Mid-ocean ridge and supra-subduction affinities in the Pindos ophiolites (Greece): implications for magma genesis in a forearc setting

www.elsevier.com/locate/lithos

Lithos 73 (2004) 229–253

Mid-ocean ridge and supra-subduction affinities in the

Pindos ophiolites (Greece): implications for magma

genesis in a forearc setting

Emilio Saccania,*, Adonis Photiadesb

aDipartimento di Scienze della Terra, Universita di Ferrara, C.so E. I d’Este 32, 44100 Ferrara, Italyb Institute of Geology and Mineral Exploration (IGME), 70 Messoghion Str., 11527 Athens, Greece

Received 2 April 2003; accepted 9 December 2003

Abstract

The Pindos ophiolitic massif is considered an important key area within the Albanide–Hellenide ophiolitic belt and is

represented by two tectonically distinct ophiolitic units: (1) a lower unit, including an intrusive and a volcanic section; and (2)

an Upper Ophiolitic Unit, mainly including mantle harzburgites. Both units share similar metamorphic soles and tectono-

sedimentary melanges at their bases.

The intrusive section of the lower unit is composed by an alternation of troctolites with various ultramafic rock-types,

including dunites, lherzolites, olivine-websterites, olivine-gabbros, anorthositic gabbros, gabbros and rare gabbronorites.

The volcanic and subvolcanic sequence of the lower unit can geochemically be subdivided into three groups of rocks: (1)

basalts and basaltic andesites of the lower pillow section showing a clear high-Ti affinity; (2) basaltic andesites of the upper

pillow section with high-Ti affinity, but showing many geochemical differences with respect to the first group; (3) very low-Ti

(boninitic) basaltic and basaltic andesitic lava flows separating the lower and upper pillow sections, and dykes widespread

throughout the Pindos ophiolites.

These different magmatic groups originated from fractional crystallization from different primary magmas, which were

generated, in turn, from partial melting of mantle sources progressively depleted by previous melt extractions. Group 1

volcanics may have derived from partial melting (ca. 20%) of an undepleted lherzolitic source, while group 2 basaltic rocks may

have derived from partial melting (ca. 10%) of a mantle that had previously experienced mid-ocean ridge basalt (MORB)

extraction. Finally, the Group 3 boninites may have derived from partial melting (ca. 12–17%) of a mantle peridotite previously

depleted by primary melt extraction of Groups 1 and 2 primary melts.

In order to explain the coexistence of these geochemically different magma groups, two petrogenetic models formerly

proposed for the Albanian ophiolites are discussed.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Ophiolites; Petrogenesis; MORB; SSZ; Greece; Jurassic

0024-4937/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.lithos.2003.12.002

* Corresponding author. Tel.: +39-532-293749; fax: +39-532-

210161.

E-mail address: [email protected] (E. Saccani).

1. Introduction

Greek ophiolites have long been the subject of

research because they represent important elements

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253230

for a reconstruction of the geodynamic evolution of

the Hellenide orogenic belt. Among these, the Pindos

Massif (Fig. 1) is characterized by one of the best

preserved ophiolitic sequences of Greece, and repre-

sents a key area for studying the genesis and tectonic

evolution of the Mesozoic Tethyan oceanic units

Fig. 1. Simplified tectonic map of the central Dinaride–Albanide–Helleni

Smith (1993), Robertson (1994), Robertson and Shallo (2000) and referen

(Jones and Robertson, 1991). According to Capedri

et al. (1980), in the Pindos Massif, two distinct

ophiolitic suites showing different magmatic affinities

can be recognized: one is represented by high-Ti

ophiolites, interpreted as having formed in a back-

arc tectonic setting, while the other includes low-Ti

de area showing the main tectono-stratigraphic units. Compiled after

ces therein. Box indicates the studied area expanded in Fig. 2.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 231

ophiolites generated in a supra-subduction zone (SSZ)

setting. In addition, the neighbouring Vourinos ophio-

litic complex, located east of the Pindos Massif (Fig.

1), is believed to be continuous with the Pindos

beneath the Meso-Hellenic molasse trough (Jones

and Robertson, 1991; Ross and Zimmerman, 1996),

and displays both low-Ti and very low-Ti magmatic

suites (Beccaluva et al., 1984). Capedri et al. (1980,

1981) and Jones and Robertson (1991) interpreted the

Pindos, Vourinos, and Othrys ophiolites as different

parts of an oceanic crust developed above a SSZ

during the Middle Jurassic.

Recent studies on Albanian ophiolites pointed out

the occurrence of two separate ophiolitic belts: a

western belt characterized by high-Ti ophiolites,

Fig. 2. Simplified geological map of the Pindos Massif. Compi

generated in a mid-ocean ridge (MOR) setting, and

an eastern belt which includes low-Ti and very low-

Ti magmatic sequences, formed in a SSZ setting

(Beccaluva et al., 1994; Shallo, 1994; Bortolotti et

al., 1996; Bebien et al., 1998). In addition, Bortolotti

et al. (1996, 2002); Bebien et al. (2000), and Hoeck

et al. (2002) have pointed out a progressive change

from MOR-type to island arc tholeiitic-type (IAT)

magmatism and the presence of very low-Ti basaltic

dykes in the Albanian western belt ophiolites. Borto-

lotti et al. (1996, 2002) have interpreted the western

belt ophiolites as a MOR-type oceanic crust trapped

in a SSZ setting.

Since the Pindos Massif can be considered the

southern prolongation of the Albanian ophiolitic belts

led after Jones and Robertson (1991), Jones et al. (1991).

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253232

(Robertson and Shallo, 2000), its ophiolitic sequences

should be critically reinvestigated in the light of

results obtained from the Albanian ophiolites, and

its petrological significance re-examined.

The purpose of this paper is thus to present new

petrological and geochemical data on the Pindos ophio-

lites in order to discuss their petrogenetic implications.

2. General geological framework of the Pindos

ophiolites

The Pindos Massif is located in the Subpelagonian

Zone (Fig. 1), and represents the main ophiolitic

outcrop of the Hellenides (2500 km2). Its geological

structure is rather complicated since the ophiolitic

successions are dismembered in a series of west-verg-

Fig. 3. Simplified reconstructed stratigraphy (not to scale) of the

ing, imbricated thrust sheets together with platform

carbonates, pelagic and turbiditic sediments and me-

lange units. These thrust sheets are emplaced over the

autochthonous Maastrichtian–Eocene Pindos Flysch

(Fig. 2). Jones and Robertson (1991) subdivided the

Pindos complex into four tectono-stratigraphic units:

(1) the Jurassic Pindos Ophiolite Group; (2) Late

Triassic–Late Jurassic Avdella Melange, representing

the sub-ophiolitic melange; (3) Late Jurassic–Late

Cretaceous deep-sea sedimentary Dio Dendra Group;

(4) Late Cretaceous shallow-water limestones of the

Orliakas Group. The Pindos Ophiolite Group is the

uppermost tectonic unit, and is overlain by the Eo-

cene–Miocene molasse deposits of the Meso-Hellenic

Trough (Fig. 2). It can be subdivided into three sub-

units (from bottom to top): (i) the Aspropotamos

Complex; (ii) Loumnitsa Unit; and (iii) Dramala Com-

Pindos Ophiolitic Group. Modified after Jones et al. (1991).

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 233

plex. According to Jones and Robertson (1991), the

fragmentation of the Pindos oceanic units occurred

during their Late Jurassic emplacement.

