0�46 0�60 0�46Ba 131 53 140 450 683 323 467 904 853 834La 7�84 15�7 10�6 11�3 3�30 3�17 3�53 90�0 87�0 89�5Ce 12�5 39�3 26�6 31�1 9�59 10�1 10�0 182 176 180Pr 1�21 4�82 3�07 4�29 1�50 1�71 1�61 20�0 19�0 19�6Nd 4�19 20�7 12�2 21�3 8�15 9�62 8�55 80�4 78�0 80�2Sm 0�57 3�79 2�13 4�56 2�05 2�58 2�16 13�3 13 13�3Eu 0�16 1�24 0�85 1�44 0�67 0�89 0�75 3�96 3�92 4�00Gd 0�38 3�26 1�88 4�09 2�09 2�8 2�26 10�6 10�4 10�7Tb 0�05 0�44 0�28 0�50 0�27 0�38 0�3 1�26 1�25 1�26Dy 0�31 2�49 1�69 2�57 1�47 2�06 1�62 6�34 6�22 6�25Ho 0�05 0�44 0�31 0�40 0�24 0�34 0�26 1�00 0�97 0�97Er 0�14 1�06 0�79 0�85 0�51 0�74 0�57 2�19 2�16 2�12Tm 0�02 0�14 0�11 0�10 0�06 0�09 0�06 0�26 0�25 0�25Yb 0�13 0�85 0�68 0�57 0�34 0�49 0�35 1�52 1�48 1�48Lu 0�02 0�13 0�10 0�08 0�05 0�07 0�05 0�22 0�21 0�21Hf 0�03 0�33 0�20 5�29 6�07 10�6 4�19 6�37 6�21 5�98Ta 0�48 1�46 0�77 2�12 3�04 3�29 2�58 7�56 7�76 7�70Pb 0�55 1�04 1�66 0�75 0�44 0�42 0�45 4�58 4�51 4�25Th 2�17 5�43 0�84 0�99 0�25 0�21 0�28 11�6 11�1 11�4U 0�18 0�46 0�27 0�21 0�06 0�08 0�09 3�18 2�95 2�84

Sample numbering as in Table 2.


at University of C

ambridge on A


Dow

nloaded from


Spinel peridotiteLherzolite, harzburgite and Ol-websteriteLherzolite, harzburgite and one orthopyroxene-rich peri-

dotite (sample Int-14-6) have Mg# varying from 88�5 to

91�0, whereas Al2O3 and CaO contents vary between

0�98 and 2�54 wt %, and 1�02 and 2�7 wt %, respectively.

The whole-rock Ca/Al ratio is highly variable and

reaches values as high as 3�1 (sample IL-17-4). The ma-

jority of the In Teria lherzolites and harzburgites form a

group within the off-cratonic mantle array (Fig. 4a) with

compositions comparable with those of the most fertile

Cap Verde peridotite xenoliths (Bonadiman et al., 2005)

(Fig. 4b and c). Only three samples show distinct com-

positions. Two samples (Int-17-1 and Int-17-6), are more

refractory and plot within or close to the field of Canary

Island peridotite xenoliths (Neumann et al., 2004; Fig.

4b and c); the Ol-websterite (sample Int-14-6) has low

Al2O3, CaO and MgO contents compared with the other

peridotites (Fig. 4).

Lu varies from 0�012 to 0�13 ppm (Table 3), and is

positively correlated with Al2O3 and negatively with

MgO. Although showing variable REE contents and

distributions, all spinel peridotites are more enriched

in light REE (LREE) than the garnet peridotite (Fig. 5a

and c). On the basis of normalized REE patterns three

subgroups can be distinguished (Fig. 5). The first

group (group 1) comprises most of the lherzolites and

harzburgites and is characterized by almost straight

positive slope from heavy REE (HREE) to LREE (Fig.

5c). The other two groups (groups 2 and 3) are distin-

guished by concave-upwards REE patterns. The se-

cond group comprises the lherzolites Erem-2 and Int-

15-1; Erem-2 is the sample containing relicts of pri-

mary clinopyroxene. These two samples have a

roughly flat to slightly middle REE (MREE)-depleted

MREE–HREE segment [0�67< (Eu/Lu)N< 0�74] and are

enriched in LREE relative to MREE [(La/Sm)N� 8]. The

third group includes the lherzolite samples IL-17-18

and Int-17-13 and the Ol-websterite sample Int-14-6.

These samples are distinguished from the second

group by positive (Eu/Lu)N ratios, between 1�5 and 3�3,

and lesser LREE enrichment [6< (La/Sm)N� 8). On

Primitive Mantle (PM)-normalized diagrams group 1

peridotites are mostly characterized by negative

anomalies of Zr, Hf, and Ti relative to the MREE.

Conversely, several samples show positive anomalies

of Nb and Ta relative to Th, U, and LREE; the anoma-

lies tend to be marked in samples with lower Th, U,

and LREE contents. Except for a positive spike in Th in

one sample, group 1 peridotites also tend to be

depleted in Th and U relative to Ba and LREE. Group 2

samples show markedly distinct trace element pat-

terns, characterized by a strong negative anomaly in

Nb and Ta, whereas Zr, Hf, and Ti do not show signifi-

cant anomalies. Group 2 samples are also distin-

guished by selective enrichment of Th, U, Pb, and Sr.

In contrast, group 3 is more akin to group 1, particu-

larly in showing negative anomalies of Zr, Hf, and Ti,

and lacking substantial anomalies of Nb and Ta.

Wehrlite and Ol-clinopyroxenitesThe wehrlite and Ol-clinopyroxenites (Il-17-21, Int-17-14

and Il-17-16) have low bulk-rock MgO contents (21�2–

28�1 wt %), relatively high Al2O3 (3�19–5�01 wt %), and

elevated CaO contents (8�50–12�3 wt %; Fig. 4a, Table 2).

0

2

4

6

20 25 30 35 40 45 50 55

Ahaggar

Canary

Cape-Verde

(a)

Int-14-6

Canary

Ahaggar

Cape-Verde

(c)PM

Int-14-6

off-craton

cratonInt-14-6

IL-17-21

Int-17-14

Int-17-16

melting-trend

(b)grt. peridotitesp. peridotiteErem-2Wehrlite & Ol-cpx

0

2

4

35 40 45 50

CaO

(wt%

)A

l2O3 (

wt%

)

MgO (wt%)

Al2O

3 (w

t%)

0

1

2

3

4

35 40 45 50

PM

Fig. 4. Whole-rock major element compositions of peridotitexenoliths from In Teria compared with (a) off-craton xenoliths(small grey dots) and on-craton mantle xenoliths (small blackdots) from worldwide localities [data from Canil (2004)], and (b,c) the residual trend after melt extraction (black) from a fertilelherzolite source (PM) (Niu et al., 1997; Beccaluva et al., 2007).Fields for Cape Verde xenolith (Bonadiman et al., 2005);Tenerife, Lanzarote and Hierro xenoliths from the CanaryIslands (Neumann et al., 2004); and xenoliths from the Ahaggardistrict (Beccaluva et al., 2007) are shown for comparison.These xenoliths are anhydrous peridotites and metasomatizedspinel lherzolites and harzburgites. PM, Primitive Mantle com-position from McDonough & Sun (1995). grt, garnet; sp, spinel;Ol-cpx, Ol-clinopyroxenite.


at University of C

ambridge on A


Dow

nloaded from


Their normalized REE and trace element patterns

(Fig. 5c and d) are almost indistinguishable from perido-

tite group 1 patterns, except for higher concentrations

of most elements.

PyroxeniteThe pyroxenites contain between 6�95 and 11�6 wt %

Al2O3 and between 6�0 and 14�9 wt % CaO, and their

MgO ranges between 11�5 and 16�9 wt % (Table 2). They

PM

nor

mal

ized melilitite

grt peridotite

PM

nor

mal

ized

pyroxenite

spinel peridotite

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu Gd

TbDy

HoEr

TmYb

Lu

0.1

1

10

100

1000

0.1

1

10

100

1000

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

0.1

1

10

100

1000


Ch

norm

aliz

edC

h no

rmal

ized

spinel peridotite

melilitite

grt peridotite

pyroxenite

(a)

(c)

(g)

(b)

(d)

(h)

(f)

PM

nor

mal

ized

PM

nor

mal

ized

Ch

norm

aliz

edC

h no

rmal

ized


spinel peridotitespinel peridotite

0.1

1

10

100

1000


0.1

1

10

100

1000

0.1

1

10

100

1000

0.1

1

10

100

1000

0.1

1

10

100

1000

Ti

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu Gd

TbDy

HoEr

TmYb

LuTi

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu Gd

TbDy

HoEr

TmYb

LuTi

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu Gd

TbDy

HoEr

TmYb

LuTi

group 1

group 2

group 3

(e) Erem-2

Ol-cpx (Il-17-16)

Il-17-22

wehrlite (Il-17-21)

Fig. 5. Whole-rock REE and trace element compositions of (a, b) host melilitite and garnet peridotite; (c–f) spinel peridotite and(g, h) pyroxenite xenoliths. Spinel peridotites are separated into three groups that have different chemical compositions (see textfor discussion). Ol-clinopyroxenites and wehrlite are shown by open triangles (c, d) and lherzolite Erem-2 (with primary clinopyrox-ene) by grey filled circles (e, f). REE and whole-rock trace element contents are normalized to chondrite (Ch) and Primitive Mantle(PM), respectively; normalizing values after McDonough & Sun (1995).


at University of C

ambridge on A


Dow

nloaded from


show convex-upward REE patterns, characterized

by positive slopes from the HREE to the MREE [5�6<Eu/

Lu)N< 7�4] and a roughly flat LREE segment [(La/

Sm)N¼ 0�77–1�55; Fig. 5g]. The PM-normalized trace

element patterns are distinguished by positive HFSE

anomalies in Nb, Ta, and Ti in all samples, and Zr–Hf in

two samples. The pyroxenites are also distinctively en-

riched in Rb and Ba relative to Th and U (Fig. 5h).

MelilititeThe compositions of the five analysed host-rock melili-

tites are homogeneous (Table 2). The samples are

undersaturated in SiO2 (�33�5 wt %) and have low total

alkali (K2OþNa2O) contents varying from 3�0 to 4�7 wt

%. The high Mg (MgO¼ 14�2–15�3 wt %), high Ni and Cr

contents indicate that these lavas have not been exten-

sively modified by fractional crystallization (Dautria

et al., 1992). The chondrite-normalized REE patterns

show a smooth positive slope from HREE to LREE

[48� (La/Yb)N� 70; Fig. 5a]. The PM-normalized trace

element patterns are convex upwards, with high but

relatively unfractionated abundances of the most in-

compatible elements (Rb to LREE), except for Pb, which

shows a negative anomaly relative to the LREE [(Pb/

Ce)N¼0�28]. Among the HFSE, only Zr and Hf show a

noticeable negative anomaly.

MINERAL CHEMISTRY

Major elementsGarnet peridotiteGarnet has a homogeneous, pyrope-rich composition

(Mg#¼0�80; Cr#¼ 0�04, Supplementary Data Table A2);

TiO2 increases from core (0�19 wt %) to rim (0�35 wt %).

Clinopyroxenes have high TiO2 contents (1�16 wt %) and

show some compositional similarities to secondary

clinopyroxenes from Manzaz in the Ahaggar district

(Beccaluva et al., 2007) and primary clinopyroxenes

from Cape Verde (Bonadiman et al., 2005). The ortho-

pyroxene porphyroclasts (Opx-I) are zoned; for in-

stance, Al2O3 increases from core to rim (3�10 to 4�09 wt

%, respectively; Supplementary Data Table A2), as does

Mg# (from 0�87 to 0�89). Olivine has a homogeneous

composition with low Mg#¼ 0�86. The phlogopite (Phl-

II) has Mg# 0�86 and a high TiO2 content (5�50 wt %,

Supplementary Data Table A2).

Using garnet–pyroxene geothermobarometry (Nickel

& Green, 1985; Brey & Kohler, 1990) core equilibrium

temperature and pressure have been estimated at

1050–1100�C and 2�5–2�7 GPa, respectively (Dautria

et al. 1992). However, pyroxene and garnet rim compos-

itions yield temperatures and pressures of 1210–1240�C

and 2�6–2�7 GPa.

Spinel peridotiteLherzolite, harzburgite and Ol-websterite. Primary clino-

pyroxenes (Cpx-I) from lherzolite Erem-2 have high

Al2O3 (6�78 wt %) and Na2O (1�74 wt %; Supplementary

Data Table A2) contents and resemble primary clinopyr-

oxene from the Ahaggar district (Beccaluva et al., 2007)

(Fig. 6a). The primary orthopyroxene (Opx-I) Mg# varies

from 0�87 to 0�91, and its Al2O3 content varies from 0�53

to 4�86 wt %. Ol-I have Mg# between 0�89 and 0�94 and

low CaO contents (0�2 wt %). In the secondary mineral

assemblage (II), the amphibole (Amph-II) is pargasite to

edenite with very variable Mg# (between 0�89 and 0�96)

and Na2O (3�30–5�10 wt %) contents, and low contents

of TiO2 (0�15–0�35 wt %) and K2O (0�06–0�60 wt %). The

phlogopite (Phl-II) shows less variability (Mg#¼ 0�92–

0�93) and is TiO2-poor (0�27–0–28 wt %) and K2O-rich

(6�8–7�0 wt %). Clinopyroxenes from the third assem-

blage (Cpx-III, Fig. 6a) have Mg# between 0�92 and 0�93,

high CaO (18�5–23�5 wt %), low but variable Al2O3 (0�83–

3�09 wt %), and low TiO2 (0�08–0�25 wt %) contents.

These compositions are similar to those of finely disse-

minated secondary clinopyroxene replacing orthopyr-

oxene in Cape Verde mantle xenoliths (Bonadiman

et al., 2005) and secondary large poikilitic clinopyroxene

from the Canary Islands (Neumann et al., 2002; Fig. 6a).

Ol-III displays a large variation in Mg# (0�89–0�94), en-

compassing the composition of Ol-I; however, Ol-III has

on average a more Mg-rich composition (Mg# 0�92)

Clinopyroxene

Olivine

(a)

(b)

17

19

21

23

25

0 2 4 6 8

CaO

(wt%

)

Al2O3 (wt%)

0

0.1

0.2

0.3

0.4

0.85 0.87 0.89 0.91 0.93 0.95 0.97

CaO

(wt%

)

Mg #

Cape VerdeCpx-II

Cpx-III

grt peridotite

Cpx-I (Erem-2)

Cape Verde

spinel peridotites

Ol-III

grt peridotite

Ol-III (Il-17-16)

Ol-Ispinel peridotites

Cpx-III (Il-17-16)Cpx-I (Il-17-16)

Cpx-I (Canary)

Ahaggar

Cape Verde

Ahaggar

Canary

AhaggarCpx-II

Fig. 6. Major element composition: Al2O3, CaO (wt %) and Mg#of (a) clinopyroxene and (b) olivine. Represented data are from11 samples analyzed in this study and three samples fromJ.-M. Dautria (unpublished data). Fields of olivine and clinopyr-oxene from Cape Verde (Bonadiman et al., 2005), Manzaz,Ahaggar (Beccaluva et al., 2007) and the Canary Islands(Neumann et al., 2004) are shown for comparison. Grt, garnet.Cpx-II indicates secondary clinopyroxenes from other studies.


at University of C

ambridge on A


Dow

nloaded from







than Ol-I (Mg# 0�90, Fig. 6b). Olivine neoblasts within re-

action patches have high CaO contents (0�12–0�24 wt %)

compared with Ol-I CaO contents (0�01–0�03 wt %). The

glass analysed within the melt pockets is siliceous (up

to 68 wt % SiO2) and peraluminous (up to 25 wt %

Al2O3) in composition, and has low MgO, FeO

and CaO contents, resembling feldspar in its compos-

ition. This glass may have formed at the expense of

feldspar (Dautria et al., 1992). Its composition is similar

to that observed in many mantle xenolith suites (e.g.

Frey & Green, 1974; Ionov et al., 1993b; Coltorti et al.,

2000).Using the Brey & Kohler (1990) geothermometer cali-

bration on Opx-I and Cpx-I (sample Erem-2), Dautria

et al. (1992) obtained a core equilibrium temperature

between 770 and 870�C and a rim temperature increas-

ing up to 1080�C. According to the regional geotherm

(Lesquer et al., 1990), such temperatures would yield a

pressure range between 1�6 and 2�1 GPa. However the

absence of garnet indicates that pressure should be

lower than 1�9 GPa (Green & Ringwood, 1967; Wallace

& Green, 1988).

Wehrlite and Ol-clinopyroxenite. Primary olivine in

wehrlite has the same composition as Ol-I in spinel

peridotite with Mg# 0�89–0�90 and 0�01–0�03 wt % CaO

(Supplementary Data Table A2). The amphibole (Amph-

II) has a similar composition to Amph-II in the spinel

peridotites with Mg# 0�89, Na2O of 3�87 wt %, and low

TiO2 (0�27 wt %) and K2O (0�35 wt %) contents.

0.01

0.1

1

10

100

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

Spinel peridotite

Garnet peridotite

PM

nor

mal

ized

Pyroxenite

Spinel peridotite

Garnet peridotite

Pyroxenite

etipogolhPenexoryponilC

PM

nor

mal

ized

PM

nor

mal

ized

PM

nor

mal

ized

PM

nor

mal

ized

PM

nor

mal

ized

Cpx-III

(a)

(b)

(c)

(d)

(e)

(f)

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

0.0001

0.001

0.01

0.1

1

10

100

1000

0.0001

0.001

0.01

0.1

1

10

100

1000

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

0.01

0.1

1

10

100

Cpx-I(Erem-2)

0.01

0.1

1

10

100

0.0001

0.001

0.01

0.1

1

10

100

1000

megacryst

Fig. 7. Primitive Mantle normalized trace element patterns of clinopyroxene and phlogopite in the garnet peridotite, spinel perido-tites and pyroxenite xenoliths from In Teria. Values are normalized to the Primitive Mantle of McDonough & Sun (1995). Each pat-tern represents an average of several analyses in different grains (three on average). Error bars represent one standard deviationand are reported for representative samples. In (f) the phlogopite megacryst was analysed by solution ICP-MS.


at University of C

ambridge on A


Dow

nloaded from



Ol-clinopyroxenite Il-17-16 contains clinopyroxene

phenocrysts and clinopyroxene in symplectites, which

have low Al2O3 (0�80–1�13 wt %) and high CaO (22�8–

23�4 wt %) contents (Fig. 2j; Supplementary Data Table

A2). Cpx-III forming the fine-grained reaction matrix dis-

plays a similar composition to that in other spinel peri-

dotites. This sample contains olivine neoblasts (Ol-III)

with a high CaO content (0�23 wt %) as observed in Ol-III

from spinel peridotites, but with lower Mg# (0�89)

(Fig. 6b).

PyroxeniteThe clinopyroxene porphyroclast cores are homoge-

neous in composition with Mg# from 0�77 to 0�78.

Phlogopite has Mg# varying between 0�73 and 0�74, and

TiO2 between 2�77 and 6�18 wt %, which is the typical

composition for upper mantle phlogopite (e.g. Delaney

et al., 1980; Ionov et al., 1997). Two distinct phlogopite

zoning trends from core to rim were observed: the first

shows a decrease in TiO2 (6�18–5�58 wt %) associated

with an increase in Al2O, Na2O and Mg# (e.g. sample

Int-17-89-100), and the second an increase in TiO2 from

2�77 to 4�76 wt % (sample Il-17-27). The amphiboles

have Mg# 0�70–0�72 and high K2O contents from 2�34 to

2�56 wt %. Temperature estimates for crystallization of

the pyroxenites range between 1100 and 1180�C

(Boissiere & Megartsi, 1982).

Trace elementsGarnet peridotiteGarnet has a high HREE content (LuN¼ 9�27) and a

strong LREE depletion relative to the HREE and MREE

[(La/Lu)N¼ 0�002; Supplementary Data Table A3].

Clinopyroxene has a convex PM-normalized REE pat-

tern, in which the MREE are enriched relative to the

HREE [(Eu/Lu)N¼7�1; Fig. 7a] and LREE [(La/

Sm)N¼ 0�56]. It is characterized by relatively unfractio-

nated abundances of LILE except for Pb, which shows a

negative anomaly [(Pb/Ce)N¼ 0�11]. The HSFE show

barely detectable positive anomalies [e.g. (Nb/

Ce)N¼0�16]. Orthopyroxene REE patterns display a rela-

tively flat HREE to MREE segment [(Eu/Lu)N¼1�01] and

a negative slope from MREE to LREE [(La/Sm)N¼0�13]

with well-marked Zr–Hf and Ti anomalies [(Zr/

Sm)N< 3�14; Fig. 8a]. Olivine displays enrichment from

Dy to Lu [(Dy/Lu)N¼ 0�19]. Phlogopite is characterized

by low REE abundances [(Lu)N¼ 0�010; (La)N¼ 0�028],

but prominent positive HFSE anomalies [(Nb/

La)N¼422; (Zr/Sm)N¼ 27�8; Fig. 7d] and strong Cs, Rb

and Ba enrichment, but low U–Th content

(Supplementary Data Table A3).

Spinel peridotiteLherzolite, harzburgite and Ol-websterite. The primary

clinopyroxene from sample Erem-2 displays a LREE-

PM

nor

mal

ized

spinel peridotite

garnet peridotite spinel peridotite

pyroxenite

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

PM

nor

mal

ized

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

RbBa

ThU

NbTa

LaCe

PbPr

SrNd

ZrHf

SmEu

TiGd

TbDy

HoY

ErTm

YbLu

PM

nor

mal

ized

Orthopyroxene Amphibole(a) (c)

(b) (d)

PM

nor

mal

ized

0.001

0.01

0.1

1

10

Erem-2

0.1

1

10

100

0.1

1

10

100

0.001

0.01

0.1

1

10

100

Fig. 8. Primitive Mantle normalized trace element patterns of orthopyroxene from garnet and spinel peridotites and amphibolefrom spinel peridotites and pyroxenites. Values are normalized to the Primitive Mantle (PM) values of McDonough & Sun (1995).Each pattern represents the average of several analyses (three on average). Error bars represent one standard deviation and are re-ported for representative samples.


at University of C

ambridge on A


Dow

nloaded from






depleted PM-normalized trace element pattern

[1�5� (La/Yb)N�4�3; Fig. 7b]. Such REE patterns can be

interpreted in terms of melt extraction after 7–12% frac-

tional or batch partial melting, respectively [estimated

using modal Cpx: 20% initial clinopyroxene

(YbN¼ 1�03)]. Erem-2 Cpx-I is similar to that reported

from the Manzaz volcanic district (Central Hoggar) by

Beccaluva et al. (2007). The trace element patterns of

metasomatic clinopyroxenes from melt pockets (Cpx-III)

show a continuous positive slope from HREE to LREE

[1�1< (Eu/Lu)N< 4�7; 0�7< (La/Sm)N<6�4] with well-

marked negative Zr–Hf and Ti anomalies but variable

Nb–Ta abundances compared with the neighbouring

REE [0�28< (Nb/La)N<10�5]. Opx–I from sample

Erem-2 displays a general negative slope and positive

Ti and Zr–Hf anomalies [(Zr/Sm)N< 2�4] (Fig. 8b).

Orthopyroxenes from the rest of the spinel peridotites

display a concave PM-normalized pattern and negative

Zr–Hf anomalies [0�12< (Zr/Sm)N< 0�75]. They show a

positive slope from MREE to LREE [2�1< (La/

Sm)N< 11�3]. The slope of the orthopyroxene trace

element patterns and the Zr–Hf anomalies are opposite

to those of orthopyroxene from the garnet peridotite.

Olivines display a large range of variation and enrich-

ment from Dy to Lu [0�08< (Dy/Lu)N< 0�45]. Amphibole

(Amph-II) is enriched in LREE relative to MREE and

HREE [3�3< (La/Sm)N< 11�9] and its PM-normalized REE

patterns mostly mimic those of Cpx-III. Amphiboles dis-

play negative Ti and Zr–Hf anomalies [(Zr/Sm)N¼ 0�10–

0�34], but positive Nb–Ta anomalies [2�1< (Nb/

La)N<6�5]. Damp/cpx(Zr) and Damp/cpx(Hf) are around

1�4 6 0�4 and 1�1 6 0�3, respectively, whereas Damp/

cpx(Nb) and Damp/cpx(Ta) are highly variable between 0�7and 40 and between 0�8 and 34, respectively. Previous

studies (e.g. Ionov et al., 1997) have reported a similar

variability and range of Damp/cpx for the HFSE. However,

LILE partitioning between Amph-II and Cpx-III is ex-

tremely variable and overall lower than estimated in

previous studies. For instance, despite high Ba contents

(53–519 ppm), Damp/cpx(Ba) varies between 0�8 and 456

and barely overlaps with the literature range of between

200 and 3000 (Adam et al., 1995; Brenan et al., 1995;

Table 4: Pb–Sr–Nd isotopic compositions of whole-rock garnet and spinel peridotites, pyroxenites, melilitites and phlogopitemegacrysts

206Pb/204Pb 207Pb/204Pb 208Pb/204Pb 87Sr/86Sr eSr(i) Classical leaching Stronger leaching

87Sr/86Sr eSr(i)143Nd/144Nd eNd(i)

Garnet peridotiteInt-14-7 18�5241 6 19 15�6101 6 19 38�3263 6 45 0�703303 6 4 –20 0�703049 6 5 –20 0�512946 6 5 6�2Int-14-7r 18�5496 6 6 15�6054 6 6 38�3364 6 15 0�703551 6 7 –13 0�512929 6 4 6�1Spinel peridotiteErem-2 18�5893 6 4 15�6070 6 4 38�3779 6 14 0�704299 6 6 –2�9 0�513091 6 6 9Il-17-4 19�7623 6 4 15�6782 6 4 39�4682 6 13 0�704419 6 8 –0�9 0�512958 6 7 6�5Il-17-5 19�6090 6 6 15�6727 6 6 39�3158 6 17 0�704472 6 19 –0�2 0�512924 6 5 5�7Il-17-13 19�8158 6 4 15�6712 6 3 39�4382 6 11 0�703694 6 10 –11�7 0�703359 6 4 –21 0�512928 6 3 5�9Il-17-14 19�8840 6 4 15�6751 6 3 39�5007 6 10 0�703753 6 5 –7 0�512916 6 2 5�6Il-17-14r 19�8150 6 3 15�6737 6 3 39�4453 6 9 0�703725 6 19 –10�7 0�512915 6 3 5�9Il-17-18 19�9803 6 5 15�6659 6 5 39�5092 6 17 0�703396 6 3 –15�5 0�512979 6 4 6�9Int-15-1 19�1522 6 4 15�6788 6 5 39�1135 6 10 0�704925 6 9 6 0�704063 6 3 –19 0�513100 6 10 9�1Int-17-1 18�6650 6 4 15�6091 6 4 38�4934 6 11 0�705307 6 3 11�6 0�703104 6 5 –19 0�512927 6 4 5�8Int-17-1r 18�6503 6 2 15�6080 6 2 38�4879 6 52 0�705452 6 12 14 0�512926 6 4 6�1Int-17-6 19�0202 6 26 15�6373 6 21 38�7406 6 56 0�704476 6 11 –0�4 0�703868 6 9 –20 0�512889 6 8 5�1Int-17-7 18�3595 6 4 15�6003 6 4 38�1725 6 13 0�703364 6 4 –16 0�703008 6 9 –21 0�512960 6 3 6�5Int-17-12 19�8433 6 5 15�6650 6 5 39�4140 6 13 0�703600 6 6 –13 0�703037 6 8 –20 0�512964 6 4 6�6Int-17-13 19�7701 6 6 15�6561 6 4 39�3444 6 12 0�705026 6 5 8 0�512986 6 3 7Int-14-6 18�7054 6 10 15�6138 6 9 38�4810 6 28 0�703836 6 18 –9�8 0�703093 6 6 –20 0�512912 6 4 5�6Ol-clinopyroxeniteIl-17-16 19�8029 6 5 15�6631 6 4 39�3954 6 16 0�704696 6 50 3�1 0�512936 6 4 6Il-17-16r 19�8137 6 5 15�6617 6 4 39�3286 6 13 0�704550 6 8 1 0�512940 6 4 6�4WehrliteIl-17-21 19�8922 6 5 15�6686 6 5 39�4490 6 14 0�703391 6 6 –15 0�512939 6 4 6�1PyroxeniteIl-17-22 19�6459 6 5 15�6416 6 4 39�2549 6 14 0�703267 6 4 –18 0�512955 6 2 6�3Il-17-24 19�7378 6 5 15�6550 6 4 39�3305 6 11 0�703448 6 6 –20�9 0�512936 6 3 5�9Il-17-27 19�4690 6 3 15�6662 6 3 39�2097 6 12 0�703495 6 12 –17�5 0�512917 6 4 5�5Int-17-100 19�5189 6 4 15�6695 6 4 39�2885 6 93 0�704991 6 6 4�3 0�705019 6 5 –21 0�512965 6 3 6�5Melilitite8814 19�6014 6 9 15�6417 6 8 39�2337 6 21 0�703164 6 6 –19�1 0�512950 6 4 6�38815 19�6408 6 6 15�6410 6 5 39�2496 6 13 0�703227 6 4 –18�6 0�512931 6 6 5�98816 19�6569 6 6 15�6374 6 6 39�2458 6 16 0�703148 6 3 –19�3 0�512953 6 4 6�4PhlogopiteInt-4 18�9286 6 3 15�5530 6 3 38�8550 6 9 0�704234 6 6 –40 0�512783 6 10 3Int-14 19�0874 6 5 15�6878 6 5 39�5024 6 14 0�704213 6 12 –40 0�512715 6 3 1�7Int-17 18�6836 6 6 15�6333 6 5 38�5971 6 16 0�703820 6 10 –46 0�512699 6 8 1�4

Errors on isotopic ratios are 2r. Italic lines represent the duplicates.


at University of C

ambridge on A


Dow

nloaded from


Ionov et al., 1997; Gregoire et al., 2000; Powell et al.,

2004). Similarly, Damp/cpx(Rb) is always lower than three

for the In Teria peridotites, significantly below natural

and experimental data (100–800; Adam et al., 1995;

Brenan et al., 1995; Ionov et al., 1997; Gregoire et al.,

2000; Powell et al., 2004). In contrast, we obtain a Damp/

cpx(Sr) of about 1�8 6 0�7, which compares well with pre-

vious data (e.g. Ionov et al., 1997; Gregoire et al., 2000;

Powell et al., 2004) and experimental determinations

(e.g. Adam et al., 1995; Brenan et al., 1995). Phlogopite

is characterized by low REE abundances (Lu< 0�7 ppb).

PM-normalized patterns show an overall positive slope

from HREE to LREE–Th [7< (La/Sm)N� 22], marked by

pronounced positive HFSE anomalies [2< (Zr/

Sm))N� 27; 280< (Nb/La)N�14684] and strong LILE en-

richment relative to REE [e.g. 157< (Sr/Nd)N� 607;

7583< (Ba/La)N� 74080]. However, positive Zr and Hf

anomalies are mainly due to low HREE–MREE contents,

as Zr and Hf concentrations in phlogopite remain low

compared with Cpx, and yield a Dphl/cpx(Zr,Hf)�0�05,

lower than reported in the literature (e.g. Adam et al.,

1995; La Tourette et al., 1995; Ionov et al., 1997;

Gregoire et al., 2000). This is not the case for Nb and Ta,

which significantly partition into phlogopite relative to

clinopyroxene [e.g. 1�3<Dphl/cpx(Nb)� 16�5]. These are

within the spread of values reported from previous nat-

ural and experimental studies (e.g. Adam et al., 1995; La

Tourette et al., 1995; Ionov et al., 1997; Gregoire et al.,

2000). Whereas Cs and Ba partition strongly into

phlogopite relative to Cpx, Sr and Rb prefer Cpx, and U,

Th and Pb partitioning is variable [e.g. 0�4�Dphl/

cpx(U)�3�6].

Wehrlite and Ol-clinopyroxenite. Clinopyroxene trace

element patterns show a positive slope from the HREE

to the LREE relative to the MREE [(La/Sm)N� 1�9 6 0�5;

(Eu/Lu)N�3�3 6 0�2], similar to the patterns observed

for Cpx-III in other peridotites. The HFSE systematics

are also comparable with those of peridotite Cpx-III,

with negative Zr–Hf and Ti anomalies relative to the

MREE [e.g. (Zr/Sm)N�0�07] and positive Nb–Ta anoma-

lies relative to the LREE (e.g. (Nb/La)N� 2�28). The in-

compatible trace element patterns of amphibole

(Fig. 8c) are almost indistinguishable from those of

clinopyroxene (Fig. 7b).

PyroxeniteClinopyroxenes in pyroxenites have a general convex-

downward PM-normalized trace element pattern with a

positive slope from the HREE to MREE and a negative

slope from the MREE to LREE, with positive Zr–Hf and

both negative and positive Ti anomalies [(Zr/

Sm)N¼ 1�35–3�12] (Fig. 7c). These anomalies are the op-

posite of those observed in clinopyroxenes from spinel

peridotite. The Nb–Ta positive anomaly is well marked

only in sample Il-17-104 and is absent or weak for the

three other pyroxenites. Amphibole REE contents and

patterns are similar to those of clinopyroxene (Fig. 8d).

Like clinopyroxene, amphibole exhibits positive Zr–Hf

[(Zr/Sm)N¼ 2�33–4�03] and Ti anomalies. The opposite

was observed for the peridotite amphibole and clino-

pyroxene. Amphibole showing both Rb–Ba enrichment

and depletion has been observed in the same pyroxen-

ite sample (Fig. 8d). Phlogopites have low REE contents

and show an overall positive slope from MREE to LREE.

Their extended trace elements patterns are similar to

those described for peridotite xenoliths (especially for

the garnet peridotite; Fig. 7d and f) and are marked by

prominent positive HFSE anomalies [e.g. 160� (Nb/

La)N<6300), 27< (Zr/Sm)N< 237] and high Cs, Rb, Ba

and Sr abundances (e.g. Sr and Ba up to�170 ppm

and�2287 ppm, respectively). In contrast to phlogopite

in spinel peridotites, that in the pyroxenites shows low

U and Th abundances, as in the garnet peridotites. A

phlogopite megacryst analyzed by solution ICP-MS

shows a similar pattern (Fig. 7f; Supplementary Data

Table A2), except for the HREE. We note that DPhl/

Amp(Nb,Ta) and DPhl/Amp(Zr,Hf) are in the range of 0�6–2

and 0�04–0�09, respectively, as reported by Gregoire

et al. (2000) and Moine et al. (2001) (and references

therein). Similarly, the partitioning of Rb (9–22), Ba

(7–20), and Sr (0�1–0�5) between these two phases is

also broadly similar to that reported in previous studies.