The Aspropotamos Complex is highly dismem-

bered; nonetheless, its ‘‘apparent’’ stratigraphic se-

quence (Fig. 3) includes layered ultramafic and

mafic cumulates, isotropic mafic intrusive, and mafic

extrusive rocks (Montigny et al., 1973). The ultra-

mafic–mafic cumulates are represented by a dunite–

anorthosite–troctolite–gabbro series, a dunite–wehr-

lite–olivine gabbro series, and layered gabbros,

passing upward into isotropic gabbros, and small

plagiogranites bodies.

A sheeted dyke complex marks the transition be-

tween the intrusive and extrusive sequences; the latter

consists of pillow and massive basaltic lavas, basaltic

breccias and hyaloclastites. The mafic rocks of the

Aspropotamos Complex display contrasting geochem-

ical signatures. In fact, Capedri et al. (1980) situated in

the Metsovo area the presence of both mid-ocean ridge

basalts (MORBs) and island arc tholeiitic (IAT) vol-

canic rocks, whereas Kostopoulos (1988) documented

a progressive trend from MORB to IAT, and finally to

boninite in the central sector of Pindos.

The Loumnitsa Unit represents the metamorphic

sole of the Pindos ophiolites (Fig. 3); it consists of

metabasites and metasedimentary rocks ranging from

greenschist to amphibolite facies. Rocks of this Unit

occur at the base of both the Dramala and Aspropo-

tamos Complexes, as well as in variably sized blocks

in the Avdella Melange, in which two main types of

mafic protoliths are found: typical mid-ocean ridge

basalts (MORBs) and within-plate basalts (WPBs)

(Jones and Robertson, 1991). Various datings have

been performed with different methods (Thuizat et al.,

1981; Spray et al., 1984, and references therein); ages

range from 163F 3 to 172F 5 Ma (see these papers

for analytical methods).

The Dramala Complex consists mainly of harzbur-

gitic tectonites, and subordinate dunites, pyroxenites

and ultramafic cumulates (Jones and Robertson, 1991).

3. Tectono-stratigraphic relationships between the

Pindos ophiolitic units

A detailed discussion on tectono-stratigraphic rela-

tionships between the different ophiolitic units of the

Pindos Massif is beyond the scope of the present

study. Nonetheless, for the purpose of this paper, a

critical examination of the ophiolitic pseudostratigra-

phy so far presented in the literature (Kemp and

McCaig, 1984; Jones and Robertson, 1991) should

be made.

In the ophiolitic pseudostratigraphy proposed by

Kemp and McCaig (1984) and Jones and Robertson

(1991), the Dramala Complex is considered the

mantle section of the Pindos ophiolites. However,

this Complex occupies the upper tectonic position

and, together with the metamorphic sole (Loumnitsa

Unit) and Avdella Melange at its base (Fig. 3), is

overthrust onto the Aspropotamos Complex. More-

over, the Dramala Complex, formed by mantle harz-

burgites, probably represents a mantle section

developed in a SSZ setting (Bonatti and Michael,

1989). As a result, the genetic relationships between

the harzburgites of the Dramala Complex and the

magmatic sequence represented in the Aspropotamos

Complex are ambiguous, since both MORB and SSZ

geochemical signatures are found in the mafic rocks

of the Aspropotamos Complex (Capedri et al., 1980;

Kostopoulos, 1988).

In addition, the presence of the Avdella Melange

and metamorphic sole at the base of both the Aspro-

potamos and Dramala Complexes (Fig. 3) suggests

that these units represent major, distinct ophiolitic

thrust sheets.

Most of the tectono-stratigraphic and petrological

features observed in the Pindos ophiolites are com-

monly found in many Subpelagonian ophiolites (Fig.

1), from the Mirdita Zone, in Albania (Bortolotti et al.,

1996; Bebien et al., 2000), to Othrys (Rassios, 1990);

consequently, their significance should be considered

on a regional scale.

For all these reasons, we prefer to regard the

Dramala Complex as the Upper Ophiolitic Unit,

Aspropotamos Complex as the Lower Ophiolitic Unit,

and Loumnitsa Unit as the metamorphic sole (Fig. 3).

4. Sampling and methods

Forty eight representative samples from the Pindos

ophiolites were selected for petrographical and geo-

chemical analysis. Sampling was mainly focused on

the Lower Ophiolitic Unit, though mantle tectonites

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253234

from the Upper Ophiolitic Unit were also collected.

The volcanic Lower Ophiolitic Unit was sampled in

detail along the Aspropotamos River section, where a

complete rock sequence (from ultramafic cumulates to

volcanics) crops out.

Bulk rock major and trace element (Zn, Ni, Co,

Cr, V, Rb, Sr, Ba, Zr, Y) analyzes were performed

on pressed powder pellets using an automated

Philips PW1400 X-ray fluorescence (XRF) spec-

trometer. The matrix correction methods proposed

by Franzini et al. (1972) were applied. Accuracy for

trace elements is better than 10%; detection limits

are 1–2 ppm.

Rare Earth Elements (REE), Sc, Nb, Hf, Ta, Th,

and U were determined by inductively coupled plas-

ma-mass spectrometry (ICP-MS) using a VG Elemen-

tal Plasma Quad PQ2 Plus. Accuracy and detection

limits were calculated by analyzing a set of interna-

tional standards, including JP-1, JGb-1, BHVO-1,

UB-N, BE-N, BR, GSR-3, and AN-G. Detection

limits are (in ppm): Sc = 0.29; Y, Nb, Hf, Ta = 0.02;

REE < 0.014; Th, U = 0.011. Accuracy for analyzed

elements is in the range of 0.9–7.9 relative %, with

the exception of Gd (10.2 relative %). All analyses

were performed at the Department of Earth Sciences

of the University of Ferrara.

Electron microprobe analyses were performed at

the University of Florence using a JEOL-JXA 8600

automated microanalyser. The operative conditions

were sample current of 10 nA and accelerating poten-

tial of 15 kV. Counting time was 100 s per peak and

20 s per background positions.

5. Petrography of the Pindos ophiolites

5.1. Mantle tectonites of the Upper Ophiolitic Unit

Mantle tectonites are represented by harzburgites.

They mainly show porphyroclastic texture with

coarse-sized orthopyroxene crystals, though grano-

blastic textures are locally observed. Olivine is often

characterized by kink banding, and orthopyroxene

frequently bears clinopyroxene exsolution lamellae.

Lobate or subhedral Cr-spinel represents the main

accessory phase. Small modal amounts of diopside

(3–4%) are observed in some samples. The analyzed

harzburgites are fresh to moderately altered; hydrated

associations of serpentine minerals, chlorite, talc and

opaques are the main secondary minerals. Cumulate

rocks associated with mantle tectonites of the Upper

Ophiolitic Unit (Jones and Robertson, 1991) have not

been studied in this paper.

5.2. Cumulate rocks of the Lower Ophiolitic Unit

The lower cumulate sequence of the Lower Ophio-

litic Unit is characterized by an alternation of trocto-

lites with dunites, lherzolites, olivine-websterites and

olivine-gabbros, while the upper part of the cumulitic

sequence is dominated by anorthositic gabbros,

gabbros and rare gabbronorites.

Olivine and plagioclase are the predominant cu-

mulus phases in the melanocratic and leucocratic

rocks, respectively. Cumulus olivine usually forms

large rounded grains in troctolites and ultramafic

cumulates, or smaller anhedral grains in gabbros.

Plagioclase occurs as large euhedral cumulus grains

in gabbros and troctolites, or as small intercumulus

anhedral grains in mela-troctolites. Clinopyroxene is

commonly observed in almost all rock types. In

lherzolites and olivine-websterites, clinopyroxene is

commonly found either as large poikilitic or subhedral

crystals, ranging in volume from about 20% to 50%,

while in troctolites, only small modal amounts (5–

10%) of poikilitic clinopyroxene can be observed. In

olivine-gabbros and gabbros, clinopyroxene shows

subhedral texture.