Sr, Nd AND Pb ISOTOPIC DATA

Whole-rock isotopic compositions of mantle xenoliths,

phlogopite megacrysts and host melilitites are reported

in Table 4. The 87Sr/86Sr of the In Teria xenoliths varies

widely from 0�70327 to 0�70503 and does not correlate

with 143Nd/144Nd (þ5�5< eNd<þ6�6; Fig. 9a). Two spinel

peridotites (samples Erem-2 and Int-15-1) have higher143Nd/144Nd values of 0�51309 and 0�51310, respectively

(Table 4). In Sr–Nd isotope space, 11 mantle xenoliths

samples (one garnet peridotite, three pyroxenites and

seven spinel peridotites) and the melilitites are closely

grouped and plot close to the HIMU mantle end-mem-

ber. This group partly overlaps the fields for Ahaggar al-

kali basalts (Allegre et al., 1981), Morocco peridotites

(Raffone et al., 2009; Wittig et al., 2010; Natali et al.,

2013) and is close to the field for Libyan alkali basalts

(Beccaluva et al., 2008). Whereas the existing data for

northern Africa mantle xenoliths define an array charac-

terized by variable 143Nd/144Nd at relative low87Sr/86Sr—extending up to high 143Nd/144Nd values re-

cording time-integrated mantle depletion—the In Teria

xenoliths show higher and more variable 87Sr/86Sr val-

ues at nearly constant 143Nd/144Nd (Fig. 9a). Another

group of samples (eight spinel peridotites and one pyr-

oxenite) from In Teria have higher 87Sr/86Sr values in

the range 0�7045–0�7055 (Fig. 9a). The field for Canary

Islands peridotites and pyroxenites shows the same

characteristics, with variable 87Sr/86Sr values at con-

stant 143Nd/144Nd values (Whitehouse & Neumann,

1995; Neumann et al., 2002, 2015). For all In Teria

samples, the 87Sr/86Sr values are not correlated with

whole-rock Sr contents, suggesting that the high


at University of C

ambridge on A


Dow

nloaded from




Canary

islands

(peridotites &

pyroxenites)

Canary

islands

(peridotites &

pyroxenites)

Canary

islands

(peridotites &

pyroxenites)

NH

RL

NHRL

207 P

b/20

4 Pb

208 P

b/20

4 Pb

143 N

d/14

4 Nd

HIMU

HIMU

HIMU

DMM

DMM

EM1

DMM

EM2

EM1

EM2

garnet peridotite

phlogopitemelilititespyroxenites

spinel peridotites

EM10.5124

0.5126

0.5128

0.5130

0.5132

0.5134

0.5136

0.5138

0.702 0.703 0.704 0.705 0.706

37.0

37.5

38.0

38.5

39.0

39.5

40.0

40.5

17 18 19 20 21 22

15.4

15.5

15.6

15.7

15.8

17 18 19 20 21 22

Ahaggar

(alkali basalts)

Ahaggar

(tholeiites)

Ahaggar

(Manzaz

peridotites)

Morocco

(peridotites)

206Pb/204Pb

87Sr/86Sr

Libya

(peridotites)

Libya

(alkali basalts)(a)

(b)

(c)

Ol-cpxlherz + harzb

Ahaggar

(tholeiites)

Morocco

(peridotites)

Morocco

(peridotites)

Ahaggar

(Manzaz

peridotites)

Ahaggar

(Manzaz

peridotites)

Libya

(alkali basalts)

Libya

(peridotites)

Libya

(peridotites)

Ahaggar

(tholeiites)

Ahaggar

(alkali basalts)

Ahaggar

(alkali basalts)

Erem-2 Int-15-1

Libya

(alkali basalts)

lherz Erem-2

Sicily

(peridotites &

pyroxenites)

gg

gg

Fig. 9. Variation of 143Nd/144Nd vs 87Sr/86Sr (a), 207Pb/204Pb vs 206Pb/204Pb (b) and 208Pb/204Pb vs 206Pb/204Pb (c) for in Teria melili-tites, pyroxenites, spinel peridotites, the garnet peridotite (Int-14-7) and phlogopite megacrysts. The following fields are shown forcomparison: Ahaggar alkali basalts (Allegre et al., 1981; Dupuy et al., 1993); Ahaggar tholeiitic basalts (Aıt-Hamou, 2000; Aıt-Hamou


at University of C

ambridge on A


Dow

nloaded from


87Sr/86Sr values are not a secondary weathering effect.

The three phlogopites analysed yield low 143Nd/144Nd

values (þ1�4< eNd<þ3) at relatively high 87Sr/86Sr

(0�7038–0�7042) and plot in the field of tholeiitic basalts

from central Ahaggar (Aıt-Hamou et al., 2000).

Compared with In Teria whole-rock xenoliths, some of

the separated clinopyroxenes from the xenoliths from

Libya and Ahaggar (Manzaz district) show significantly

higher 143Nd/144Nd values (Beccaluva et al., 2007, 2008).

The Pb isotopic compositions of the In Terra xeno-

liths are in the range of 18�36–19�98 for 206Pb/204Pb,

15�60–15�69 for 207Pb/204Pb, and 38�17–39�51 for208Pb/204Pb (Table 4). In 207Pb/204Pb–206Pb/204Pb and208Pb/204Pb–206Pb/204Pb space, the samples define a

relatively large range (Fig. 9b and c). Seven spinel peri-

dotites, all the pyroxenites and the melilitites are char-

acterized by high 206Pb/204Pb and 208Pb/204Pb ratios

trending towards the HIMU end-member. This group of

samples overlaps the fields defined by alkali basalts

from Ahaggar and Libya (Allegre et al., 1981; Dupuy

et al., 1993; Beccaluva et al., 2008) and the fields of

Manzaz and Libyan peridotites (Beccaluva et al., 2007,

2008). The garnet-bearing peridotite and four spinel

peridotites show significantly lower 206Pb/204Pb and208Pb/204Pb values (<18�76 and <38�50, respectively)

associated with high 207Pb/204Pb ratios (>15�60) (Fig. 9b

and c) and partly overlap the field of Libyan peridotites

(Beccaluva et al., 2008). The three phlogopite mega-

crysts display intermediate 206Pb/204Pb values (18�68–

19�09), a large variation in 208Pb/204Pb and 207Pb/204Pb,

and plot between the two xenolith groups identified

previously (Fig. 9b and c).

DISCUSSION

Mantle deformationThe association of porphyroclastic microstructures,

marked by elongated olivine crystals with common sub-

grains, and the well-developed olivine and orthopyroxene

CPO indicate that the studied xenoliths are deformed. In

the garnet peridotite, CPO are characterized by the paral-

lelism of clinopyroxene and orthopyroxene [001] maxima

and the olivine [100] maxima (Fig. 3a). This suggests that

all the minerals accommodated the same deformation,

consistent with petrographic observations indicating that

the clinopyroxene is primary (Fig. 2a). In all spinel perido-

tites, the parallelism of olivine [100] and orthopyroxene

[001] maxima implies that the primary phases (Ol-I and

Opx-I) have recorded the same deformation event. In con-

trast to garnet peridotite, in most spinel lherzolites, the

lack of correlation between clinopyroxene and amphibole

CPO, together with petrographic observations, suggests

that clinopyroxene and amphibole are secondary and that

this metasomatic addition post-dated the main deform-

ation episode. An orthopyroxene crystallographic control

on the growth of clinopyroxene and amphibole might ex-

plain the partial correlation between clinopyroxene and

amphibole CPO and the olivine and orthopyroxene CPO in

the Type A spinel peridotites (Fig. 3b). An implication of

these observations is that prior to the metasomatic event,

the shallow lithospheric mantle beneath In Teria had a re-

fractory harzburgitic composition, whereas deeper levels

had more fertile compositions.

Analysis of the olivine CPO patterns also singles out

the garnet peridotite. The latter has a fiber-[010] olivine

(Fig. 3a), which implies simultaneous activation of

the [100](010) and [001](010) slip systems or a

Clinopyroxene

Orthopyroxene

pyroxenites

spinel peridotitesgarnet peridotite

Cpx-III

Cpx-I

(Ti/Eu)N

(b)

(a)

Erem 2

Cpx-I

0

2

4

6

8

10

12

14

16

18

20

0 0.5 1 1.5 2 2.5 3 3.5

0.001

0.01

0.1

1

10

0 10 20 30 40

(Ce/

Yb) N

100

(Ce/

Yb) N

Fig. 10. Chondrite-normalized (Ce/Yb)N vs. (Ti/Eu)N for (a)clinopyroxene and (b) orthopyroxene (Opx-I) from In Teria gar-net peridotite, spinel peridotites and pyroxenites. These dia-grams are based on those proposed by Coltorti et al. (1999).Chondrite normalizing values are from McDonough & Sun(1995). Cpx-I, primary clinopyroxene; Cpx-III, clinopyroxenefrom the third metasomatic assemblage in the spinel perido-tites (stage 3).

et al., 2000); alkali basalts from Libya (Beccaluva et al., 2008),peridotite xenoliths from Manzaz, Ahaggar (Beccaluva et al.,2007); peridotite xenoliths from the Middle Atlas, Morocco(Raffone et al., 2009; Wittig et al., 2010; Natali et al., 2013); peri-dotite xenoliths from Libya (Beccaluva et al., 2008); peridotitexenoliths from Sicily (Bianchini et al., 2010); mantle xenolithsfrom Tenerife, Hierro and Fuerteventura, Canary Islands(Whitehouse & Neumann, 1995; Neumann et al., 2002, 2015).HIMU, DMM, EM2 and EM1 mantle end-member compositionsare from Zindler & Hart (1986) and Eisele et al. (2002). TheNorthern Hemisphere Reference Line (NHRL) is from Hart(1984).


at University of C

ambridge on A


Dow

nloaded from


transpressional deformation (Tommasi et al., 1999,

2000). However, the strong [001] maximum of the pyr-

oxene CPO is not consistent with transpressional de-

formation. We suggest therefore that this olivine CPO

records deformation at high stress or under high-pres-

sure conditions (e.g. Jung & Karato, 2001; Mainprice

et al., 2005; Demouchy et al., 2013). The spinel perido-

tites, on the other hand, show olivine CPO with ortho-

rhombic or fiber-[100] symmetries, consistent with

dominant activation of the (010)[100] slip system, sug-

gesting deformation at low-stress, high-temperature,

low-pressure conditions (Tommasi et al., 2000).

Metasomatic imprintsBased on petrographic evidence and geochemical data,

three metasomatic imprints are recognized in the In

Teria xenoliths, as follows.

1. The garnet and spinel peridotites were first affected

by crystallization of amphibole and/or phlogopite at

the expense of orthopyroxene, a feature that was

ascribed by Dautria et al. (1992) to reactive porous

percolation of a hydrous alkaline melt. Although

it shows a lesser extent of modal metasomatism

(crystallization of phlogopite; Fig. 2b), the garnet

peridotite (sample Int-14-7) is characterized by a con-

vex-upward normalized REE pattern in clinopyroxene

(Fig. 7a), which suggests equilibration with a LREE-

enriched melt (Irving, 1980; Irving & Frey, 1984;

Bodinier et al., 1987). Moderate LREE enrichment

without significant negative HFSE anomalies (or with

positive anomalies) in clinopyroxene (or whole-rocks)

is widely considered as evidence for mantle meta-

somatism involving alkaline silicate melts (e.g.

Downes, 2001; Bodinier et al., 2004; Powell et al.,

2004; Kaeser et al., 2006). Together the observations

on the garnet and spinel peridotites suggest that the

lithospheric mantle beneath In Teria was pervasively

metasomatized by hydrous alkaline melts.

2. The second stage (stage 2) of metasomatism is akin

to the percolation of carbonatitic melt and has been

well documented by Dautria et al. (1992). It is re-

stricted to the spinel peridotites, where it is respon-

sible for the crystallization of Cr-rich diopside at the

expense of orthopyroxene and amphibole. It results

in the formation of microgranular aggregates (for-

mer melt pockets) composed of secondary clino-

pyroxene, olivine, Al- and Cr-spinel, sulphides,

K-feldspar and glass, which extensively replace theprimary phase assemblages (Fig. 2h and i). Most of

the major and trace element characteristics of these

samples and their minerals are typical of mantle

metasomatism by carbonate melt (Woolley &

Kempe, 1989; Yaxley et al., 1991; Ionov et al., 1993a;

Hauri & Hart, 1994; Blundy & Dalton, 2000;

Chakhmouradian, 2006). These characteristics not-

ably include a strong enrichment of highly incompat-

ible trace elements in whole-rocks and pyroxenes

(Opx-I and Cpx-III), negative HFSE (Ta, Zr, Hf and Ti)

anomalies in Cpx-III, and negative Zr and Hf anoma-

lies in Opx-I (Figs 5 and 7). Ol-III have high Mg# and

are enriched in CaO compared with Ol-I, and Cpx-III

are enriched in Cr (Fig. 6). Moreover, secondary

clinopyroxene (Cpx-III) and coexisting orthopyrox-

ene (Opx-I) show evolutionary trends in Fig. 10

marked by a strong increase in (Ce/Nb)N at rela-

tively low (Ti/Eu)N, confirming interaction with a car-

bonatitic melt. The possibility that the melt pockets

were formed by interaction with the host lava during

the exhumation of the peridotite xenoliths is

excluded because only spinel peridotites show this

interaction (i.e. it is not observed in the garnet peri-

dotite and pyroxenites).

3. The metasomatic history of the In Teria xenolithsuite concludes with the crystallization of pyroxen-

ites (stage 3). These rocks, which are predominant

among the xenolith suite, crystallized in the spinel

stability field, lack the carbonate-melt imprint (stage

2) observed in all spinel peridotites and resemble

the amphibole-pyroxenite veins found in Pyrenean

peridotites (Bodinier et al., 1987, 2004) with respect

to their major and trace element characteristics

(Figs 6 and 11). Trace element similarities between

the In Teria and Pyrenean pyroxenites include

0

5

10

15

20

25

30 35 40 45 50 55 60

pyroxenite xenoliths (this study)literature data for mantle pyroxenites

to carbonatite

SiO2 (wt%)

CaO

(wt%

)

melilitites (this study)

(3) (2)

(1)

amphibole pyroxenites veins from Lherz

Fig. 11. Variation of CaO vs SiO2 for the In Teria pyroxenitexenoliths and host melilitites, compared with literature data forpyroxenites and basaltic to carbonatitic mantle melts (shadedfields). The pyroxenite dataset includes samples from the fol-lowing xenolith suites: Hawai (Frey, 1980); Australia (Griffin &O’Reilly, 1986; O’Reilly & Griffin, 1987; O’Reilly et al., 1988);Sidamo, Ethiopia (Bedini, 1994); orogenic lherzolite massifs(Lherz, France, Bodinier et al., 1987, 1990; McPhail et al., 1990;Ronda, Spain, Garrido & Bodinier, 1999); ophiolites (Thailand,Orberger et al., 1995; New Caledonia, J. L. Bodinier, unpub-lished data); the intrusive pyroxenite–carbonatite body ofTamazeght, Morocco (Marks et al., 2008). The overlap betweensome of the In Teria pyroxenites and the amphibole-bearingpyroxenite veins from Lherz (Bodinier et al., 1987; Bodinier,1989) should be noted. Shaded lava fields: (1) South African al-kaline basalts and olivine melilitites (Janney et al., 2002); (2)tholeiites, transitional basalts, basanites and nephelinites fromSicily, Italy (Beccaluva et al., 1998); (3) undersaturated alkalineto carbonatitic melts from Morocco (Wagner et al., 2003).


at University of C

ambridge on A


Dow

nloaded from


convex-upward normalized REE patterns and posi-

tive HFSE anomalies in whole-rocks and amphibole

(Nb, Ta, Zr, Hf, Ti) and, to a lesser degree, in clino-

pyroxene (Zr, Hf, 6 Ta). In Fig. 10a, the pyroxenites

show a trend defined by a strong increase in the (Ti/

Eu)N ratio of clinopyroxene correlated with a moder-

ate increase in (Ce/Yb)N. This evolutionary trend di-

verges markedly from that defined by the secondary

clinopyroxene in the spinel peridotites. The In Teria

pyroxenites probably represent crystal segregates

from hydrous alkaline melts crystallized in vein

conduits—rather than in large intrusions.

Metasomatism: several events or multi-stagemelt–rock interactionMicrostructural observations indicate that all metasom-

atic events postdate the last deformation event, re-

corded by olivine in both garnet and spinel peridotites,

that affected the lithospheric mantle in this region. The

alignment of the phlogopite–amphibole aggregates in

the foliation, but the undeformed character of the crys-

tals and the dispersion of crystallographic axes of

Amph-II and Cpx-III (e.g. Type B in Fig. 3d) indicate that

although melt infiltration postdated the deformation, it

was at least partially controlled by the pre-existing

structure of the peridotites. A stronger structural control

of the pre-existing microstructure on the reaction prod-

ucts is also indicated by the similarity of Amph-II and

Cpx-III CPO in Type A spinel lherzolites (Fig. 3b), which

suggests topotaxial growth of at least part of the meta-

somatic products.