Accessory phases are represented by spinels occur-

ring as both cumulus and intercumulus grains.

Alteration observed in the analyzed samples ranges

from moderate in gabbros, to severe in ultramafic

types. Olivine is usually serpentinized, plagioclase is

transformed into prehnite and albite, and clinopyrox-

ene is commonly altered into actinolite and chlorite.

The crystallization order is: olivine +Cr-spinel!plagioclase! plagioclase + clinopyroxene! ortho-

pyroxene! Fe–Ti oxides; that is, the typical MORB

sequence (Beccaluva et al., 1979).

5.3. Volcanic sequence of the Lower Ophiolitic Unit

The volcanic sequence of the Lower Ophiolitic

Unit is largely represented by pillowed basalt and

basaltic andesite. Nevertheless, in the Aspropotamos

River section, some massive lava flows are interlay-

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 235

ered in the upper part of the pillow series (Fig. 3).

Texturally, the pillow lavas are aphyric to variably

plagioclase-porphyritic; the groundmass is intersertal

to subophitic, with laths of plagioclase and intersti-

tial clinopyroxene. Quartz-filled amygdules are ob-

served in a few samples from the upper part of the

sequence. Massive lavas are represented by basaltic

andesites; they are moderately porphyritic, with

clinopyroxene and minor plagioclase phenocrysts

settled in an intergranular groundmass with micro-

lites of clinopyroxene and plagioclase. Fe–Ti oxides

are present in both pillow and massive lavas. All

rock types are altered; common alteration products

are albite, prehnite and minor calcite (replacing

plagioclase) and actinolite (replacing clinopyroxene).

Nonetheless, relicts of fresh clinopyroxene are lo-

cally preserved.

The crystallization order in the pillow lavas is: pla-

gioclase! clinopyroxene! Fe–Ti oxides, whereas in

the massive lava flows, it is: Cr-spinel! clinopyro-

xene! clinopyroxene + plagioclase! Fe–Ti oxides.

Textures observed in the various lava types studied

in this paper are similar to those of equivalent rocks

from the Mirdita–Subpelagonian Zone.

5.4. Dykes

Dykes are found in all crustal levels represented in

the Lower Ophiolitic Unit, as well as in the mantle

harzburgites of the Upper Ophiolitic Unit. However,

no textural variations are observed between dykes

cross-cutting the different units. Textures range from

fine to medium grained, aphyric to moderately por-

phyritic, with clinopyroxene and possibly olivine as

phenocrysts set in either intergranular or subophitic

groundmass, in which microlites of clinopyroxene and

plagioclase can be recognized. Accessory phases are

interstitial Fe–Ti oxides and, locally, Cr-spinel micro-

phenocrysts. The dykes are generally fine grained near

the margins, and become medium grained towards the

cores. Dykes are pervasively modified by hydrother-

mal alteration, and are currently composed of albite,

chlorite, quartz, pyrite, clay minerals and calcite.

Clinopyroxene is usually the only phase that remains

partially unaltered.

The observed crystallization order is: olivine +

Cr-spinel! clinopyroxene! clinopyroxene + plagio-

clase! Fe–Ti oxides.

6. Geochemistry of the Pindos ophiolites

The discussion on the geochemical and petroge-

netic features of the studied rocks is mainly based

on those elements which are immobile during meta-

morphic and alteration processes. Previous works

(Pearce and Norry, 1979; Beccaluva et al., 1979)

indicate that the transition metals (V, Cr, Mn, Fe,

Co, Ni, Zn), Mg, Y, and the high field strength

(HFS) elements (Zr, Nb, Ti, Hf, P and REE) are

relatively immobile and largely reflect magmatic

abundances. By contrast, large ion lithophile (LIL)

elements (Ba, Rb, K, and Sr) have generally expe-

rienced metasomatic and hydrothermal mobilization

in most of the samples.

6.1. Mantle harzburgites (Upper Ophiolitic Unit)

Two groups of chemically distinguishable harzbur-

gites can be identified (Table 1), though these do not

entirely correspond to the two groups recognized on

petrographical bases (i.e., cpx-bearing and cpx-free

harzburgites). One group reflects a more refractory

nature, and is characterized by lower TiO2 ( < 0.01%),

Al2O3 (0.24–1.35%), CaO (0.03–1.52%), and V

(28–59 ppm) contents coupled with higher concen-

trations of Co (114–122 ppm) and Cr (2577–4002

ppm). REE are generally very depleted, ranging in

concentration from 0.04 to 0.3 times chondritic values

(Fig. 4B); nonetheless, the U-shaped pattern accounts

for light REE (LREE) enrichment due to subduction-

related fluids. By contrast, the other group displays a

relatively less refractory character testified by the

TiO2 (0.04–0.08%), Al2O3 (1.79–2.86%), CaO

(1.21–2.31%), V (52–70 ppm), Co (96–103 ppm),

and Cr (2398–2677 ppm) concentrations. REE pat-

terns (Fig. 4B) are characterized by a strong LREE

depletion with respect to heavy REE (HREE). The

more refractory nature of the first group with respect

to the second is also testified by the Mg# [defined as

Mg# = 100*Mg/(Mg + Fe2+)], which are in the ranges

91.9–92.5 and 91.4–91.8, respectively.

6.2. Intrusive cumulate sequence (Lower Ophiolitic

Unit)

The ultramafic varieties include dunites, plagio-

clase-lherzolites and olivine-websterites, whose chem-

Table 1

Bulk-rock major and trace element analyses of selected samples from the Pindos ophiolites; (a) XRF analyses, (b) ICP-MS analyses

Upper unit Lower unit

Mantle sequence Intrusive sequence

Sample EP6 EP22 EP17 EP28 EP26 EP33 EP35 EP9 EP29

Rock Hz(1) Hz(2) Du Pl-Lh Ol-Wb M-Troct Troct(3) Gb Gb

(a) SiO2 42.86 41.13 34.02 42.24 47.61 39.69 42.03 51.23 42.93

TiO2 0.01 0.06 0.00 0.03 0.05 0.07 0.05 0.59 0.07

Al2O3 0.35 2.40 0.07 1.68 1.62 10.10 21.59 17.06 18.34

Fe2O3 – – – – – – 0.61 0.82 0.48

FeO 6.78 6.50 5.64 7.22 6.18 5.92 4.08 5.47 3.22

MnO 0.13 0.13 0.12 0.14 0.14 0.12 0.11 0.13 0.11

MgO 44.28 40.50 44.73 36.87 27.96 31.27 17.93 10.41 15.99

CaO 0.56 2.23 0.17 5.25 11.92 5.39 9.30 8.63 11.89

Na2O 0.00 0.00 0.00 0.00 0.00 0.23 0.79 3.73 0.83

K2O 0.00 0.00 0.00 0.00 0.00 0.02 0.01 0.22 0.01

P2O5 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

L.O.I. 5.03 7.04 15.24 6.56 4.53 7.19 3.49 1.69 6.13

Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00

Mg# 92.1 91.7 93.4 90.1 89.0 90.4 88.7 77.2 89.9

Ni 2367 1934 1598 913 460 1420 691 100 313

Co 115 96 106 96 77 81 54 27 38

Cr 2625 2580 3614 2897 1796 1616 754 116 968

V 31 70 15 62 136 43 20 173 73

Rb n.d. n.d. n.d. n.d. n.d. n.d. 2 3 n.d.

Sr n.d. 24 2 4 4 30 74 137 89

Y n.d. n.d. 2 3 3 5 2 21 4

Zr n.d. n.d. n.d. n.d. n.d. 4 2 46 n.d.

Nb n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

Ba n.d. 6 5 6 5 3 10 10 11

Th n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

La n.d. n.d. n.d. n.d. n.d. n.d. n.d. 2 n.d.

Ce n.d. n.d. n.d. 8 10 7 4 14 3

Pb n.d. 7 2 6 4 4 5 7 6

Zn 42 42 36 39 33 44 28 30 21

(b) Sc 5.16 8.64 – – – – – – –

Y 1.32 2.12 – – – – – – –

Nb 0.42 0.06 – – – – – – –

La 0.02 0.03 – – – – – – –

Ce 0.04 0.06 – – – – – – –

Pr 0.01 0.02 – – – – – – –

Nd 0.02 0.17 – – – – – – –

Sm 0.01 0.12 – – – – – – –

Eu n.d 0.05 – – – – – – –

Gd 0.01 0.22 – – – – – – –

Tb n.d 0.06 – – – – – – –

Dy 0.02 0.38 – – – – – – –

Ho 0.01 0.10 – – – – – – –

Er 0.02 0.30 – – – – – – –

Tm n.d 0.05 – – – – – – –

Yb 0.03 0.32 – – – – – – –

Lu 0.01 0.05 – – – – – – –

Hf 0.02 0.09 – – – – – – –

Ta 0.47 0.06 – – – – – – –

Th 0.01 0.01 – – – – – – –

U 0.00 0.00 – – – – – – –

Zr/Y – – – – – 0.8 1.0 1.2 –

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253236

Table 1 (continued )

Lower unit

Lower pillow sequence Massive lavas Upper pillow sequence

Sample EP8 EP41 EP44 EP45 EP46 EP47 EP48 EP49 EP50

Rock HT-B HT-B HT-B HT-B VLT-BA VLT-BA HT-BA HT-BA HT-BA(3)

(a) SiO2 47.72 49.00 48.00 47.50 54.85 54.90 55.34 60.75 64.26

TiO2 0.98 1.15 1.28 1.25 0.39 0.22 1.21 0.92 0.75

Al2O3 17.75 15.18 17.36 16.78 16.28 15.56 14.91 13.77 13.50

Fe2O3 1.11 1.12 1.10 1.15 1.25 1.20 1.42 1.09 0.92

FeO 7.41 7.46 7.32 7.65 8.32 8.02 9.43 7.27 6.11

MnO 0.15 0.15 0.15 0.16 0.16 0.15 0.17 0.15 0.13

MgO 10.41 10.83 9.62 9.57 6.52 7.47 7.59 4.41 4.10

CaO 9.50 8.66 10.33 10.95 6.16 7.39 2.23 3.25 2.27

Na2O 2.72 3.58 3.06 2.87 3.53 2.42 3.72 4.17 6.05

K2O 0.16 0.12 0.19 0.06 0.57 0.19 1.31 1.88 0.09

P2O5 0.11 0.11 0.10 0.09 0.01 0.00 0.10 0.06 0.06

L.O.I. 1.98 2.64 1.48 1.96 1.96 2.48 2.56 2.29 1.77

Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00

Mg# 71.4 72.1 70.1 69.0 58.3 62.4 58.9 51.9 54.5

Ni 127 133 142 140 24 76 12 11 11

Co 41 39 36 39 36 39 30 23 24

Cr 333 372 411 399 47 138 16 19 18

V 205 238 255 247 243 263 314 307 255

Rb n.d. 2 7 5 10 4 21 42 n.d.

Sr 141 174 153 155 125 111 107 139 122

Y 30 31 30 29 16 9 27 24 23

Zr 90 102 109 107 38 17 68 67 54

Nb 2 5 n.d. 2 n.d. n.d. 2 n.d. n.d.

Ba 6 28 16 22 39 30 113 131 18

Th 2 n.d. 3 n.d. n.d. n.d. n.d. n.d. n.d.

La 2 5 6 5 2 n.d. n.d. 2 2

Ce 7 3 4 7 4 n.d. 7 6 5

Pb 3 6 32 31 33 33 27 25 6

Zn 74 72 76 78 69 76 89 79 69

(b) Sc – – – 31.7 23.1 37.9 29.1 29.6 27.3

Y – – – 28.4 15.3 8.1 26.8 21.8 21.0

Nb – – – 3.64 1.04 1.17 2.47 2.35 1.33

La – – – 3.52 1.33 0.88 2.41 2.11 1.99

Ce – – – 9.93 3.14 1.76 6.23 5.85 4.71

Pr – – – 1.65 0.48 0.26 1.05 1.12 0.92

Nd – – – 8.55 2.63 1.10 5.77 5.91 4.85

Sm – – – 2.56 0.92 0.40 1.95 2.02 1.72

Eu – – – 0.88 0.33 0.14 0.70 0.65 0.62

Gd – – – 3.28 1.46 0.70 2.77 2.80 2.48

Tb – – – 0.62 0.30 0.16 0.55 0.56 0.51

Dy – – – 3.98 2.10 1.24 3.67 3.80 3.46

Ho – – – 0.88 0.49 0.33 0.83 0.85 0.80

Er – – – 2.62 1.50 1.05 2.52 2.55 2.39

Tm – – – 0.38 0.23 0.18 0.38 0.40 0.37

Yb – – – 2.64 1.66 1.32 2.65 2.65 2.43

Lu – – – 0.38 0.24 0.21 0.38 0.40 0.37

Hf – – – 2.61 1.01 0.53 2.02 2.36 1.58

Ta – – – 0.25 0.09 0.11 0.17 0.16 0.09

Th – – – 0.23 0.28 0.24 0.27 0.25 0.18

U – – – 0.07 0.15 0.12 0.13 0.12 0.16

Zr/Y 3.0 3.3 3.6 3.7 2.4 1.9 2.5 2.8 2.3

(continued on next page)

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 237

Table 1 (continued )

Lower unit

Dykes

Sample EP30 EP31 EP32 EP39 EP43 EP54 EP36 EP38 EP42 EP40

Rock VLT-B VLT-B VLT-B VLT-B VLT-B VLT-B VLT-BA VLT-BA VLT-BA VLT-BA

(a) SiO2 44.89 47.71 50.88 52.86 50.08 50.66 56.90 56.04 53.65 55.05

TiO2 0.18 0.15 0.15 0.16 0.16 0.14 0.26 0.28 0.19 0.27

Al2O3 10.90 10.50 10.29 10.06 13.99 11.90 13.93 14.75 15.08 16.09

Fe2O3 0.97 1.15 1.10 1.06 1.08 1.13 1.19 1.34 1.15 1.16

FeO 6.49 7.66 7.32 7.09 7.21 7.56 7.94 8.91 7.64 7.73

MnO 0.15 0.16 0.16 0.15 0.16 0.15 0.15 0.16 0.15 0.15

MgO 19.23 17.08 18.69 16.94 14.35 16.24 8.15 7.23 7.55 7.31

CaO 11.77 11.52 8.12 8.89 7.24 6.86 7.44 8.24 9.68 5.67

Na2O 0.12 0.65 0.68 0.86 1.99 2.41 2.15 1.75 2.74 4.30

K2O 0.02 0.02 0.08 0.03 0.10 0.20 0.05 0.04 0.01 0.23

P2O5 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

L.O.I. 5.27 3.41 2.53 1.89 3.64 2.75 1.84 1.27 2.16 2.03

Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00

Mg# 84.1 79.9 82.0 81.0 78.0 79.3 64.7 59.1 63.8 62.8

Ni 223 290 310 228 187 268 62 42 56 52

Co 45 54 52 49 42 51 38 37 29 35

Cr 849 1185 1348 1082 658 1105 272 72 74 111

V 202 217 231 212 264 233 311 299 290 249

Rb n.d. n.d. 2 n.d. n.d. 3 n.d. n.d. n.d. 2

Sr 9 85 29 41 82 52 85 88 126 107

Y 8 8 6 7 8 7 12 14 7 11

Zr 12 10 11 11 12 10 19 23 14 22

Nb n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

Ba 8 38 17 8 22 21 22 10 9 13

Th n.d. n.d. 2 n.d. n.d. n.d. n.d. n.d. n.d. n.d.