Our geochemical data show striking similarities be-

tween the mineral compositions of peridotites affected

solely by stage 1 metasomatism and those of stage 3

pyroxenites. This may be exemplified by comparing the

trace element signatures of clinopyroxene and phlogo-

pite in the garnet peridotite sample and in the pyroxen-

ites (Figs 7 and 10a). Moreover, our data show the clear

geochemical affinity of the pyroxenite xenoliths (stage 3)

with the melilitite host lavas. As illustrated in Fig. 11, the

In Teria pyroxenites differ from other mantle pyroxenites

(including, to some degree, the Pyrenean veins) in their

lower SiO2 contents and as such are compositionally

akin to strongly undersaturated and carbonatitic mag-

mas. The equilibrium melt calculated from the trace

element composition of clinopyroxene in pyroxenite is

comparable with the host melilitites—as well as with the

garnet peridotite equilibrium melt [Fig. 12; partition coef-

ficients (KD) from Hart & Dunn (1993)]. The only signifi-

cant differences are the positive HFSE anomalies

(particularly for Zr and Hf) that are observed in the pyrox-

enite equilibrium melt (but not in the garnet peridotite

equilibrium melt). These anomalies probably reflect the

inadequacy of available experimental clinopyroxene/

melt KD values for clinopyroxenes crystallized from

strongly SiO2-undersaturated alkaline melts.

The microstructural observations and the geochem-

ical affinity of the xenoliths with the melilitite host lavas

favour a scenario whereby the different metasomatic im-

prints record successive stages of interaction between

lithospheric mantle and sublithospheric melts. The evo-

lution from stage 1 to stage 2 (silicate/hydrous to carbon-

ate melt) may result from a differentiation process

involving melt ingress down a lithospheric thermal gradi-

ent associated with melt–rock reactions at decreasing

melt mass (Bedini et al., 1997; Bodinier et al., 2004).

mel

t / P

rimiti

ve M

antle

Ba Th U Nb Ta La Ce Pr Sr Nd Zr Hf SmEu Ti Gd Tb Dy Ho Y Er TmYb Lu

melt: Cpx-III sp peridotites

melt: Cpx pyroxenite (Il-17-27)

melt: Cpx-Isp peridotite(Erem-2)

melilitite (8814)

melt: Cpx-I grt peridotite (Int-14-7)

0.1

1

10

100

1000

10000

Fig. 12. Calculated compositions of theoretical melts in equilibrium with Cpx-I (Erem-2) and Cpx-III in spinel peridotites, the garnetperidotite and a pyroxenite sample. We used the partition coefficients between Cpx and basaltic melt of Hart & Dunn (1993) for thespinel peridotite, the garnet peridotite and the pyroxenite, and the partition coefficients between Cpx and carbonatitic melt ofKlemme et al. (1995) for secondary Cpx (Cpx-III) in the other spinel peridotites. A host melilitite (sample 8814) is shown for compari-son. sp, spinel; grt, garnet.


at University of C

ambridge on A


Dow

nloaded from


Gradual solidification of the melt, also referred to as ‘per-

colative fractional crystallization’ (Harte et al., 1993),

would be responsible for marked changes in its compos-

ition, culminating in the individualization of volatile-rich

and/or carbonated small melt fractions that are expected

to be strongly enriched in highly incompatible elements

(McKenzie, 1989). Equilibrium melts calculated from the

trace element composition of In Teria Cpx-III are enriched

in incompatible elements and display negative HFSE

anomalies in their PM-normalized patterns [Fig. 12; KD

from Klemme et al. (1995)]. This signature is similar to

that obtained by numerical simulations of reactive por-

ous flow at decreasing melt mass (Bedini et al., 1997;

Ionov et al., 2002a).

Trace element modelling predicts that the combin-

ation of chromatographic effects related to melt trans-

port and ‘source’ effects related to melt–rock reactions

results in spatial decoupling between melt–rock reac-

tions and their trace element expression (Godard et al.,

1995). This may be responsible for transient depletion

of several incompatible elements in porous-flow sys-

tems, owing to buffering of the percolating melt by the

depleted peridotite protolith (see, e.g. fig. 13 of Bodinier

et al., 2008). In contrast to other pyroxenes from the In

Teria xenolith suite, Cpx-I and Opx-I of the spinel peri-

dotite Erem-2 are depleted in LREE (Figs 7, 8, 10 and

12). The unique trace element signature of this sample

may represent a remnant of the pre-metasomatic,

depleted composition of the lithospheric mantle be-

neath In Teria. Alternatively, it may also be explained

by transient buffering of the percolating melt by the

peridotite protolith.

Melt compositions appear to have varied with perco-

lative differentiation from siliceous and hydrous, but

locally impoverished in highly incompatible trace elem-

ents relative to the infiltrated melt (e.g. sample Erem-2

equilibrium melt with cpx-I; Fig. 12), to carbonated and

strongly enriched in highly incompatible trace elem-

ents, but with negative HFSE anomalies (other spinel

peridotites equilibrium melt in Fig. 12). Clinopyroxene

trace element compositions in the garnet peridotite and

pyroxenites (Fig. 7) and the similarity between the equi-

librium melt for the garnet peridotite and the host meli-

litite (Fig. 12) indicate that the garnet peridotite has

interacted with a slightly evolved silicate melt. This may

have a deeper origin than the spinel peridotites and/or a

100 mWm-2

100 mWm-2

60 mWm-2 80 mWm-2 80 mWm-260 mWm-2

Pre

ssur

e G

Pa

Temperature ºC

garnet

4.0

spinel

3.0

2.0

1.0

0.0

700 800 900 1000 1100 1200 1300 1400 1500

700 800 900 1000 1100 1200 1300 1400 1500

plagioclase

spinel

parg out

solidus

hydrous

peridotite

solidus

anhydrous

peridotite

spinel peridotites (T from mineral cores)

grt peridotite (Int-14-7) mineral cores

grt peridotite (Int-14-7) mineral rims

spinel peridotites (T from mineral rims)

Fig. 13. Pressure and temperature estimates for the In Teria mantle xenoliths. For the spinel peridotites the temperatures were ob-tained from mineral cores and rims using the Na-in-Opx/Cpx geothermometer of Brey & Kohler (1990). The pressure conditions forthe xenoliths were constrained by the absence of plagioclase and garnet in the peridotites. For the garnet peridotite, both pressureand temperature estimates were obtained from mineral cores and rims, using the Nickel & Green (1985) Al in Opx/Grt geobarome-ter and the Brey & Kohler (1990) Na-in-Opx/Cpx geothermometer. The plagioclase–spinel and spinel–garnet phase transitions arefrom Gasparik (1987) and Gasparik (1984), respectively. The pargasite stability field (parg. out) is from Niida & Green (1999), andthe anhydrous and hydrous peridotite solidus is from Taylor & Green (1983). The geotherms for the Ahaggar shield and the In-Salah–Illizi zone are from Lesquer et al. (1990). They were calculated for surface heat flows in the range 80–100 mW m–2, mean crus-tal heat productions of 0�8mW m–3 (continuous lines) and 1�2 mW m–3 (dashed lines), a 30–35 km thick crust and no contributionfrom the mantle.


at University of C

ambridge on A


Dow

nloaded from


possible origin in the wall-rock of a magma conduit.

The garnet peridotite records higher pressure–

temperature (P–T) conditions than the spinel peridotites

and is devoid of any carbonatitic metasomatic over-

print, suggesting a deeper origin.The multi-stage metasomatic event that affected the

In Teria xenoliths was probably coupled with a transient

thermal event; that is, heating of subcontinental litho-

sphere followed by thermal relaxation. Heating is re-

corded by the compositional zoning of the primary

minerals, indicating a core to rim increase of equilib-

rium temperatures from 1050–1100� to 1210–1240�C for

the garnet peridotite sample Int-14-7 and from 770–

870�C to about 1080�C for the spinel peridotite sample

Erem-2 (Fig. 13). Combined with pressure estimates,

these rim temperatures define a rather steep thermal

gradient roughly consistent with the high heat flux

measured in the In Salah–Illizi district (>100 mW m–2;

Lesquer et al., 1990). The rim P–T estimates lie in the

interval between the hydrous and anhydrous peridotite

solidus (Fig. 13) and may therefore record lithospheric

P–T conditions during the first metasomatic stage. The

intergranular porous flow of hydrous alkaline melt

inferred for this stage requires ambient temperatures in

excess of (at least) the hydrous peridotite solidus. The

observation that phlogopite is the only metasomatic

mineral formed in the garnet peridotite is consistent

with the high ‘rim’ temperature of this sample, well be-

yond the stability field of pargasite (Niida & Green,

1999; Fig. 13). Compared with amphibole, phlogopite is

stable at higher temperatures (Wendlandt & Eggler,

1980; Mengel & Green, 1989). Conversely, the predom-

inance of pargasite over phlogopite in spinel peridotites

would reflect the lower temperatures of the shallower

lithosphere during stage 1, overlapping the amphibole

stability field (Fig. 13).

In a scenario in which carbonate melt is produced by

reactive differentiation of percolating silicate melt down

a thermal gradient, the silicate and carbonate melt

metasomatism are coeval but disconnected in space

(Bedini et al., 1997; Bodinier et al., 2004). During con-

ductive heating of the lithosphere, of an asthenospheric

origin, silicate melt metasomatism would occur in

deeper and/or hotter lithosphere whereas carbonate

melt metasomatism would affect shallower and/or

colder domains (Fig. 14a). Such an arrangement of

metasomatic aureoles related to a thermal gradient has

been observed on a small scale in vein wall-rocks within

the Lherz peridotite (Bodinier et al., 2004). In the In Teria

suite, the lack of a carbonate melt imprint in the deeper

garnet peridotite, whereas it is pervasive in the shal-

lower spinel peridotites, provides evidence for a similar,

thermally controlled arrangement of metasomatic do-

mains at lithospheric scale.

In the proposed scenario, overprinting of stage 1

metasomatism in spinel peridotites by stage 2 meta-

somatism is explained by downwards subsidence of

the metasomatic domain owing to lithospheric thermal

relaxation (Fig. 14b). The lack of any carbonate-melt im-

print in the In Teria pyroxenites, in contrast to the spinel

garnetspinel60

30

35

40

45

carbonate melt metasomatism

50

55

(a) (b) (c)

(km)

silicate melt metasomatism (reactive percolation of hydrous alkaline melt at decreasing melt mass)

continental

crust

silicate melt metasomatism

carbonate melt metasomatism

(reactive percolation of evolved carbonate

melt fractions)

pyroxenites

~1100˚C

~1100˚C

the

rma

l re

laxa

tio

n

~1100˚

Fig. 14. Schematic illustration of the three stages of metasomatism associated with the lithospheric thermal evolution proposed forthe In Teria mantle xenoliths: (a) upon conductive heating of the lithosphere by an asthenospheric source, silicate melt metasoma-tism occurs in the deeper and/or hotter parts of the lithosphere whereas carbonate melt metasomatism is restricted to shallowerand/or colder domains; (b) during subsequent thermal relaxation and downwards subsidence of the metasomatic domains of stage1, silicate melt metasomatism in the spinel peridotites is overprinted by carbonate melt metasomatism (stage 2); (c) upon furtherlithospheric cooling extensive hydraulic fracturing allows the injection of silicate melts forming a dense network of pyroxenite veins(stage 3)


at University of C

ambridge on A


Dow

nloaded from


peridotites, which are extensively metasomatized, sug-

gests that the pyroxenite veins were late stage. The em-

placement of a dense vein network—judging from the

predominance of pyroxenites within the xenolith

suite—might be associated with further lithospheric

cooling favouring extensive hydraulic fracturing

(Fig. 14c). Based on their similarity to phlogopite within

the pyroxenites, the phlogopite megacrysts may repre-

sent deep crystal segregates from the infiltrating alka-

line silicate melt (Fig. 7).

Based on the time constraints necessary to allow

successive heating and thermal relaxation of the litho-

spheric mantle, the metasomatic evolution proposed in

our model probably lasted for the whole Cenozoic era.

Stage 1 heating was possibly related to Eocene tholei-

itic magmatism, for which a lithospheric contribution

has been suggested (Maza et al., 1998), implying sub-

stantial heating. The resemblance of In Teria phlogopite

megacrysts to Ahaggar tholeiitic volcanism in terms of

their Nd–Sr isotope composition (Fig. 9) lends support

to this hypothesis and suggests that the phlogopites are

related to stage 1. The suggested subsequent evolution

would have lasted until the recent melilitite volcanism,

based on the observation that stage 3 pyroxenites prob-

ably represent mantle segregates from melilitite mag-

mas. The relationship between the suggested

lithospheric thermal evolution and the higher heat flow

observed in the In Teria region, compared with the rest

of the Ahaggar Massif (Lesquer et al., 1989), is poorly

understood. The presence of garnet peridotites in a re-

gion of higher heat flow is paradoxical. The origin of the

high heat flux is probably related to the situation of In

Teria at the southern border the Sahara Basins (Fig. 1),

characterized by a regional-scale heat flow anomaly

(Takherist & Lesquer, 1989; Lesquer et al., 1990), rather

than to the mantle upwelling beneath Ahaggar.

Origin of 87Sr enrichmentAmong the In Teria xenoliths, the enrichment in 87Sr with-

out any concomitant decrease of 143Nd/144Nd is restricted

to the spinel peridotites and one pyroxenite (sample Int-

17-100). This enrichment is not observed in the host meli-

litites, which show a narrow range of isotopic compos-

itions comparable with those of the Ahaggar alkaline

basalts (Fig. 9). Together with the lack of correlation be-

tween 87Sr/86Sr and Sr concentration, this does not not

support a hypothesis of late-stage introduction of radio-

genic Sr into the studied samples. The enrichment in 87Sr

is not related to the degree of metasomatism either, nor

to the metasomatic enrichment in incompatible trace

elements. Although extensively re-equilibrated with alka-

line-silicate and carbonate melts, respectively, the garnet

peridotite and the Ol-clinopyroxenite (sample Il-17-21) are

not significantly enriched in radiogenic Sr

(87Sr/86Sr< 0�7035; Fig. 9a). Conversely, the spinel perido-

tite Erem-2, which is the least metasomatized sample

with LREE-depleted pyroxenes (Fig. 10), and the phlogo-

pite megacrysts are relatively enriched in radiogenic Sr

0.5124

0.5125

0.5126

0.5127

0.5128

0.5129

0.5130

0.5131

0.5132

0.7030 0.7035 0.7040 0.7045 0.7050 0.7055

(b+d)

EMI

143 N

d/14

4 Nd

HIMU

(a)

(c)

(f)(g)

(h)

(e)

mixing curves

87Sr/86Sr

Ahaggar

(tholeiites)

Ahaggar

(alkali basalts)

Erem-2 Int-15-1

Fig. 15. Variation of 87Sr/86Sr vs 143Nd/144Nd for the In Teria melilitites, pyroxenite and peridotite xenoliths and phlogopite mega-crysts, compared with results of one-dimensional percolation–diffusion modelling aiming to explore the effects of different com-positional parameter values retrieved from the studied xenoliths and their host lavas (see text). The continuous lines a–h illustratemodel runs performed with compositional parameter values reflecting the range of Sr–Nd contents observed in the studied perido-tites for the protolith, and in the host melilitites and clinopyroxene equilibrium melts for the infiltrated melt. The complete set ofparameter values is given in Supplementary Data Table A5. The dashed lines show the range of mixing curves calculated with thesame compositional parameters as the one-dimensional percolation–diffusion modelling. The symbols for the In Teria xenolithsamples, and the references for the Ahaggar basalt fields and the mantle EM1 and HIMU end-members, are the same as in Fig. 9.


at University of C

ambridge on A


Dow

nloaded from



(87Sr/86Sr> 0�704 and > 0�7038, respectively). Elevated87Sr/86Sr values found in a few limestone-hosted, Ca-rich

carbonatites from Tamazert (Morocco) are clearly related

to crustal contamination (Bouabdellah et al., 2010). To

test the origin of high 87Sr/86Sr values in the In Teria xeno-

liths we performed a stronger leaching on seven spinel

peridotites, the garnet peridotite and one pyroxenite

(Table 4). The new 87Sr/86Sr compositions for most sam-

ples are less radiogenic and closer to HIMU than previ-

ously (Supplementary Data Fig. A4). These results

suggest that the high 87Sr/86Sr values could be related to

the crystallization of interstitial micro-phases from small-

volume melts during the waning stages of metasoma-

tism. Detailed studies of trace element distribution in

metasomatized mantle xenoliths have shown that a sig-

nificant proportion of the whole-rock budget of large ion

lithophile elements (LILE, including Sr) is hosted by such

intergranular components (Bedini & Bodinier, 1999).

Recently, Kourim et al. (2014) have revealed the existence

of such a LILE-enriched interstitial component in mantle

xenoliths from southwestern Ahaggar. Chromatographic

theory predicts that downstream in a reactive percolation

column (i.e. at very low melt/rock ratio), the isotopic sig-

nature of the percolating melt may be decoupled from its

trace element signature: the melt may be isotopically

equilibrated with the peridotite protolith although

strongly enriched in highly incompatible elements as a

result of reactions at decreasing melt mass (Bodinier

et al., 2004). Owing to their ability to percolate as small-

volume melts through relatively cold peridotite

(McKenzie, 1989) and precipitate LILE-enriched micro-

phases, carbonate melts such as those invoked for the In

Teria stage 2 metasomatism are probably the best candi-

dates to generate such enrichments. This may explain

why 87Sr enrichment is observed in xenolith suites that

have experienced carbonate-melt metasomatism (e.g. In

Teria and the Canary Islands; Neumann et al., 2002,

2015).