La 3 4 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

Ce 3 n.d. 2 n.d. n.d. n.d. 3 2 2 3

Pb n.d. 3 5 3 5 7 7 2 5 6

Zn 50 50 61 64 77 54 59 64 65 71

(b) Sc – – 24.7 31.6 47.5 27.8 49.3 34.9 42.4 28.3

Y – – 4.19 4.94 6.68 5.35 11.1 10.3 6.12 8.71

Nb – – 0.70 0.42 0.94 0.76 0.90 0.95 1.00 1.03

La – – 0.47 0.40 0.59 0.48 0.87 0.80 0.66 1.16

Ce – – 1.08 0.91 1.34 1.10 2.03 1.83 1.48 2.68

Pr – – 0.14 0.13 0.17 0.13 0.28 0.25 0.19 0.40

Nd – – 0.68 0.60 0.73 0.61 1.38 1.29 0.84 2.05

Sm – – 0.24 0.23 0.27 0.20 0.54 0.47 0.30 0.80

Eu – – 0.09 0.09 0.14 0.07 0.15 0.15 0.15 0.41

Gd – – 0.36 0.35 0.43 0.34 0.80 0.75 0.48 1.11

Tb – – 0.08 0.08 0.10 0.08 0.18 0.17 0.11 0.29

Dy – – 0.58 0.59 0.77 0.61 1.33 1.26 0.79 1.91

Ho – – 0.14 0.15 0.21 0.16 0.34 0.32 0.20 0.51

Er – – 0.48 0.49 0.67 0.55 1.09 1.03 0.66 1.67

Tm – – 0.07 0.08 0.12 0.09 0.18 0.17 0.11 0.26

Yb – – 0.60 0.62 0.91 0.77 1.36 1.28 0.87 1.81

Lu – – 0.08 0.09 0.16 0.11 0.22 0.19 0.14 0.31

Hf – – 0.26 0.27 0.52 0.27 0.98 0.57 0.63 0.96

Ta – – 0.17 0.05 0.10 0.08 0.08 0.07 0.16 0.25

Th – – 0.14 0.13 0.20 0.17 0.29 0.33 0.19 0.43

U – – 0.07 0.07 0.09 0.07 0.13 0.14 0.09 0.33

Zr/Y 1.5 1.3 1.8 1.6 1.5 1.4 1.6 1.6 2.0 2.0

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253238

Fig. 4. Rock/C1 chondrite (Sun, 1982) transition metal com-

positions (A) and rock/chondrite (Sun and McDonough, 1989) REE

compositions (B) for mantle harzburgites from the Upper Ophiolitic

Unit.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 239

ical compositions (Table 1) are clearly related to their

modal assemblages.

Similarly, the mafic rocks (mela-troctolites, trocto-

lites, and gabbros) cover a wide compositional range

(Table 1), depending on the highly variable modal

composition, which is due to the cumulitic nature of

these rocks. In general, the incompatible elements

(e.g., Zr, Ti, Y, and P) are very low because they do

not partition in the common cumulate phases. By

contrast, compatible elements (e.g., Ni and Cr) show

relatively high concentrations. In particular, Ni ranges

from 691 to 1455 ppm in troctolites and from 100 to

706 ppm in gabbros, while Cr ranges from 538 to

1616 ppm in troctolites and from 116 to 2492 ppm in

gabbros. According to Serri (1981), the high-Ti mag-

Notes to Table 1:

Hz = harzburgite; Du = dunite; Pl-Lh = plagioclase-lherzolite; Ol-Wb= olivi

bro; HT- = high-Ti; VL-T= very low-Ti; B = basalt; BA= basaltic andesit

(Mg+ Fe)*100; n.d. = not detected.

matic affinity of the gabbroic rocks can clearly be

deduced on the basis of the TiO2 contents and FeO/

(FeO +MgO) ratios (Table 1) only for few samples.

By contrast, most samples cannot be clearly attributed

either to a MOR or SSZ geochemical affinity on the

basis of their chemical composition (Table 1); none-

theless, the crystallization order described above,

unequivocally testifies their MOR affinity.

6.3. Volcanic sequence and dykes (Lower Ophiolitic

Unit)

On the bases of the chemical analyses, three main

geochemical groups of samples from the volcanic

sequence of the Aspropotamos River section can be

identified: (1) basalts and basaltic andesites from the

lower part of the pillow sequence, beneath the mas-

sive lava flows; (2) basaltic andesites from the upper

part of the pillow sequence, that is, above the massive

lava flows; (3) basaltic and basaltic andesitic dykes

and massive lava flows. These groups can be readily

identified in the V vs. Ti discrimination diagram of

Fig. 5 (Shervais, 1982), as well as in the variation

diagrams of Figs. 6 and 7, where selected major and

trace elements are plotted with respect to the Zr

content.

In Fig. 5 basaltic rocks from the lower pillow

sequence display Ti/V ratios typical of MORBs (i.e.,

included between 20 and 50), while the upper pillow

sequence compositions lie across the boundary be-

tween IAT and MORB compositions (i.e., Ti/V = 20).

Dykes and massive lava flows exhibit Ti/V ratios

comparable with those of boninitic rocks, that is, lower

than 10.

Zr ranges from 90 to 109 ppm in the lower pillow

sequence lavas, from 54 to 68 ppm in the upper pillow

sequence lavas, from 10 to 12 ppm in the more pri-

mitive boninitic basalts, and from 22 to 38 ppm in the

more evolved boninitic lavas, whereasMg# variation is

72–69, 58–51, 84–78, and 64–58, respectively.

The analyzed basalts from the lower pillow se-

quence are characterized by relatively uniform com-

positions (Table 1), and they display geochemical

ne-websterite; M-Troct =mela troctolite; Troct = troctolite; Gb = gab-

e; (1) =Cpx-free; (2) = Cpx-bearing; (3) =Dyke; Mg#=molar Mg/

Fig. 5. V vs. Ti/1000 discrimination diagram for mafic volcanic

rocks and dykes from the Pindos Lower Ophiolitic Unit. Modified

after Shervais (1982). Fields for MORB and IAT compositions are

also shown.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253240

features that are compatible with those of high-Ti

basalts generated at mid-ocean ridge (Beccaluva et

al., 1983). In particular, normalized high field strength

elements (HFSE) concentrations (Fig. 8A) exhibit flat

patterns at about 1 time N-MORB contents (Pearce,

1983). These basalts also show rather flat REE pat-

terns and a mild LREE depletion (LaN/SmN = 0.89),

features that are consistent with N-MORB composi-

tions (Fig. 8C).

The analyzed samples from the upper pillow se-

quence display many geochemical differences when

compared to the underlying basalts of the lower pillow

sequence. They are represented by basaltic andesites

with Mg# ranging from 58.9 to 51.9 and high SiO2

content (Table 1), which is, however, variably influ-

enced by the presence of quartz-filled amygdules.