0.5124

0.5125

0.5126

0.5127

0.5128

0.5129

0.5130

0.5131

0.5132

0.7030 0.7035 0.7040 0.7045 0.7050 0.7055

0.5124

0.5125

0.5126

0.5127

0.5128

0.5129

0.5130

0.5131

0.5132

0.7030 0.7035 0.7040 0.7045 0.7050 0.7055

0.5124

0.5125

0.5126

0.5127

0.5128

0.5129

0.5130

0.5131

0.5132

0.7030 0.7035 0.7040 0.7045 0.7050 0.70550.5124

0.5125

0.5126

0.5127

0.5128

0.5129

0.5130

0.5131

0.5132

0.7030 0.7035 0.7040 0.7045 0.7050 0.7055

10

EMIEMI

EMI EMI

143 N

d/14

4 Nd

87Sr/86Sr 87Sr/86Sr

HIMU HIMU

HIMUHIMU 14

PM 18

20

12

10

2018

1412

17

Melilitite(15.22)

Sr/Nd ratio in meltSr/Nd ratio in protolith

143 N

d/14

4 Nd

Sr-Nd concentration in protolith

PM (19.9-1.25)

(d)(c)

(b)(a)

PMx0.5PMx0.33

PMx2PMx3

Melx0.33Melx0.5

Melx3Melx2

Melilitite(1202-79)

Sr-Nd concentration in melt

(15.92)

PM

PMPM(1(15

U

Melilitite(1 22)(15.22)

Mixing curve U

MU

Ahaggar

(alkali basalts)

Ahaggar

(tholeiites)

Ahaggar

(alkali basalts)

Ahaggar

(tholeiites)

Ahaggar

(alkali basalts)

Ahaggar

(tholeiites)

Ahaggar

(alkali basalts)

Ahaggar

(tholeiites)

Erem-2 Int-15-1

Fig. 16. Variation of 87Sr/86Sr vs 143Nd/144Nd for the studied in Teria melilitites, pyroxenites and peridotite xenoliths, and phlogopitemegacrysts, compared with results of one-dimensional percolation–diffusion modelling showing the effect of compositional vari-ations in Sr and Nd (see text). The short-dashed line represents a reference experiment calculated with Primitive Mantle (PM) Sr–Nd contents as the protolith (McDonough & Sun, 1995) and the average Sr–Nd contents of the studied melilitites for the infiltratedmelt. The long-dashed line represents a mixing curve calculated with the same Nd–Sr contents. (a–d) illustrate the effects of varyingthe following: (a) the Sr–Nd concentrations in the peridotite protolith, from 0�33� to 3�PM; (b) the Sr–Nd concentrations in themelt, from 0�33�to 3� the average Sr–Nd contents in the melilitites; (c) the Sr/Nd ratios in the peridotite protolith and (d) in themelt, from 10 to 20. The other parameter values are the same as in Fig. 15 and are given in Supplementary Data Table A5. The sym-bols for the In Teria samples and the data sources for the Ahaggar basalt fields and mantle EM1 and HIMU end-members are thesame as in Fig. 9.


at University of C

ambridge on A


Dow

nloaded from




The high 87Sr/86Sr ratios recorded by the In Teria

xenoliths are most probably inherited from the pre-

metasomatic composition of the lithospheric mantle.

The highest values are comparable with the 87Sr/86Sr

ratios ascribed to the EM1 mantle end-member, a com-

position that is considered to represent the signature of

the lower continental lithosphere (Hawkesworth et al.,

1986; Mahoney et al., 1991; Milner & Le Roex, 1996).

(a)(b)

(c)

(d)

EMI

87Sr/86Sr

HIMU(b)

(a)

207 P

b/20

4 Pb

208 P

b/20

4 Pb

HIMU

DMM

DMM

EMI

Pb/Sr varies in protolith and melt

Ahaggar

(alkali

basalts)

Ahaggar

(tholeiites)

Libya

(peridotites)

Ahaggar

(Manzaz

peridotites) Libya

(alkali basalts)

Ahaggar

(Manzaz

peridotites)

Ahaggar

(alkali basalts) Ahaggar

(tholeiites)

Libya

(alkali basalts)

Libya

(peridotites)

37.5

38.0

38.5

39.0

39.5

40.0

0.702 0.703 0.704 0.705 0.706

15.45

15.50

15.55

15.60

15.65

15.70

15.75

15.80

15.85

0.702 0.703 0.704 0.705 0.706

(a)(b)

(c)

(d)

(e)(f)

(g)

Erem-2

Erem-2

Int-14-7

Fig. 17. Variation of 207Pb/204Pb and 208Pb/204Pb vs 87Sr/86Sr for the studied In Teria melilitites, pyroxenite and peridotite xenoliths,and phlogopite megacrysts, compared with results of the one-dimensional percolation–diffusion modelling showing the effect ofvarying the (Pb/Sr)melt/(Pb/Sr)peridotite ratios (see text). The model runs a–g in (a) were calculated with (Pb/Sr)melt/(Pb/Sr)peridotite val-ues in the range 1�4–0�6; the model runs a–d in (b) were obtained with a more restricted range of (Pb/Sr)melt/(Pb/Sr)peridotite values,from 1�1 to 0�7. The other parameter values are given in Supplementary Data Table A5. The symbols for the In Teria samples, andthe references for the fields of North African mantle xenoliths and Ahaggar Cenozoic volcanism, as well as for the mantle DMM,EM1 and HIMU end-members, are the same as in Fig. 9.


at University of C

ambridge on A


Dow

nloaded from



The Ahaggar intra-plate tholeiites contain an EM1-like

mantle component, generally ascribed to the involve-

ment of lower lithosphere during their genesis (Maza

et al., 1998; Aıt-Hamou et al., 2000). Conversely, the

HIMU signature found in North African alkali basalts,

and in In Teria melilitites, pyroxenites and several spi-

nel peridotites (Fig. 9), is widely considered to represent

the signature of upwelling asthenosphere or a mantle

plume (e.g. Maza et al., 1998; Aıt-Hamou et al., 2000;

Beccaluva et al., 2008). In this respect, it is worth noting

that the three analysed phlogopite megacrysts plot be-

tween HIMU–DMM and EM1 compositions, within the

field of the Ahaggar tholeiites, hence providing further

evidence for the existence of an EM1 mantle signature

beneath In Teria (Fig. 9).

However, although enriched in radiogenic Sr, the

isotopic composition of the In Teria spinel peridotites

differs markedly from the end-member EM1 signature.

Their 143Nd/144Nd ratios are too high and their208Pb/204Pb ratios too low at a given 206Pb/204Pb value

(Fig. 9). Among the analysed phlogopites, one sample

shows Pb isotopic compositions consistent with a con-

tribution from EM2 (i.e. a high 208Pb/204Pb value and a

composition comparable with that of the Ahaggar tho-

leiites) whereas the two other phlogopites are isotopic-

ally comparable with the In Teria peridotites (Fig. 9c).

These features are reminiscent of previous observa-

tions of isotopic decoupling in mantle rocks that have

been ascribed to daughter element fractionation as a re-

sult of melt–rock interactions (Ionov et al., 2002b;

Bodinier et al., 2004; Le Roux et al., 2009). The ‘instant-

aneous’ effects of daughter element fractionation as a

result of melt–rock interactions (as opposed to the

‘time-integrated’ effects of parent–daughter element

fractionation) might account for a significant part of the

isotopic variability observed in mantle rocks. Porous

melt flow and related diffusional or reaction processes

may generate isotopic covariation trends between peri-

dotite and melt end-members that diverge markedly

from mixing lines.

Percolation modelWe used the one-dimensional percolation–diffusion

model of Vasseur et al. (1991), modified to include iso-

topic homogenization (Bodinier et al., 2004), to evaluate

a process involving percolation of melt through a litho-

spheric mantle column to generate the Nd–Sr–Pb iso-

topic compositions observed in the In Teria mantle

xenolith suite. We consider in the model a protolith with

an EM1 signature, even if the mantle was probably het-

erogeneous in composition, varying between DMM and

EM1. The model incorporates the effects of melt infiltra-

tion velocity, distance of percolation, critical distance of

isotopic homogenization along the column, chemical

diffusional exchange between melt and minerals,

and mineral grain sizes. It assumes instantaneous

solid–liquid equilibrium at the surface of mineral grains,

considered to be spherical, and chemical diffusion

within the grains. Isotopic equilibrium between melt

and minerals is governed by the mass-balance equation

of isotopic homogenization at the scale of critical vol-

umes. The critical volume of isotopic homogenization is

defined as the volume of percolated peridotite in which

melt and minerals have reached isotopic equilibriumduring the critical time of percolation (i.e. the time it

takes for the melt to reach the top of the column).

The parameters used for modelling are compiled in

Supplementary Data Table A5 and the results are

shown in Figs 15–17. The only parameters that are com-

mon to all model runs are the isotopic compositions of

the melt and the peridotite protolith. The melt isotopic

composition was fixed to the composition of In Teriamelilitites for Nd and Sr. For Pb, we took slightly higher

values in the range of the Ahaggar and other North

African alkali basalts (Allegre et al., 1980; Dupuy et al.,

1993; Beccaluva et al., 2008). For the protolith we took

the EM1 values of Eisele et al. (2002) for Nd and Pb, and

a slightly higher value for Sr, although still within the

range of literature values (e.g. Hofmann, 2002).

The results of models to evaluate the influence of

compositional parameters (daughter element contentsin peridotite and melt, and mineral/melt partition coeffi-

cients) on the 143Nd/144Nd vs 87Sr/86Sr covariation are

shown in Figs 15 and 16. We first explored parameter

values retrieved from the composition of the peridotite

xenoliths, their host lavas and clinopyroxene equilib-

rium melts (Fig. 15). For the peridotite protolith, we tried

three Sr–Nd compositions reflecting the compositional

range observed in the studied peridotites: (1) the lessmetasomatized spinel peridotite Erem-2 (run a); (2) the

garnet peridotite Int-14-7, metasomatized by alkaline

silicate melt but devoid of carbonate-melt imprint (runs

b–d); (3) the spinel peridotites (average composition),

metasomatized by carbonate melt (runs e–h). For the

melt, we used either the average composition of the

analysed melilitites (runs a–g) or the theoretical melt in

equilibrium with clinopyroxene from the spinel perido-tites (run h). Because the composition of the percolating

melt evolved along with metasomatism, from silicate

(stage 1) to carbonate melt (stage 2), we tried two differ-

ent sets of KD. We used the experimental values of Hart

& Dunn (1993) for minerals/silicate melt (runs a–c) and

minerals/clinopyroxene partitioning, and those of

Klemme et al. (1995) for clinopyroxene/carbonate melt

partitioning (runs d–h). Among the other parameters,

only the protolith modal composition and mineral grainsizes are slightly variable (Supplementary Data Table

A5). The modes vary from that of an amphibole-free spi-

nel lherzolite (runs a and b) to the amphibole-bearing,

average composition of the In Teria spinel peridotites

(runs f–h). Mineral grain size is constant and moderately

coarse in most of the runs (Cpx radius¼ 0�15 mm), ex-

cept for runs g and h, for which we used the average

grain size of the spinel peridotites, characterized byvery fine clinopyroxene grains (Cpx radius¼ 0�01 mm).

The other parameters (length of the percolation column,

porosity, melt velocity, critical distance of isotopic


at University of C

ambridge on A


Dow

nloaded from





homogenization, and chemical diffusivity in minerals

and melt) are constant in all runs. Their effects were as-

sessed separately (Supplementary Data Fig. A6).

The results in Fig. 15 show a wide range of143Nd/144Nd vs 87Sr/86Sr covariation patterns, from

deeply concave to convex upward. All of these patternsdiffer markedly from the mixing curves between perido-

tite protoliths and melts, which are almost linear. The

shape of the covariation patterns and the degree of

decoupling are correlated with the difference in Sr/Nd

between melt and peridotite. Low Sr/Nd in the melt and/

or high Sr/Nd in the protolith [i.e. (Sr/Nd)melt/(Sr/

Nd)peridotite< 1] generate concave variation patterns

indicating that 87Sr/86Sr in the protolith is re-equili-brated with incoming melt at lower melt/rock ratio than143Nd/144Nd. This is well illustrated by the strongly con-

cave covariation patterns obtained in run h [(Sr/Nd)melt/

(Sr/Nd)peridotite¼ 0�805] and run a [(Sr/Nd)melt/(Sr/

Nd)peridotite¼ 0�285). Conversely, high Sr/Nd ratio in the

melt and/or low Sr/Nd ratio in the protolith [i.e. (Sr/

Nd)melt/(Sr/Nd)peridotite> 1] generate convex patterns

indicating that 143Nd/144Nd re-equilibrates at a lower

melt/rock ratio than 87Sr/86Sr. This pattern was obtainedwith the experiments b–d where (Sr/Nd)melt/(Sr/

Nd)peridotite¼ 1�072. This confirms that the Sr/Nd ratios

in the protolith and melt play a major role in the decou-

pling of Nd–Sr isotopic compositions during porous

melt flow (Ionov et al., 2002b; Le Roux et al., 2009).

Compared with the Sr/Nd values, the other parameters

that were explored through the numerical experiments

shown in Fig. 15 have only limited effects. Runs c and dillustrate the subtle effect of mineral/melt partition coef-

ficients as they were calculated with the same set of par-

ameters, except for distinct mineral/melt KD values (for

silicate- and carbonate-melt, respectively). Experiments

b and c illustrate the limited effect of adding 6% amphi-

bole to the peridotite mineralogy, whereas experiments

f and g show the effect of reducing the mean radius of

clinopyroxene grains from 0�15 to 0�01 mm.Despite their capability to decouple the 143Nd/144Nd

and 87Sr/86Sr ratios, the experiments shown in Fig. 15

do not fit the whole range of data for the In Teria xeno-

liths. The samples that are moderately enriched in 87Sr

(87Sr/86Sr<0�704) can be accounted for by the experi-

ments b–d (run with a relatively low Sr/Nd value for the

protolith, based on the composition of the garnet peri-

dotite). However, the other samples are not fitted by

these experiments and would require even lower Sr/Ndratio in the protolith, or a higher Sr/Nd value in the per-

colating melt. As a consequence, we ran further experi-

ments to explore the effect of compositional variations

in Sr and Nd (Fig. 16). For this approach, we first calcu-

lated a reference model (short-dashed lines) using the

Sr–Nd composition of Primitive Mantle (McDonough &

Sun, 1995) for the peridotite protolith and that of

the In Teria melilitites for melt [(Sr/Nd)melt/(Sr/Nd)peridotite¼ 0�956]. Then we simulated variations of

the Sr–Nd contents from 0�33 to three times the concen-

trations of the reference model (Fig. 16a and b), and

variations of the Sr/Nd ratios from 10 to 20 (Fig. 16c and

d), by varying the Nd contents (Supplementary Data

Table A5). The other parameters are the same as for ex-

periment g in Fig. 15, except for the minerals/melt parti-

tion coefficients for which we used the silicate-melt

values, instead of the carbonate-melt values.As shown by the similarity between the reference

model and the mixing line, these experiments confirm

that the decoupling of Nd and Sr isotopic compositions

is virtually non-existent when the peridotite protolith and

the incoming melt have similar Sr/Nd ratios, and this re-

sult remains true within a wide range of Sr and Nd abso-

lute concentrations (Fig. 16a and b). Conversely, the

results shown in Fig. 16b and c indicate that even moder-ate variations of the (Sr/Nd)melt/(Sr/Nd)peridotite ratio gen-

erate significant isotopic decoupling. Almost the whole

range of the Nd–Sr isotopic variations observed in the In

Teria mantle xenoliths can be accounted for by experi-

ments involving (Sr/Nd)melt/(Sr/Nd)peridotite values in the

range 1�15–1�30. This value may result from a sub-chon-

dritic Sr/Nd ratio (10–12) in the protolith or from a super-

chondritic ratio (�20) in the percolating melt. The latter

alternative is more likely, as mantle rocks rarely showsuch low Sr/Nd values, whereas elevated Sr/Nd values

can be found in carbonate melts (e.g. �58 in carbonatites

from Morocco; Wagner et al., 2003). Alternatively, the In

Teria Nd–Sr isotopic compositions can be explained by a

combination of moderately sub-chondritic Sr/Nd values

in the peridotite (e.g. �14�2 as in the garnet peridotite Int-

14-7) and moderately super-chondritic Sr/Nd values in

melt [e.g. �17�14, as in ocean island basalt (OIB);McDonough & Sun, 1995]. Moreover, the decoupling can

be amplified by a short distance of isotopic homogeniza-

tion, thus requiring lower (Sr/Nd)melt/(Sr/Nd)peridotite val-

ues. The influence of parameters on Nd–Sr decoupling is

discussed in the Supplementary Material (Fig. A6).

It is worth noting that the peculiar Sr–Nd isotopic sig-

nature of the In Teria phlogopite megacrysts can be ex-

plained by the same two-component porous-flow modelas for the spinel peridotites, but requires either lower (Sr/

Nd)melt/(Sr/Nd)peridotite values (Fig. 16c and d) or a longer

distance of isotopic homogenization (Supplementary

Data Fig. A6f), or a combination of these. This result

lends support to the hypothesis that the phlogopites and

the metasomatized spinel peridotites are related to a sin-

gle event of interaction between the lithospheric mantle

and sublithospheric melts, but the former record inter-

action at relatively high temperature involving silicatemelt, whereas the spinel peridotites record lower tem-

perature interaction with more evolved, carbonate melt

(Fig. 14). However, two spinel peridotite samples (Erem-2

and In-15-1) remain unexplained by the model, showing

elevated 143Nd/144Nd at a given 87Sr/86Sr value. The iso-

topic composition of these samples requires a compo-

nent characterized by a 143Nd/144Nd signature higher that

both the EM1 signature and the composition of theAhaggar basalts. Mantle xenoliths with Nd isotopic com-

positions recording long-term depletion in LREE have

been reported from Ahaggar and other North African


at University of C

ambridge on A


Dow

nloaded from








localities (Beccaluva et al., 2007, 2008). In addition, the208Pb/204Pb compositions of several samples also point

to the involvement of a depleted component with a

DMM-like signature (Fig. 9c).