According to their Mg#, these rocks have very low

Ni and Cr contents (11–12 and 16–19 ppm, respec-

tively), moderate to low MgO (4.1–7.59%), and high

FeOtot�/MgO ratio (1.4–1.9) in comparison with the

lower pillow sequence basalts. Nevertheless, in spite

of the more evolved nature of these rocks, also HFSE

concentrations are slightly lower (Fig. 8B), and display

rather flat patterns at 0.5–1 times N-MORB contents

with marked Cr negative anomalies (Fig. 8B). Basaltic

rocks from the upper pillow sequence show HREE

concentrations similar to those of basalts from the

lower pillow sequence, but differ from these by their

lower contents of light to medium REE (Fig. 8C) and

LREE depletion (LaN/SmN = 0.68–0.80). Nonethe-

less, the overall REE patterns presented by these

basalts are consistent with N-MORB compositions.

The dykes and massive lava flows are represented

by basalts and basaltic andesites showing rather var-

iable compositions (Table 1). Basically, two groups of

samples can be identified: more primitive basalts, and

more evolved basalts and basaltic andesites. These

groups mainly differ in their MgO, Cr and Zr contents,

which are in the range 14.35–19.23%, 658–1348

ppm, 10–12 ppm, respectively, for the first group,

and 6.52–8.15 %, 47–272 ppm, 14–38 ppm for more

evolved rocks (Table 1, Figs. 6 and 7).

However, the most striking characteristic of the

basaltic rocks in dikes and massive lava flows is their

very low content of incompatible elements (e.g., TiO2

0.14–0.39%, P2O5 < 0.01%, Y 7–16 ppm), associat-

ed with low Al2O3 (10.06–16.28%). In Fig. 9A,B, a

sharp chemical distinction between more primitive

basalts and more evolved basalts and basaltic ande-

sites can be observed: the former are characterized by

lower abundance of HFSE (0.08–0.2 times N-

MORB) and high Cr contents (3–6 times N-MORB),

while the latter display low Cr contents ( < 1 time N-

MORB) and HFSE between 0.1 and 0.6 times N-

MORB. REE display the U-shaped patterns typical of

boninites (Fig. 9C,D) ranging in concentration, for

medium-REE, between one and seven times chondrit-

ic abundance.

These features suggest close similarities with very

low-Ti (boninitic) lavas found in the forearc regions of

oceanic island arcs (Crawford et al., 1989; Falloon

and Crawford, 1991), as well as in many ophiolitic

complexes (Beccaluva and Serri, 1988; Bedard,

1999).

7. Clinopyroxene chemistry

Clinopyroxene chemistry, although controlled by

crystal chemical constraint, is strongly influenced by

the composition of magmas from which they crystal-

lize. It is thus widely accepted that clinopyroxene

Fig. 6. Variation of selected major elements vs. Zr for the Pindos Lower Ophiolitic Unit mafic ophiolitic rocks.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 241

compositions represent a suitable indicator of the

magmatic affinity of basalts from different tectonic

settings (Leterrier et al., 1982) and from different

ophiolitic types (Beccaluva et al., 1989). For this

reason, only clinopyroxenes from mafic volcanic

and subvolcanic rocks were analyzed and presented

in this paper.

Primary pyroxene crystals are scarce in volcanic

rocks, as they are usually replaced by actinolite during

ocean-floor metamorphism. Selected chemical data on

preserved crystals are presented in Table 2.

Clinopyroxenes from the high-Ti basalts of the

lower pillow sequence show compositions ranging

from diopside to augite (Fig. 10A). They are chemi-

cally distinct from clinopyroxenes of very low-Ti rocks

in their higher TiO2 (1.02–1.23 wt.%) and Al2O3

(2.28–4.71 wt.%), as well as in their lower SiO2

content (50.37–51.69 wt.%). In these rocks, clinopyr-

Fig. 7. Variation of selected trace elements vs. Zr for the Pindos Lower Ophiolitic Unit mafic ophiolitic rocks.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253242

oxenes are characterized by a marked Fe-enrichment

trend, as testified by the very high FeO content in

crystal rims (Table 2).

Clinopyroxenes from the very low-Ti basaltic

rocks of the massive lava flows and dykes, including

phenocrysts, microphenocrysts and groundmass

microlites, display quite uniform augitic compositions

(Fig. 10A); they are characterized by very low Ti

abundance (0.06–0.13 wt.%) and by considerably

high Mg# values (79.7–89.8). The Cr2O3 content

does not exhibit any correlation with Mg# and,

although generally high, is extremely variable (0.07

to 0.64 wt.%).

Clinopyroxenes have been plotted in the discrim-

ination diagram of Fig. 10B (Beccaluva et al., 1989).

In this diagram, minerals from the lower pillow

sequence reveal a close affinity with clinopyroxenes

of normal-MORBs, while minerals from both massive

Fig. 8. N-MORB normalized (Pearce, 1983) incompatible element

compositions for the lower pillow sequence (A) and upper pillow

sequence (B), and chondrite-normalized (Sun and McDonough,

1989) REE compositions (C) for both pillow series.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 243

lavas and dykes display chemical compositions com-

parable with clinopyroxenes from boninitic basalts.

8. Petrogenesis

8.1. Mantle sources

From the overall geochemical characteristics, it

appears that the basaltic rocks studied in this paper

represent different magmatic groups that may have

been originated from different mantle sources and/or

from various degrees of partial melting. The origin of

the Pindos ophiolites from different mantle source

types has also been suggested by Economou-Eliopou-

los and Vacondios (1995) on the basis of chemical

composition of chromitites included in the ultramafic

rocks of the Upper Ophiolitic Unit. According to

Pearce and Norry (1979), the Zr/Y ratios of the pillow

sequence are consistent with a genesis from primary

or slightly enriched mantle sources, whereas Zr/Y

ratios of massive lavas and dykes are consistent with

a genesis from depleted mantle source.

Fig. 11 shows the Cr vs. Y plot for the analyzed

samples. These elements have been chosen in the

attempt to depict the degrees of partial melting of

the different lava types studied in this paper, because

they are considered not to be significantly affected

either by processes causing mantle heterogeneity, or

by partial melting (Pearce, 1983). In this diagram,

three possible mantle sources, in accordance with the

model of incremental batch melting starting from a

single source proposed by Murton (1989), are con-

sidered for the genesis of MOR and boninitic rocks.

The lower pillow sequence is compatible with about

20% partial melting of a MORB source (M1), calcu-

lated according to Pearce (1983). The upper pillow

sequence is only represented by fairly fractionated

rocks; however, assuming that this sequence originat-

ed from a mantle source similar to that of the lower

pillow sequence, the possible trend of fractional

crystallization intersects the melting path at about 40

%, which is a rather excessive degree of partial

melting. Alternatively, the upper pillow sequence

may have derived from about 10% partial melting

from a mantle source that experienced previous

MORB melt extraction at about 20% (M2).

Likewise, assuming that the boninitic dykes and

massive lavas originated from a source that had previ-

ously experienced only MORB extraction (i.e., source

M2), this would require approximately 20–40% melt-

ing. Such high melting rates can result from melting in

an exceptionally hot thermal regime in a shallow

forearc mantle wedge, and/or beneath young intra-

oceanic subduction-related systems (Tatsumi and

Eggins, 1995). The requirement of very high thermal

conditions has been suggested by many authors (Dun-

can andGreen, 1980; Beccaluva et al., 1983; Beccaluva

Fig. 9. N-MORB normalized (Pearce, 1983) incompatible element compositions (A, B) and Chondrite-normalized (Sun and McDonough, 1989) REE compositions (C, D) for

boninitic dykes and massive lava flows (*). Compositional fields for typical boninites from various localities (Beccaluva and Serri, 1988) are also shown for comparison.

E.Sacca

ni,A.Photia

des

/Lith

os73(2004)229–253

244

Table 2

Representative microprobe analyses of clinopyroxenes in basalts from the lower unit of the Pindos ophiolites

Lower pillow sequence Massive lava flows Dykes

Sample EP8 EP47 EP42 EP39

Rock High-Ti basalt Very low-Ti basaltic andesite Very low-Ti basalt Very low-Ti basalt

Mineral Cpx1-1c Cpx1-1r Cpx4-1 Cpx1-1 Cpx4-1c Cpx6-1c Cpx2-1r Cpx3-1c Cpx3-2c Cpx3-1c Cpx3-1r Cpx4-1c

SiO2 50.94 50.37 51.69 53.96 52.66 54.05 53.96 53.27 53.49 53.77 53.49 53.59

TiO2 1.23 1.13 1.02 0.07 0.06 0.10 0.07 0.10 0.08 0.13 0.07 0.06

Al2O3 4.71 2.28 3.43 1.87 1.93 1.63 1.29 1.77 1.61 1.78 2.18 1.70

Cr2O3 0.64 0.02 0.13 0.63 0.89 0.07 0.55 0.25 0.64 0.48 0.47 0.23

Fe2O3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

FeO 6.15 15.41 7.67 5.43 5.80 8.17 4.48 6.01 5.14 6.42 5.65 5.81

MnO 0.14 0.46 0.21 0.16 0.21 0.28 0.16 0.12 0.13 0.16 0.15 0.12

MgO 15.74 11.57 16.07 18.22 17.04 17.13 18.81 17.61 18.00 17.58 17.79 18.05

CaO 20.45 18.47 20.08 20.32 20.77 19.92 20.71 20.28 20.70 20.47 21.16 20.45

Na2O 0.40 0.52 0.41 0.07 0.08 0.13 0.12 0.10 0.12 0.12 0.10 0.09

Oxide total 100.40 100.23 100.71 100.73 99.44 101.48 100.15 99.51 99.91 100.91 101.06 100.10

Fe2O3* 0.42 1.47 1.58 0.00 0.38 0.43 0.74 0.19 0.44 0.31 1.17 0.65

FeO* 5.77 14.09 6.25 5.43 5.46 7.78 3.81 5.84 4.75 6.14 4.59 5.23

Total* 100.44 100.38 100.87 100.73 99.48 101.52 100.22 99.53 99.95 100.94 101.18 100.16

Si 1.864 1.914 1.889 1.953 1.941 1.960 1.957 1.956 1.952 1.951 1.933 1.953

Ti 0.034 0.032 0.028 0.002 0.002 0.003 0.002 0.003 0.002 0.004 0.002 0.002

Al 0.203 0.102 0.148 0.080 0.084 0.070 0.055 0.077 0.069 0.076 0.093 0.073

Cr 0.019 0.001 0.004 0.018 0.026 0.002 0.016 0.007 0.018 0.014 0.013 0.007

Fe3+ 0.012 0.042 0.044 0.000 0.010 0.012 0.020 0.005 0.012 0.009 0.032 0.018

Fe2+ 0.177 0.448 0.191 0.164 0.168 0.236 0.116 0.179 0.145 0.186 0.139 0.159

Mn 0.004 0.015 0.007 0.005 0.007 0.009 0.005 0.004 0.004 0.005 0.005 0.004

Mg 0.858 0.655 0.875 0.983 0.936 0.926 1.017 0.964 0.979 0.951 0.958 0.980

Ca 0.802 0.752 0.786 0.788 0.820 0.774 0.805 0.798 0.809 0.796 0.819 0.798

Na 0.028 0.038 0.029 0.005 0.006 0.009 0.008 0.007 0.008 0.008 0.007 0.006

Cation total 4.000 4.000 4.000 3.998 4.000 4.000 4.000 4.000 4.000 4.000 4.000 4.000

Mg# 82.9 59.4 82.1 85.7 84.8 79.7 89.8 84.3 87.1 83.6 87.3 86.0

Wollastonite 43.7 40.5 42.5 40.7 42.6 40.0 41.6 41.1 41.9 41.2 42.8 41.2

Enstatite 46.7 35.3 47.3 50.8 48.6 47.8 52.5 49.7 50.6 49.2 50.0 50.6

Ferrosilite 9.6 24.1 10.3 8.5 8.8 12.2 6.0 9.2 7.5 9.6 7.2 8.2

c = crystal core; r = crystal rim.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 245

and Serri, 1988) to explain the genesis of similarly

depleted melts through a multiple-stage melting model.

Alternatively, using a source (M3) calculated by

Murton (1989) as the residue after about 12% melt

extraction from M2, the melting curve intersects with

the compositions of dykes and massive lava flows at

partial melting degrees ranging from 10% to 20%,

which are compatible with reasonably lower thermal

conditions.

In summary, the petrologically more realistic sce-

nario for the genesis of the different primary lava

groups in the Pindos ophiolites is in agreement with

an incremental batch melting model in which lower

pillow sequence MORBs were generated by 20%

partial melting of an undepleted lherzolitic mantle

source leaving, as a residue, depleted lherzolitic to

harzburgitic mantle compositions. These, in turn, can

be considered as the probable mantle source for the

upper pillow sequence MORBs (for ca. 10% partial

melting), and possibly for some boninitic lavas (>20%

partial melting). The consequent melt extraction left

as a residue moderately depleted harzburgitic compo-

sitions, which can be considered the source for the

more Y and incompatible elements depleted boninites

corresponding to 12–17% partial melting. The final

boninitic-type melt extraction left as a residue, ex-

tremely depleted harzburgitic to dunitic compositions.

Although there is no evidence that mantle harzbur-

gites of the Upper Ophiolitic Unit are either the source

or the residua of the various melts in the Lower

Fig. 10. (A) Groundmass and phenocryst clinopyroxene compositions, expressed in terms of wollastonite–enstatite– ferrosilite, and (B) TiO2–

Na2O–SiO2/100 (wt.%) discrimination diagram (Beccaluva et al., 1989) for clinopyroxenes in basaltic rocks from the Pindos Lower Unit of

ophiolites. Fields representing clinopyroxene compositions in basalts from Modern oceanic settings are reported for comparison. Abbreviations:

LPS= lower pillo sequence; MLF=massive lava flow; NM=normal MORB; EM= enriched MORB; WOPB=within-plate oceanic basalts;

ICB= Iceland basalts; IAT= island-arc tholeiites; BON=boninites; BA-A= intraoceanic forearc basalts and basaltic andesites.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253246

ophiolitic Unit, from a petrological point of view, this

model is also supported by the composition of mantle

harzburgites studied in this paper. In fact, in Fig. 11,

less refractory cpx-bearing harzburgites plot close the

theoretical M2 source and can be considered as the

residua after MORB extraction, whereas, more refrac-

tory, cpx-free harzburgites plot close the theoretical

M3 source and can be considered as the residua after

the extraction of primary melts of the Upper Pillow

Series. Nonetheless, the incompatible element com-

position of the Upper Pillow Series basalts and

boninites indicates that their relative mantle sources

underwent LILE and LREE enrichment by SSZ fluids.

This enrichment is modest for the Upper Pillow series

basalts (Fig. 8) and for their postulated mantle source

(sample EP22 in Fig. 4), and is more evident in

boninitic rocks (Fig. 9) and their inferred mantle

source (sample EP6 in Fig. 4).

8.2. Fractional crystallization

The good correlation, within each single group,

between Zr and many major and trace elements (Figs.

6 and 7) indicates fractional crystallization as the main

evolutionary process for all the studied magmatic

groups. This is also suggested by the Cr vs. Y diagram

(Fig. 11), where all samples display Cr decreasing

trends with increasing Y, which are compatible with

the fractional crystallization paths (Pearce, 1983).

However, the different variations of the selected

elements in the three recognized lava types presented

in Figs. 6 and 7 account for different fractional

crystallization processes.