To evaluate how much of the Pb isotopic systematics

can be explained by the two-component porous-flow

model, and how much requires a third, DMM compo-

nent, we carried out experiments simulating covari-

ations of 207Pb/204Pb and 208Pb/204Pb vs 87Sr/86Sr (Fig.

17). We ran several experiments involving various com-

binations of Pb/Sr ratios in the protolith and melt

(Supplementary Data Table A5) by varying the Pb con-

tent in the protolith and melt (207Pb/204Pb vs 87Sr/86Sr,

Fig. 17a), or in the melt only (208Pb/204Pb vs 87Sr/86Sr,

Fig. 17b). The other parameters, including the Sr con-

tent in the protolith [¼ PM after McDonough & Sun

(1995)] and melt (¼ In Teria melilitite), are the same as

in the reference model in Fig. 16. The results for the207Pb/204Pb vs 87Sr/86Sr covariation (Fig. 17a) show that

the whole In Teria array can be accounted for by the

two-component model within a narrow range of (Pb/

Sr)melt/(Pb/Sr)peridotite values between 0�7 and 0�9 (runs

c–f). For Pb/Sr values in melt consistent with the In Teria

melilitites (Pb/Sr¼ 0�00371 6 0�00015) and OIB values

(0�0048; McDonough & Sun, 1995), this constrains the

Pb/Sr values of the peridotite protolith in the range

0�004–0�007, consistent with the lower range observed

in the In Teria spinel peridotites.

On the 208Pb/204Pb vs 87Sr/86Sr diagram (Fig. 17b),

however, although the majority of the spinel peridotites

may be accounted for by the two-component model

within the same range of (Pb/Sr)melt/(Pb/Sr)peridotite val-

ues as for 208Pb/204Pb (runs c and d), several samples

characterized by lower 208Pb/204Pb values than EM1 re-

quire a third component with a DMM signature. These

samples notably include the garnet peridotite, the

poorly metasomatized spinel peridotite Erem-2, four

other spinel peridotites and a phlogopite megacryst. As

the 208Pb/204Pb composition of these samples is distinct

from that of the volcanism in In Teria and in the whole

Ahaggar region, the third, low 208Pb/204Pb component

probably records isotopic heterogeneities in the mantle

protolith, rather than in the incoming melt. The litho-

spheric mantle beneath In Teria was probably not uni-

formly EM1-like before the onset of metasomatism; it

probably included a DMM peridotite component, as

well as some peridotites with elevated 143Nd/144Nd val-

ues recording long-term LREE depletion.

Implications for the origin and composition ofthe lithospheric mantle and for the nature oflithosphere–asthenosphere interactions beneaththe Ahaggar SwellThe In Teria mantle xenoliths differ from their counter-

parts from the Ahaggar Swell and the majority of North

African mantle xenoliths in three respects: (1) the pres-

ence of rare garnet peridotites; (2) an extensive carbon-

ate-melt metasomatic imprint; (3) an enriched 87Sr/86Sr

signature. The elevated 87Sr/86Sr compositions are con-

sidered to be inherited from an EM1 lithospheric proto-

lith providing the source of the metasomatic melt. The

existence of an EM1 signature in the lithospheric mantle

of the Tuareg Shield was first inferred from the Pb–Sr–

Nd isotopic systematics of the Ahaggar tholeiiticvolcanism (Allegre et al., 1981; Maza et al., 1998; Aıt-

Hamou et al., 2000).

In contrast to other xenolith localities from the

Ahaggar domain (Manzaz, Beccaluva et al., 2007;

Tahalgha, Kourim et al., 2014), In Teria is not situated

on the Tuareg Shield but stands on the western margin

of the ‘Saharan Metacraton’ (Abdelsalam et al., 2002,

and references herein) (Fig. 1a). The latter is consideredas a former Archaean to Paleoproterozoic craton that

was remobilized (‘decratonized’) during the

Neoproterozoic, but behaved as a relatively rigid block

before this time. Hence the possibility cannot be

excluded that the lithospheric mantle beneath In Teria

was originally different (thicker and possibly more en-

riched) than that sampled by the Ahaggar mantle xeno-

liths. However, the existence of an EM1 signature both

in the Ahaggar continental tholeiites and in the In Teriaxenoliths and megacrysts suggests that this signature

was widely distributed in the lithospheric mantle be-

neath the Tuareg Shield and adjacent areas. For

Beccaluva et al. (2007), the absence of this signature in

the Ahaggar (Manzaz) xenoliths is thus ‘perplexing’.

Those researchers interpreted the mantle beneath the

Ahaggar Swell as newly accreted material, owing to the

lateral displacement and replacement of old litho-spheric mantle by upwelling asthenosphere. They also

suggested that the In Teria xenoliths could ‘plausibly

represent the (EM1-metasomatized) older cratonic litho-

spheric mantle’.

In fact, the In Teria mantle xenoliths differ from true

cratonic mantle in several respects (e.g. their relatively

low Mg# values compared with kimberlite-hosted peri-

dotite xenoliths and their dominantly fertile, lherzolitecompositions) and are much more similar to mantle

xenoliths from younger lithospheric domains (Menzies,

1990; O’Reilly et al., 2001, and references herein). This

lends support to the hypothesis that the destabilization

of the Saharan Metacraton in the Neoproterozoic was

associated with extensive rejuvenation of the litho-

sphere, which occurred either regionally as a result of

its delamination and thermo-mechanical erosion after

thickening (Ashwal & Burke, 1989; Black & Liegeois,1993; Abdelsalam et al., 2011) or more locally along

mega-shear zones (Liegeois et al., 2003; Fezaa et al.,

2010). Despite their strong metasomatic imprint, the In

Teria xenoliths have nevertheless preserved some min-

eralogical and geochemical features that may be

ascribed to the original Pan-African lithosphere, prior to

the Cenozoic events. These include remnants of the

EM1 lithospheric isotopic signature, as predicted byBeccaluva et al. (2007). The EM1 signature was possibly

inherited from subduction processes that occurred dur-

ing the Pan-African orogeny in conjunction with the


at University of C

ambridge on A


Dow

nloaded from



amalgamation of lithospheric terranes (Black et al.,

1994; Liegeois et al., 1994; Caby, 2003).

Apart from the probable existence of a widespread

lithospheric EM1 signature, the nature of the litho-

spheric mantle beneath the Tuareg Shield and adjacent

areas prior to the Cenozoic asthenosphere–lithosphereinteractions is otherwise virtually unknown. In particu-

lar, the lateral variations in lithospheric thickness and

composition that would be expected from the juxtapos-

ition of lithospheric blocks of different provenances and

ages (e.g. Liegeois et al., 2005) are poorly constrained.

In addition to the scarcity of data, the lack of informa-

tion also reflects the modifications that have affected

the lithospheric mantle during the Cenozoic. Thesemodifications were particularly severe in the central

part of the Ahaggar Swell where the lithosphere was

strongly modified by extensive interaction with plume

melts (Dautria & Lesquer, 1989) or, possibly, rejuven-

ated by removal of old lithospheric material by upwell-

ing asthenosphere (Beccaluva et al., 2007). The

modification of the lithosphere was less intense in the

In Teria district where melt–rock interactions culmi-

nated in the pervasive infiltration of small-volume car-bonate melts. The persistence of a relatively thick

thermal boundary layer beneath In Teria would have

facilitated the formation of carbonate melts by incipient

lithospheric melting during transient heating and/or

through the evolution of primary silicate melts via melt–

rock reactions (Wallace & Green, 1988; Dalton & Wood,

1993; Sweeney, 1994; Yaxley & Green, 1996).

In scenarios involving the impingement of a mantleplume head centred on the Ahaggar Swell, or any alter-

native involving upper mantle diapiric instabilities

(Davies & Bunge, 2006; Lustrino et al., 2007) or shallow

asthenospheric upwelling (Beccaluva et al., 2007), the

evolution recorded by the In Teria mantle xenolith suite

may simply reflect the attenuated effects of this process

owing to the distal location of In Teria, to the NE of the

Ahaggar Swell (Fig. 1). As noted by Aıt Hamou &Dautria (1994), this scheme is supported for the

Ahaggar Swell by the spatio-temporal evolution of the

volcanism, varying from volumetrically predominant

Eocene to Oligocene plateau tholeiites in the central

part of the Swell (Taharaq district) to smaller volumes

of Miocene to Quaternary undersaturated alkaline lavas

in the outer parts. The isotopic signature of the In Teria

melilitites is virtually indistinguishable from that of the

Ahaggar alkaline lavas (Fig. 9) and thus this recent vol-canic district might be viewed as the most distal mani-

festation of the asthenosphere–lithosphere interactions

related to the Ahaggar Swell. However, available heat

flow data (Lesquer et al., 1989, 1990; Takherist &

Lesquer, 1989) do not support this model. In Teria is sit-

uated on an east–west high heat flow axis that trends

from Libya towards the Canary Islands, between the

Sahara Basins to the north and the Ahaggar Swell andthe West African Craton to the south (Lesquer et al.,

1990). This anomaly is a prominent feature of the heat

flow pattern in northwestern Africa, showing its

maximum values (100–120 mW m–2) in southern

Algeria. As suggested by Lesquer et al. (1990), the In

Teria volcanism might be related to the mantle proc-

esses that are responsible for this anomaly. In this re-

spect, it may be worth noting that mantle xenoliths

comparable with those from In Teria—and also brought

to the surface by dominantly undersaturated lavas—are

found along the western extension of the anomaly, in

southern Morocco (Jbel Saghro, Ibhi et al., 2002) and in

the Canary Islands (e.g. Neumann et al., 2002). The simi-

larities notably include the presence of abundant reac-

tion aggregates containing microgranular secondary

phases and glass, and evidence for interaction with car-

bonate melts. Alkaline pyroxenites and wehrlites are

common, sometimes accompanied by carbonated peri-

dotites or carbonatites. The significance of the North

Saharan thermal anomaly is poorly understood, how-

ever. Edge-driven convection (Missenard & Cadoux,

2012) was recently proposed for southern Morocco,

where the western termination of the anomaly runs

along the northern border of the West African Craton. In

Algeria, however, this model is consistent with the ob-

servation that the anomaly cross-cuts at a high angle

the north–south lithospheric structures resulting from

terrane amalgamation and shear-zone activity during

the Pan-African orogeny (Black et al., 1994).

CONCLUSIONS

The In Teria xenolith suite records three different meta-

somatic imprints, which all postdate the last deform-

ation episode in the lithospheric mantle of this region.

Stage 1 is a diffuse infiltration of hydrous alkaline melts

affecting the spinel and garnet peridotites. Melt percola-

tion and the crystallization of the reaction products dur-

ing this episode are controlled by the pre-existing

deformation microstructure of the peridotite. This first

metasomatic stage was probably coupled with a heat-

ing of the deep subcontinental lithosphere by melts of

an asthenospheric origin. Stage 2 is a carbonate melt

metasomatic imprint restricted to the spinel peridotites,

which largely overprints stage 1. Finally, stage 3 corres-

ponds to the crystallization of pyroxenites, which are

probably crystal segregates from hydrous alkaline

melts crystallized in vein conduits. The three metasom-

atic imprints represent successive stages of interaction

between the lithosphere and sublithospheric melts.

The In Teria peridotites with enriched 87Sr/86Sr com-

positions, as well as phlogopite megacrysts, plotting

within the field of the Ahaggar tholeiites, show evidence

for the involvement of an EM1-like component, prob-

ably inherited from the lithospheric mantle beneath In

Teria. However, the isotopic composition of the In Teria

spinel peridotites differs from the end-member EM1 sig-

nature, which might be accounted for by isotopic

decoupling in mantle rocks. Assuming that the Ahaggar

alkali basalts and the In Teria melilitites represent melts

of essentially sublithospheric origin, numerical model-

ling simulating porous melt percolation through an


at University of C

ambridge on A


Dow

nloaded from


EM1 lithospheric mantle can reproduce the high87Sr/86Sr values observed in some peridotite xenoliths.

These high 87Sr/86Sr values may be related to the crys-

tallization of interstitial micro-phases from small-vol-

ume melts and chromatographic effects. Conversely,

the HIMU signature observed in the In Teria melilitites,

pyroxenites and several spinel peridotites, as well as

North African alkali basalts, is widely considered to rep-

resent a mantle plume signature. Additionally, the208Pb/204Pb composition of several samples points to

the involvement of a depleted component with a DMM

signature. The lithospheric mantle beneath In Teria was

probably not uniformly EM1-like; it probably included a

DMM peridotite component as well as some peridotites

with elevated 143Nd/144Nd values recording long-term

LREE depletion.

The In Teria mantle xenoliths differ from North

African and Ahaggar Swell mantle xenoliths by the

presence of garnet peridotite, the extensive carbonate

melt metasomatic imprint, and their high 87Sr/86Sr

ratios. The last are considered to be inherited from an

EM1-like lithospheric mantle protolith. The similar EM1

signature in Ahaggar continental tholeiites suggests

that this signature extends over the entire Tuareg

Shield and adjacent areas and that the generation of the

tholeiites involved melting of the base of the litho-

sphere. The EM1 signature was possibly inherited from

subduction processes that occurred during the Pan-

African orogeny. In Teria mantle xenoliths therefore

have geochemical characteristics similar to those of

young lithospheric mantle rather than cratonic mantle

domains, supporting the hypothesis that the destabil-

ization of the Saharan Metacraton in the Neoproterozoic

was associated with extensive lithosphere rejuvenation.

ACKNOWLEDGEMENTS

C. Nevado and D. Delmas supplied high-quality pol-

ished thin sections for EBSD measurements. D.

Mainprice is thanked for providing software for analyz-

ing and plotting CPO data. We are also grateful to P.

Verdou for his help during thermal ionization mass

spectrometry analyses at the GIS laboratory (Nımes)

and to P. Telouk for his help during MC-ICP-MS ana-

lyses at the Service Commun National of the Ecole

Normale Superieure de Lyon. We thank editors M.

Wilson and D. Weiss for constructive comments and we

also gratefully acknowledge G. Bianchini, J. Berger and

an anonymous reviewer for providing constructive

reviews.

FUNDING

This study was performed as part of a collaborative

multi-disciplinary research project on the Hoggar region

chiefly involving the Faculte des Sciences de la Terre,

de Geographie et d’Amenagement du Territoire

(FSTGAT, USTHB, Algiers) and Geosciences

Montpellier, funded by the Centre National de la

Recherche Scientifique (CNRS, France: INSU and

DERCI), the Direction de la Post-Graduation Recherche-

Formation (DPGRF, Algeria), and the French Ministry of

Foreign Affairs (MAE), through a CNRS–DPGRF PICS

co-operation project (2008–2010), an MAE bilateral co-

operation project in the frame of the Tassili Hubert

Curien programme (2009–2012), and several INSU re-

search grants (SEDIT 2007, Actions Coordonnees 2008,

and SYSTER 2010–2011 programmes). M.-A.K.’s post-

doctoral research in Montpellier was funded by the

Institut National des Sciences de l’Univers–Centre

National de la Recherche Scientifique (INSU–CNRS,

France). M.A.K. also thanks the Australian Research

Council (Grant DP0878453) and the Swiss National

Science Foundation (SNSF) from Ambizione fund NSF-

26083906 for funding part of this research. The EBSD–

SEM national facility in Montpellier is supported by the

INSU–CNRS and by the Conseil Regional Languedoc–

Roussillon, France.

SUPPLEMENTARY DATA

Supplementary data for this paper are available at

Journal of Petrology online.

REFERENCES

Abdelsalam, M. G., Liegeois, J. P. & Stern, R. J. (2002). The

Saharan Metacraton. Journal of African Earth Sciences 34,

119–136.Abdelsalam, M. G., Gao, S. S. & Liegeois, J. P. (2011). Upper

mantle structure of the Saharan Metacraton. Journal of

African Earth Sciences 60, 328–336.Adam, J., Green, T. H. & Sie, S. H. (1995). Proton microprobe

determined partitioning of Rb, Sr, Ba, Y, Zr, Nb and Ta be-

tween experimentally produced amphiboles and silicate

melts with variable F content. Chemical Geology 109, 29–49.Aıt-Hamou, F. (2000). Un exemple de ‘point chaud’ intra-contin-

ental en contexte de plaque quasi-stationnaire: etude petro-

logique et geochimique du djebel Taharaq et evolution du

volcanisme cenozoıque de l’Ahaggar (Sahara algerien).

Universite Montpellier 2, 240 pp.Aıt-Hamou, F. & Dautria, J. M. (1994). Le magmatisme

Cenozoique du Hoggar: une synthese des donnees disponi-

bles. Mise au point sur l’hypothese d’un point chaud.

Bulletin du Service de la Carte Geologique en Algerie 5,

49–68.Aıt-Hamou, F., Dautria, J. M., Cantagrel, J. M., Dostal, J. &

Briqueu, L. (2000). Nouvelles donnees geochronologiques

et isotopiques sur le volcanisme Cenozoıque de l’Ahaggar

(Sahara algerien): des arguments en faveur du panache.

PhD thesis, Comptes Rendus de l’Academie des Sciences

330, 829–836.Allegre, C. J., Brevart, O., Dupre, B. & Minster, J.-F. (1980).

Isotopic and chemical effects produced in a continuously dif-

ferentiating convecting Earth mantle. Philosophical

Transactions of the Royal Society of London 297, 447–477.Allegre, C. J., Dupre, B., Lambret, B. & Richard, P. (1981). The

subcontinental versus suboceanic debate, I. Lead–

neodymium–strontium isotopes in primary alkali basalts

from a shield area: The Ahaggar volcanic suite. Earth and

Planetary Science Letters 52, 85–92.


at University of C

ambridge on A


Dow

nloaded from



Ashwal, L. D. & Burke, K. (1989). African lithospheric structure,

volcanism and topography. Earth and Planetary Science

Letters 96, 8–14.Ayadi, A., Dorbath, C., Lesquer, A. & Bezzeghoud, M. (2000).

Crustal and upper mantle velocity structure of the Hoggar

swell (Central Sahara). Physics of the Earth and Planetary

Interiors 118, 111–123.Beccaluva, L., Siena, F., Coltorti, M., Di Grande, A., Lo Giudice,

A., Macciotta, G., Tassinari, R. & Vaccaro, C. (1998).

Nephelinitic to tholeitic magma generation in a transten-

sional tectonic setting: an integrated model for the Iblean

volcanism, Sicily. Journal of Petrology 39, 1547–1576.Beccaluva, L., Azzouni-Sekkal, A., Benhallou, A., Bianchini, G.,

Ellam, R. M., Marzola, M., Siean, F. & Stuart, F. M. (2007).

Intracratonic asthenosphere upwelling and lithosphere reju-

venation beneath the Hoggar swell (Algeria): Evidence from

HIMU metasomatised lherzolite mantle xenoliths. Earth and

Planetary Science Letters 260, 482–494.Beccaluva, L., Bianchini, G., Ellam, R. M., Marzola, M., Oun, K.

M., Siena, F. & Stuart, F. M. (2008). The role of HIMU meta-

somatic components in the North African lithospheric man-

tle: petrological evidence from the Gharyan lherzolite

xenoliths, NW Libya. In: Coltorti, M. & Gregoire, M. (eds)

Metasomatism in Oceanic and Continental Lithosphere

Mantle. Geological Society, London, Special Publications

293, 253–277.Bedini, R. M. (1994). Studio geochimico-petrologico degli xeno-

liti ultrafemici di mega (Sidamo, Ethiopia—rift Africano

orientale—): Implicazioni sui processi di interazione

magma–roccia nel mantello litosferico subcontinentale. PhD

thesis, Dipartimento di Scienze Mineralogiche dell’Bologna,

Universita di Bologna.Bedini, R. M. & Bodinier, J. L. (1999). Distribution of incompat-

ible trace elements between the constituents of spinel peri-

dotite xenoliths: ICP-MS data from the East African Rift.

Geochimica et Cosmochimica Acta 63, 3883–3900.Bedini, R. M., Bodinier, J. L., Dautria, J. M. & Morten, L. (1997).

Evolution of LILE-enriched small melt fractions in the litho-

spheric mantle; a case study from the East African Rift. Earth

and Planetary Science Letters 153, 67–83.Begg, G. C., Griffin, W. L., Natapov, L. M., O’Reilly, S. Y., Grand,

S. P., O’Neill, C. J., Hronsky, J. M. A., Poudjom Djomani, Y.

H., Swain, C. J., Deen, T. & Bowden, P. (2009). The litho-

spheric architecture of Africa: Seismic tomography, mantle

petrology, and tectonic evolution. Geosphere 5, 23–53.Bertrand, J. M., Michard, A., Boullier, A. M. & Dautel, D. (1986).

Structure and U/Pb geochronology of Central Hoggar

(Algeria): a reappraisal of its Pan-African evolution.

Tectonophysics 5, 955–972.Bestmann, M. & Prior, D. J. (2003). Intragranular dynamic

recrystallization in naturally deformed calcite marble: diffu-

sion accommodated grain boundary sliding as a result of

subgrain rotation recrystallization. J Struct Geol 25,

1597–1613.Bianchini, G., Yoshikawa, M. & Sapienza, G. T. (2010).

Comparative study of ultramafic xenoliths and associated

lavas from South-Eastern Sicily: nature of the lithospheric

mantle insights on magma genesis. Mineralogy and

Petrology 98, 111–121.Black, R. & Liegeois, J. P. (1993). Craton, mobile belts, alkaline

rocks and continental lithospheric mantle: The Pan-African

testimony. Journal of the Geological Society, London 150,

89–98.Black, R., Latouche, L., Liegeois, J. P., Caby, R. & Bertrand, J.-M.

(1994). Pan-Africa displaced terranes in the Tuareg Shield

(central Sahara). Geology 22, 641–644.

Blundy, J. D. & Dalton, J. (2000). Experimental comparison of

trace element partitioning between clinopyroxene and melt

in carbonate and silicate systems, and implications for man-

tle metasomatism. Contributions to Mineralogy and

Petrology 139, 356–371.Bodinier, J. L. (1989). Distribution des terres rares dans les mas-

sifs lherzolitiques de Lanzo et de l’Ariege. Montpellier:

Centre Geologique et Geophysique de Montpellier,

Universite des Sciences et Techniques du Languedoc,

177 pp.Bodinier, J. L., Guiraud, M., Fabries, F., Dostal, J. & Dupuy, C.

(1987). Petrogenesis of layered pyroxenites from the Lherz,

Freychinede and Prades ultramafic bodies (Ariege, French

Pyrenees). Geochimica et Cosmochimica Acta 51, 279–290.Bodinier, J. L., Vasseur, G., Vernieres, J., Dupuy, C. & Fabries,

J. (1990). Mechanisms of mantle metasomatism: geochem-

ical evidence from the Lherz orogenic peridotite. Journal of

Petrology 31, 597–628.Bodinier, J. L., Menzies, M. A., Shimizu, N., Frey, F. A. &

McPherson, E. (2004). Silicate, hydrous and carbonate meta-

somatism at Lherz, France: Contemporaneous derivatives of

silicate melt–harzburgite reaction. Journal of Petrology 45,

299–320.Bodinier, J. L., Garrido, C., Chanefo, I., Bruguier, O. & Gervilla,

F. (2008). Origin of pyroxenite–peridotite veined mantle by

refertilization reactions: evidence from the Ronda peridotite

(Southern Spain). Journal of Petrology 49, 999–1025.Boissiere, G. & Megartsi, M. (1982). Petrologie des nodules de

pyroxenolites associes �a la rushayite d’In Teria (NE d’Illizi,

ex Fort Polignac), Algerie. Bulletin de Mineralogie 105,

89–98.Bonadiman, C., Beccaluva, L., Coltorti, M. & Siena, F. (2005).

Kimberlite-like metasomatism and ‘garnet signature’ in spi-

nel-peridotite xenoliths from Sal, Cape Verde Archipelago:

Relics of a subcontinental mantle domain within the Atlantic

Oceanic lithosphere?. Journal of Petrology 46, 2465–2493.Bouabdellah, M., Hoernle, K., Kchit, A., Duggen, S., Hauff, F.,

Klugel, A., Lowry, D. & Beaudoin, G. (2010). Petrogenesis of

the Eocene Tamazert continental carbonatites (Central High

Atlas, Morocco): Implications for a common source for the

Tamazert and Canary and Cape Verde Island carbonatites.

Journal of Petrology 51, 1655–1686.Brenan, J. M., Shaw, H. F., Ryerson, F. J. & Phinney, D. L.

(1995). Experimental determination of trace-element parti-

tioning between pargasite and a synthetic hydrous andesitic

melt. Earth and Planetary Science Letters 135, 1–11.Brey, G. P. & Kohler, T. (1990). Geothermobarometry in four-

phase lherzolites II. New thermobarometers, and practical

assessment of existing thermobarometers. Journal of

Petrology 31, 1353–1378.Caby, R. (2003). Terrane assembly and geodynamic evolution

of central–western Hoggar: a synthesis. Journal of African

Earth Sciences 37, 133–159.Caby, R., Bertrand, J. M. & Black, R. (1981). Pan-African closure

and continental collision in the Hoggar–Iforas segment,

Central Sahara. In: Kroner, A. (ed.) Precambrian Plate

Tectonics. Amsterdam: Elsevier, pp. 407–434.Canil, D. (2004). Mildly incompatible elements in peridotites

and the origin of mantle lithosphere. Lithos 77, 375–393.Chakhmouradian, A. R. (2006). High-field-strength elements in

carbonatitic rocks: Geochemistry, crystal chemistry and sig-

nificance for constraining the source of carbonatites.

Chemical Geology 235, 138–160.Coltorti, M., Bonadiman, C., Hinton, R. W., Siena, F. & Upton, B.

G. J. (1999). Carbonatite metasomatism of the oceanic

upper mantle: evidence from clinopyroxenes and glasses in


at University of C

ambridge on A


Dow

nloaded from


ultramafic xenoliths of Grande Comore, Indian Ocean.

Journal of Petrology 40, 133–165.Coltorti, M., Beccaluva, L., Bonadiman, C., Salvini, L. & Siena, F.

(2000). Glasses in mantle xenoliths as geochemical indica-

tors of metasomatic agents. Earth and Planetary ScienceLetters 183, 303–320.

Conrad, G. (1969). L’evolution continentale post-hercyniennedu Sahara algerien (Saoura, Erg Checch-Tanezrouft, Ahnet-

Mouydir). Paris: CNRS-Paris.Crough, S. T. (1981). Free-air gravity over the Ahaggar massif,

North West Africa: evidence for alteration of the lithosphere.Tectonophysics 77, 189–202.

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.Dautria, J. M. & Lesquer, A. (1989). An example of the relation-

ship between rift and dome: recent geodynamic evolution ofthe Hoggar swell and of its nearby regions (Central Sahara,

Southern Algeria and Eastern Niger). Tectonophysics 163,45–61.

Dautria, J. M., Dupuy, C., Takherist, D. & Dostal, J. (1992).Carbonate metasomatism in the lithospheric mantle: perido-

titic xenoliths from a melilitic district of the Sahara basin.Contributions to Mineralogy and Petrology 111, 37–52.

Davies, J. H. & Bunge, H. P. (2006). Are splash plumes the origin

of minor hotspots?. Geology 34, 349–352.Delaney, J. S., Smith, J. V., Carswell, D. A. & Dawson, J. B.

(1980). Chemistry of micas from kimberlites and xenoliths.2. Primary-textured and secondary-textured micas from

peridotite xenoliths. Geochimica et Cosmochimica Acta 44,857–872.

Demouchy, S., Tommasi, A., Boffa-Ballaran, T. & Cordier, P.(2013). Low strength of Earth’s uppermost mantle inferred

from tri-axial deformation experiments on dry olivine crys-tals. Physics of the Earth and Planetary Interiors 220, 37–49.

Downes, H. (2001). Formation and modification of the shallow

sub-continental lithospheric mantle: a review of geochem-ical evidence from ultramafic xenoliths suites and tectonic-ally emplaced ultramafic massifs of Western and CentralEurope. Journal of Petrology 42, 233–250.

Dupuy, C., Chikhaoui, M. & Dostal, J. (1993). Trace element andisotopic geochemistry of Cenozoic alkali basaltic lavas fromthe Atakor (Central Sahara). Geochemical Journal 27,131–145.

Ebinger, C. J. & Sleep, N. H. (1998). Cenozoic magmatismthroughout east Africa resulting from impact of a singleplume. Nature 395, 788–791.

Eisele, J., Sharma, M., Galer, S. J. G., Blichert-Toft, J., Devey, C.W. & Hofmann, A. W. (2002). The role of sediment recyclingin EM-1 inferred from Os, Pb, Hf, Nd, Sr isotope and traceelement systematics of the Pitcairn hotspot. Earth andPlanetary Science Letters 196, 197–212.

Fezaa, N., Liegeois, J. P., Abdallah, N., Cherfouh, E. H., DeWaele, B., Bruguier, O. & Ouabadi, A. (2010). Late Ediacarangeological evolution (575–555 Ma) of the Djanet Terrane,

Eastern Hoggar, Algeria, evidence for Murzukian intraconti-nental episode. Precambrian Research 180, 299–327.

Forte, A. M., Quere, S., Moucha, R., Simmons, N. A., Grand, S.P., Mitrovica, J. X. & Rowley, D. B. (2010). Joint seismic–

geodynamic–mineral physical modelling of African geody-namics: A reconciliation of deep-mantle convection withsurface geophysical constraints. Earth and PlanetaryScience Letters 295, 329–341.

Frey, F. A. (1980). The origin of pyroxenites and garnet pyroxen-ites from Salt Lake Crater, Oahu, Hawaii—Trace element evi-dence. American Journal of Science 280, 427–449.

Frey, F. A. & Green, D. H. (1974). The mineralogy, geochemistry

and origin of lherzolite inclusions in Victorian basanites.Geochimica et Cosmochimica Acta 38, 1023–1059.

Garrido, C. J. & Bodinier, J. L. (1999). Diversity of mafic rocks inthe Ronda peridotite: evidence for pervasive melt–rock reac-tion during heating of subcontinental lithosphere by upwell-

ing asthenosphere. Journal of Petrology 40, 729–754.Gasparik, T. (1984). Experimentally determined stability of

clinopyroxeneþgarnetþ corundum in the system CaO–

MgO–Al2O3–SiO2. American Mineralogist 69, 1025–1035.Gasparik, T. (1987). Orthopyroxene thermometry in simple and

complex systems. Contributions to Mineralogy and

Petrology 96, 357–370.Girod, M. (1971). Le massif volcanique de l’Atakor (Hoggar,

Sahara Algerien). Paris: CNRS, 155 pp.Godard, G., Bodinier, J. L. & Vasseur, G. (1995). Effects of min-

eralogical reactions on trace element redistributions in man-

tle rocks during percolation processes: A chromatographicapproach. Earth and Planetary Science Letters 133, 449–461.

Green, T. H. & Ringwood, A. E. (1967). An experimental investi-

gation of the gabbro to eclogite transformation and its petro-logical applications. Geochimica et Cosmochimica Acta 31,767–833.

Gregoire, M., Moine, B. N., O’Reilly, S. Y., Cottin, J. Y. & Giret,A. (2000). Trace element residence and partitioning in man-tle xenoliths metasomatised by high alkaline silicate and

carbonate-rich melts (Kerguelen Islands, Indian Ocean).Journal of Petrology 41, 477–509.

Griffin, W. L. & O’Reilly, S. Y. (1986). Sapphirine in a mantle-derived xenolith from Delegate, Australia. MineralogicalMagazine 50, 635–640.

Guiraud, R. & Bellion, Y. (1995). Late Carboniferous to Recentgeodynamic evolution of the West Gondwanian cratonicTethyan margins. In: Nairn, A., Ricou, L. E., Vrielynck, B. &

Dercourt, J. (eds) The Ocean Basins and Margins 8, theTethys Ocean. New York: Plenum, pp. 101–124.

Guiraud, R., Bosworth, W., Thierry, J. & Delplanque, A. (2005).

Phanerozoic geological evolution of Northern and CentralAfrica: An overview. Journal of African Earth Sciences 43,

83–143.Hart, S. R. (1984). A large scale isotope anomaly in the

Southern Hemisphere mantle. Nature 309, 753–757.Hart, S. R. & Dunn, T. (1993). Experimental cpx/melt partitioning

of 24 trace elements. Contributions to Mineralogy andPetrology 113, 1–8.

Harte, B., Hunter, R. H. & Kinny, P. D. (1993). Melt geometry,movement and crystallization, in relation to mantle dykes,veins and metasomatism. Philosophical Transactions of the

Royal Society of London, Series A 342, 1–21.Hauri, E. H. & Hart, S. R. (1994). Constraints on melt migration

from mantle plumes: A trace element study of peridotitexenoliths from Savai’i, Western Samoa. Journal ofGeophysical Research 99, 24301–24321.

Hawkesworth, C. J., Mantovani, E., Taylor, P. N. & Palacz, Z.(1986). Evidence from the Parana of south Brazil for a contin-ental contribution to Dupal basalts. Nature 322, 356–359.

Hofmann, A. W. (2002). Sampling mantle heterogeneitythrough oceanic basalts: isotopes and trace elements. In:Turekian, K. & Holland, H. (eds) Geochemistry of the Mantle

and Core. Amsterdam: Elsevier, pp. 61–101.Ibhi, A., Nachit, H., Abia, E. H. & Hernandez, J. (2002). Intervention

des segregats carbonatitiques dans la petrogenese des nephe-

linites �a pyroxene de Jbel Saghro (Anti-Atlas, Maroc). Bulletinde la Societe Geologique de France 173, 37–43.

Ionov, D. A., Savoyant, L. & Dupuy, C. (1992). Application of theICP-MS technique to trace element analysis of peridotitesand their minerals. Geostandards Newsletter 16, 311–315.


at University of C

ambridge on A


Dow

nloaded from


Ionov, D., Dupuy, C., O’Reilly, S., Kopylova, M. G. & Genshaft,

Y. S. (1993a). Carbonated peridotite xenoliths fromSpitsbergen: implications for trace element signature ofmantle carbonate metasomatism. Earth and PlanetaryScience Letters 119, 283–297.

Ionov, D., Dupuy, C., O’Reilly, S. Y., Kopylova, M. G. &Genshaft, Y. S. (1993b). Carbonated peridotite xenolithsfrom Spitsbergen: implications for trace element signatureof mantle carbonate metasomatism. Earth and PlanetaryScience Letters 119, 283–297.

Ionov, D., Griffin, D. W. & O’Reilly, S. Y. (1997). Volatile-bearingminerals and lithophile trace elements in the upper mantle.Chemical Geology 141, 153–184.

Ionov, D., Bodinier, J. L., Mukasa, S. B. & Zanetti, A. (2002a).

Mechanisms and sources of mantle metasomatism: Majorand trace element compositions of peridotite xenoliths fromSpitsbergen in the context of numerical modelling. Journalof Petrology 43, 2219–2259.

Ionov, D. A., Mukasa, S. B. & Bodinier, J. L. (2002b). Sr–Nd–Pb

isotopic compositions of peridotite xenoliths fromSpitzbergen: numerical modeling indicates Sr–Nd decou-pling in the mantle by melt percolation metasomatism.Journal of Petrology 43, 2261–2278.

Irving, A. J. (1980). Petrology and geochemistry of compositeultramafic xenoliths in alkalic basalts and implications formagmatic processes within the mantle. American Journal ofScience 280, 389–426.

Irving, A. J. & Frey, F. A. (1984). Trace element abundances in

megacrysts and their host basalts: constraints on partitioncoefficients and megacryst genesis. Geochimica etCosmochimica Acta 48, 1201–1221.

Janney, P. E., Le Roex, A. P., Carlson, R. W. & Viljoen, K. S.

(2002). A chemical and multi-isotope study of the WesternCape Olivine Melilitite Province, South Africa: Implicationsfor the sources of kimberlites and the origin of the HIMU sig-nature in Africa. Journal of Petrology 43, 2339–2370.

Jung, H. & Karato, S. I. (2001). Water-induced fabric transitions

in olivine. Science 1460–1464.Kaeser, B., Kalt, A. & Pettke, T. (2006). Evolution of the litho-

spheric mantle beneath the Marsabit volcanic field(Northern Kenya): Constraints from textural, P–T and geo-

chemical studies on xenoliths. Journal of Petrology 47,2149–2184.

Kechid, S. A. & Megartsi, M. (2005). Petrogenese des xenolitesmafiques et ultramafiques des laves �a melilite d’In Teria(Illizi, Algerie). Bulletin du Service Geologique de l’Algerie

16, 127–149.Khaldi, A., Nacer, J.-E., Badani, A., Van Der Beek, P., Labrin, E.,

Chahdane, R. & Bentchakal, M. (2006). Age trace de fissionCretace Inferieur du plissement �a grand rayon de courbure

du Hoggar (Sahara Central, Algerie). Abstracts book of the4th International Geological Correlation Program IGCP-485(Algiers, December 2006), p. 15.

King, S. D. & Anderson, D. L. (1995). An alternative mechanismof flood basalt formation. Earth and Planetary Science

Letters 136, 269–279.King, S. D. & Anderson, D. L. (1998). Edge-driven convection.

Earth and Planetary Science Letters 160, 289–296.Klemme, S., van der Laan, S. R., Foley, S. F. & Gunther, D.

(1995). Experimentally determined trace and minor elementpartitioning between clinopyroxene and carbonatite meltunder upper mantle conditions. Earth and Planetary ScienceLetters 133, 439–448.

Kourim, F., Bodinier, J. L., Alard, O., Bendaoud, A., Vauchez, A.& Dautria, J. M. (2014). Nature and evolution of the litho-spheric mantle beneath the Hoggar Swell (Algeria): a recordfrom mantle xenoliths. Journal of Petrology 55, 2249–2280.

La Tourette, R. L., Hervig, R. L. & Holloway, J. R. (1995). Trace

element partitioning between amphibole, phlogopite and

basanite melt. Earth and Planetary Science Letters 135,

13–30.Le Roux, V., Bodinier, J. L., Alard, O., O’Reilly, S. & Griffin, W. L.

(2009). Isotopic decoupling during porous melt flow: A case

study in the Lherz peridotite. Earth and Planetary Science

Letters 279, 76–85.Lesquer, A., Beltrao, J. F. & De Abreu, F. A. M. (1984).

Proterozoic links between Northeastern Brazil and West

Africa: A plate tectonic model based on gravity data.

Tectonophysics 110, 9–26Lesquer, A., Bourmatte, A., Ly, S. & Dautria, J. M. (1989). First

heat flow determination from the Central Sahara: relation-

ship with the Pan-African belt and Hoggar domal uplift.

Journal of African Earth Sciences 9, 41–48.Lesquer, A., Takherist, D., Dautria, J. M. & Hadiouche, O. (1990).

Geophysical and petrological evidence for the presence of

an ‘anomalous’ upper mantle beneath the Sahara basins

(Algeria). Earth and Planetary Science Letters 96, 407–418.Liegeois, J. P., Black, R., Navez, J. & Latouche, L. (1994). Early

and late Pan-African orogenies in the Aır assembly of ter-

ranes (Tuareg shield, Niger). Precambrian Research 67,

59–88.Liegeois, J. P., Latouche, L., Mustapha, B., Navez, J. & Guiraud,

M. (2003). The LATEA metacraton (Central Hoggar, Tuareg

shield, Algeria): Behaviour of an old passive margin during

the Pan-African orogeny. Journal of African Earth Sciences

37, 161–190.Liegeois, J. P., Benhallou, A., Azzouni-Sekkal, A., Yahiaoui, R. &

Bonin, B. (2005). The Hoggar swell and volcanism:

Reactivation of the Precambrian Tuareg shield during

Alpine convergence and West African Cenozoic volcanism.

In: Foulger, G. R., Netland, J. H., Presnall, D. C. & Anderson,

D. L. (eds) Plates, Plumes and Paradigms. Geological

Society of America, Special Papers 388, 379–400.Luais, B., Telouk, P. & Albarede, F. (1997). Precise and accurate

neodymium isotopic measurements by plasma-source

mass spectrometry. Geochimica et Cosmochimica Acta 61,

4847–4854.Lustrino, M. & Wilson, M. (2007). The circum-Mediterranean

anorogenic Cenozoic igneous province. Earth-Science

Reviews 81, 1–65.Lustrino, M., Melluso, L. and Morra, V. (2007). The geochemical

peculiarity of ‘Plio-Quaternary’ volcanic rocks of Sardinia in

the circum-Mediterranean area, In: Beccaluva, L., Bianchini,

G., and Wilson, M., (eds), Cenozoic Volcanism in the

Mediterranean Area: Geological Society of America, Special

Papers 418, p. 277–301.Mahoney, J. J., Nicollet, C. & Dupuy, C. (1991). Madagascar

basalts: tracking oceanic and continental sources. Earth and

Planetary Science Letters 104, 350–363.Mainprice, D., Tommasi, A., Couvy, H. & Cordier, P. (2005).

Pressure sensibility of olivine slip systems and seismic an-

isotropy of Earth’s upper mantle. Nature 433, 731–733.Marks, M. A. W., Schilling, J., Coulson, I. M., Wenzel, T. & Markl,

G. (2008). The alkaline–peralkaline Tamazeght complex, High

Atlas mountains, Morocco: Mineral chemistry and petrologi-

cal constraints for derivation from compositionally heteroge-

neous mantle source. Journal of Petrology 49, 1097–1131.Maza, M., Briqueu, L., Dautria, J. M. & Bosch, D. (1998). Le com-

plexe annulaire d’age Oligocene de l’Achkal (Hoggar

Central, Sud Algerie): temoin de la transition au Cenozoıque

entre les magmatismes tholeitique et alcalin. Evidences par

les isotopes du Sr, Nd et Pb. Comptes Rendus de

l’Academie des Sciences 327, 167–172.


at University of C

ambridge on A


Dow

nloaded from


McDonough, W. F. & Sun, S. S. (1995). The composition of the

Earth. Chemical Geology 120, 223–253.McKenzie, D. (1989). Some remarks on the movement of small

melt fractions in the mantle. Earth and Planetary Science

Letters 95, 53–72.McPhail, D. C., Berman, R. G. & Greenwood, H. J. (1990).

Experimental and theoretical constraints on aluminum sub-stitution in magnesian chlorite, and a thermodynamical

model for H2O in magnesian cordierite. CanadianMineralogist 28, 859–874.

Megartsi, M. (1972). Etude des structures circulaires du nord-est d’Illizi (ex Fort-Polignac) Sahara Nord-Oriental. Alger:

PhD thesis, Faculte des Sciences d’Alger, 114 pp.Mengel, K. & Green, D. H. (1989). Stability of amphibole and

phlogopite in metasomatized peridotite under water-satu-

rated and water-undersaturated conditions. In: Ross, A. (ed.)Kimberlites and Related Rocks. Geological Society ofAustralia, Special Publication 14, 571–581.

Menzies, M. A. (1990). Archaean, Proterozoic and Phanerozoic

lithospheres. In: Menzies, M. A. (ed.) Continental Mantle.Oxford: Clarendon Press, pp. 67–86.

Milner, S. C. & Le Roex, A. P. (1996). Isotope characteristics ofthe Okenyenya igneous complex, northwestern Namibia:

constraints on the composition of the early Tristan plumeand the origin of the EM 1 mantle component. Earth andPlanetary Science Letters 141, 277–291.

Missenard, Y. & Cadoux, A. (2012). Can Moroccan Atlas litho-spheric thinning and volcanism be induced by edge-drivenconvection?. Terra Nova 24, 27–33.

Moine, B. N., Gregoire, M., O’Reilly, S. Y., Sheppard, S. M. F. &

Cottin, J. Y. (2001). High field strength element fractionationin the upper mantle: evidence from amphibole-rich compos-ite mantle xenoliths from the Kerguelen Islands (IndianOcean). Journal of Petrology 42, 2145–2167.

Natali, C., Beccaluva, L., Bianchini, G., Ellam, R. M., Siena, F. &Stuart, F. M. (2013). Carbonated alkali-silicate metasoma-tism in the North Africa lithosphere: Evidence from MiddleAtlas spinel-lherzolites, Morocco. Journal of South

American Earth Sciences 41, 113–121.Neumann, E. R., Wulff-Pedersen, E., Pearson, N. J. & Spencer,

E. A. (2002). Mantle xenoliths from Tenerife (CanaryIslands): Evidence for reactions between mantle peridotites

and silicic carbonatite melts inducing Ca metasomatism.Journal of Petrology 43, 825–857.

Neumann, E. R., Griffin, W. L., Pearson, N. J. & O’Reilly, S. Y.

(2004). The evolution of the upper mantle beneath theCanary Islands: Information from trace elements and Sr iso-tope ratios in minerals in mantle xenoliths. Journal ofPetrology 45, 2573–2612.

Neumann, E.-R., Abu El-Rus, M. A., Tiepolo, M., Ottilini, L., Vannucci,R. & Whitehouse, M. (2015). Serpentinization and deserpentini-zation reactions in the upper mantle beneath Fuerteventura re-vealed by peridotite xenoliths with fibrous orthopyroxene and

mottled olivine. Journal of Petrology 56, 3–31.Nickel, K. G. & Green, D. H. (1985). Empirical geothermobarom-

etry for garnet peridotites and implications for the nature ofthe lithosphere, kimberlites and diamonds. Earth and

Planetary Science Letters 73, 158–170.Niida, K. & Green, D. H. (1999). Stability and chemical compos-

ition of pargasitic amphibole in MORB pyrolite under uppermantle conditions. Contributions to Mineralogy and

Petrology 135, 18–40.Niu, Y. L., Langmuir, C. H. & Kinzler, R. J. (1997). The origin of

abyssal peridotites: a new perspective. Earth and Planetary

Science Letters 152, 251–265.Orberger, B., Lorand, J. P., Girardeau, J., Mercier, J.-C. C. &

Pitragool, S. (1995). Petrogenesis of ultramafic rocks and

associated chromitites in the Nan Uttaradit ophiolite,

Northern Thailand. Lithos 35, 153–182.O’Reilly, S. Y. & Griffin, W. L. (1987). Eastern Australia: 4000 km

of mantle samples. In: Nixon, P. H. (ed.) Mantle Xenoliths.

Chichester: John Wiley, pp. 267–280.O’Reilly, S. Y., Griffin, W. L. & Stabel, A. (1988). Evolution of

Phanerozoic eastern Australian lithosphere: isotope evi-

dence for magmatic and tectonic underplating. Journal of

Petrology, Special Volume, 89–108.O’Reilly, S. Y., Griffin, W. L., Poudjom Djomani, Y. H. & Morgan,

P. (2001). Are lithospheres forever? Tracking changes in sub-

continental lithospheric mantle through time. GSA Today

11, 4–10.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 for

NIST SRM 610 and NIST SRM 612 glass reference materials.

Geostandards Newsletter 21, 114–144.Pin, C., Briot, D., Bassin, C. & Poitrasson, F. (1994). Concomitant

extraction of Sr and Sm–Nd for isotopic analysis in silicate

samples based on specific extraction chromatography.

Analytica Chimica Acta 298, 209–217.Piromallo, C. & Faccenna, C. (2004). How deep can we find the

traces of Alpine subduction?. Geophysical Research Letters

31, 1–4.Powell, W., Zhang, M., O’Reilly, S. Y. & Tiepolo, M. (2004).

Mantle amphibole trace-element and isotopic signatures

trace multiple metasomatic episodes in lithospheric mantle,

western Victoria, Australia. Lithos 75, 141–171.Prior, D. J., Wheeler, J., Peruzzo, L., Spiess, R. & Storey, C.

(2002). Some garnet microstructures: an illustration of the

potential of orientation maps and misorientation analysis in

microstructural studies. J Struct Geol 24, 999–1011.Raffone, N., Chazot, G., Pin, C., Vannucci, R. & Zanetti, A. (2009).

Metasomatism in the lithospheric mantle beneath Middle

Atlas (Morocco) and the origin of Fe- and Mg-rich wehrlites.

Journal of Petrology 50, 197–249.Remy, J. M. (1959). Etude geologique et volcanologique du

Sud-Est de l’Amadror en Ahaggar (Sahara Central). PhD

Thesis, Paris, 189 pp.Rossi, P. L., Lucchini, F. & Savelli, C. (1979). Donnees geologi-

ques et radiometriques sur la mise en place de la Tellerteba

(Hoggar). 10th Colloque de la Geologie Africaine.

Montpellier, p. 143.Sebai, A., Stutzmann, E., Montagner, J.-P., Sicilia, D. & Beucler,

E. (2006). Anisotropic structure of the African upper mantle

from Rayleigh and Love wave tomography. Physics of the

Earth and Planetary Interiors 155, 48–62.Sweeney, R. J. (1994). Carbonatite melt compositions in the

Earth’s mantle. Earth and Planetary Science Letters 128,

259–270.Takherist, D. & Lesquer, A. (1989). Mise en evidence d’impor-

tantes variation regionales du flux de chaleur en Algerie.

Canadian Journal of Earth Sciences 26, 615–626.Taylor, W. R. & Green, D. H. (1983). Measurement of reduced

peridotite–C–O–H solidus and implications for redox melting

of the mantle. Nature 332, 349–352.Tommasi, A., Tikoff, B. & Vauchez, A. (1999). Upper mantle tec-

tonics: Three-dimensional deformation, olivine crystallo-

graphic fabrics and seismic properties. Earth and Planetary

Science Letters. 168, 173–186.Tommasi, A., Mainprice, D., Canova, G. & Chastel, Y. (2000).

Viscoplastic self-consistent and equilibrium-based model-

ing of olivine lattice preferred orientations: Implications for

the upper mantle seismic anisotropy. Journal of

Geophysical Research 105, 7893–7908.


at University of C

ambridge on A


Dow

nloaded from


Van Achterbergh, E., Ryan, C. G., Jackson, S. E. & Griffin, W. L.(2001). Data Reduction for LA-ICPMS. In: Sylvester, P. (ed.)Laser-ablation in the Earth Sciences: Principles andApplications. Ottawa, Ontario: Mineralogical Association ofCanada.

Vasseur, G., Vernieres, J. & Bodinier, J. L. (1991). Modelling oftrace element transfer between mantle melt and heterogra-nular peridotite matrix. Journal of Petrology, SpecialLherzolite Issue, 41–54.

Wagner, C., Mokhtari, A., Deloule, E. & Chabaux, F. (2003).Carbonatite and alkaline magmatism in Taourirt (Morocco):Petrological, geochemical and Sr–Nd isotope characteris-tics. Journal of Petrology 44, 937–965.

Wallace, M. E. & Green, D. H. (1988). An experimental determin-ation of primary carbonatite magma composition. Nature335, 343–346.

Wendlandt, R. F. & Eggler, D. H. (1980). Stability of sani-dineþ forsterite and its bearing on the genesis of potassicmagmas and the distribution of potassium in the uppermantle. Earth and Planetary Science Letters 51, 215–220.

White, W. M., Albarede, F. & Telouk, P. (2000). High-precisionanalysis of Pb isotope ratios by multi-collector ICP-MS.Chemical Geology 167, 257–270.

Whitehouse, M. J. & Neumann, E. R. (1995). Sr–Nd–Pb iso-tope data for ultramafic xenoliths from Hierro,

Canary Islands: Melt inflitration processes in the uppermantle. Contributions to Mineralogy and Petrology 119,239–246.

Wittig, N., Pearson, D. G., Duggen, S., Baker, J. A. & Hoernle,K. (2010). Tracing the metasomatic and magmatic evolu-tion of continental mantle roots with Sr, Nd, Hf and Pbisotopes: a case study of Middle Atlas (Morocco) perido-tite xenoliths. Geochimica et Cosmochimica Acta 74,1417–1435.

Woolley, A. R. & Kempe, D. R. C. (1989). Carbonatites: nomen-clature, average chemical compositions, and element distri-bution. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution.London: Unwin Hyman, pp. 1–14.

Yaxley, G. M. & Green, D. H. (1996). Experimental reconstruc-tion of sodic dolomitic carbonatite melts from metasoma-tised lithosphere. Contributions to Mineralogy andPetrology 124, 359–369.

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.

Zhao, D. (2007). Seismic images under 60 hotspots: Search formantle plumes. Gondwana Research 12, 335–355.

Zindler, A. & Hart, S. (1986). Chemical geodynamics. AnnualReview of Earth and Planetary Sciences 14, 493–571.


at University of C

ambridge on A


Dow

nloaded from


metasomatized mantle xenoliths as a record of the ... · france; 2department of applied geology,...

Documents