According to the petrographical observations, sam-

ples from both lower and upper pillow sequences are

considered to have experienced early fractional crys-

tallization of olivine and plagioclase, as also con-

firmed by the nature of the cumulate sequence. This

is testified by the positive correlation between Zr and

CaO/Al2O3 (Fig. 6)—which indicates the early crys-

tallization of plagioclase that increases the CaO/Al2O3

ratio—as well as by the decrease of MgO with

increasing Zr. The marked increase of TiO2, FeOtot,

CaO, V, and Y reflects the late crystallization of

clinopyroxene and Fe–Ti oxides. It is commonly

accepted that, during crystallization, Ni and Cr are

mainly distributed within the early mafic minerals

(i.e., olivine, Cr-spinel and clinopyroxene). One of

the major differences between the lower and upper

parts of the pillow sequence is the lesser Cr and Ni

content displayed by the upper pillow sequence with

respect to the lower pillow sequence (Figs. 7 and 8 and

Fig. 11. Cr vs. Y diagram for the Pindos Lower Ophiolitic Unit

mafic volcanic rocks (modified after Pearce, 1983). Mantle source

compositions and melting paths for incremental batch melting are

from Murton (1989). Figures indicate the percentage of melting.

M1: calculated MORB source; M2: residue after 20% MORB melt

extraction; M3: residue after 12% melt extraction from M2. See text

for explanation.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 247

Table 1). This aspect can be accounted for either by a

lower abundance of these elements in the possible

mantle source, or by early fractionation of Cr- and Ni-

bearing phases. However, the former case is in con-

trast with the more refractory nature of the mantle

source postulated for the upper pillow sequence in a

previous section, because Cr abundance is not con-

sidered to be significantly modified during the pro-

gressive mantle source depletion (Pearce, 1983). The

early fractionation of Cr and Ni-bearing phases is thus

the simplest process for justifying the low Cr and Ni

abundance in the upper pillow sequence.

By contrast, the sharp decrease of MgO, CaO/

Al2O3, Ni, Co, Cr, and Sc, coupled with the increase

of SiO2 with increasing Zr (Figs. 6 and 7) observed in

very low-Ti dykes and massive lavas, is in accordance

with the early crystallization of olivine and Cr-spinels

followed by clinopyroxene and, later, by plagioclase. In

particular, the negative correlation between Zr and the

CaO/Al2O3 ratio (Fig. 6) and Sc (Fig. 7) testifies for the

early crystallization of clinopyroxene. This conclusion

confirms the petrographical observations in basalts and

basaltic andesites, where clinopyroxene is abundant

and is the sole constituent of the phenocryst assemb-

lages. This fractionation was more substantial in the

samples with very low Zr contents (10–12 ppm), that

is, in the less evolved rocks, which are also character-

ized by very highMg# (about 78 to 84). In addition, the

sharp decrease of Cr suggests that Cr-spinel and clino-

pyroxene played an important role during fractional

crystallization. The crystallization of mafic phases can

be depicted in the Cr vs. Ni diagram of Fig. 12. The

negative correlation between Cr and Ni with respect to

Zr (Fig. 7) indicates that these elements are similarly

incorporated in different minerals during fractional

crystallization. However, the different correlation be-

tween Cr and Ni shown in Fig. 12 indicates that Ni and

Cr are not hosted in the same phases; in particular, Ni is

preferably partitioned in olivine (now completely al-

tered), while Cr is hosted in Cr-spinel and clinopy-

roxene. From Fig. 12, it can be observed that the

fractionation trend for the more primitive basalts is in

accordance with the crystallization of olivine + Cr-

spinel + clinopyroxene, while the more evolved basalts

and basaltic andesites are dominated by the crystalli-

zation of olivine and clinopyroxene.

The different crystallization orders in the MORB

pillow sequence and boninite varieties are also illus-

trated in the Ti vs. Zr diagram (Fig. 13). This diagram

documents the scarce influence of clinopyroxene

fractional crystallization in both lower and upper

pillow sequences. By contrast, the very low-Ti vari-

eties identify a trend subparallel to the clinopyroxene

vector, pointing out the main influence of this phase.

9. Tectonic setting for the Pindos ophiolite genesis

The data presented here indicate that the Pindos

ophiolites include mantle (Upper Ophiolitic Unit) and

crustal (Lower Ophiolitic Unit) sections whose geo-

chemical characteristics can be referred to apparently

contrasting tectonic settings. In particular, the Upper

Ophiolitic Unit is mainly represented by mantle harz-

Fig. 12. Ni vs. Cr diagram for the Pindos Lower Ophiolitic Unit mafic volcanic rocks. Fractional crystallization trends for olivine +Cr-

spinel + clinopyroxene and olivine + clinopyroxene are shown.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253248

burgites possibly representing a SSZ mantle portion.

The intrusive and lower extrusive rocks of the Lower

Ophiolitic Unit display high-Ti geochemical affinity,

Fig. 13. Ti vs. Zr diagram for the Pindos Lower Ophiolitic Unit mafic vo

shown. Abbreviations: opx = orthopyroxene, plag = plagioclase, ol = olivin

and represent a crustal portion generated in a MOR

spreading centre. By contrast, the massive lava flows

interlayered in the upper part of the volcanic sequence,

lcanic rock. Fractional crystallization trends for mineral phases are

e, cpx = clinopyroxene; mgt =magnetite.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 249

and the dykes cross-cutting the whole sequence show

very low-Ti affinities, and are interpreted as generated

in a SSZ setting. Finally, basaltic rocks showing

depleted MORB geochemical characteristics top the

volcanic sequence. According to Kostopoulos (1988),

the timing of these magmatic events indicates a general

progressive evolution from MORB to SSZ magma-

tism. Nevertheless, as observed in the Albanian ophio-

lites (Bortolotti et al., 1996, 2002; Bebien et al., 2000;

Hoeck et al., 2002), the partial interlayering between

MOR and SSZ basaltic rocks in the Pindos ophiolites

implies the contemporaneous presence of two different

types of magmatism.

A variety of possible timings of boninitic eruptions

relative to associated MORBs and/or IATs has been

described in the literature. In some cases, boninite lavas

overlie IATs, as in the Mariana forearc (Hickey and

Frey, 1982; Crawford et al., 1989), while in the Karmoy

ophiolite complex (Norway), boninites overlie both

IATs and MORBs (Pedersen and Hertogen, 1990). By

contrast, in Guam, boninites are topped by island arc

tholeiites (Reagan and Meijer, 1984; Crawford et al.,

1989). In other cases, boninite magmatism occurred

both contemporaneously and after the emplacement of

island arc tholeiites (Beccaluva and Serri, 1988). A

number of cases in which MORB and IAT volcanic

rocks are interlayered has been described, as in the

Zambales (Yumul, 1996), Halmahera (Ballantyne,

1992), and Taitao ophiolites (Klein and Kastens, 1995).

These different timings usually reflect different

histories of island arc evolution. Typically, the evolu-

tion from IAT and boninite to MORB compositions is

associated with incipient opening of back-arc basins,

whereas the evolution from MORB to IAT and bonin-

ite is commonly related to the early stages of island-

arc formation (proto-forearc), and generally reflects

progressive melting of a mantle source becoming

increasingly depleted.

It has been suggested that the MORB–boninite

association in the Pindos ophiolites is related to the

early stages of a subduction (Jones et al., 1991;

Kostopoulos, 1988), or to subduction which devel-

oped following an asymmetric ridge collapse (Clift

and Dixon, 1998), or else to the transition from island

arc to back-arc settings (Capedri et al., 1980).

Nonetheless, the range of compositions for the

Pindos lavas (and associated mantle sources), as well

as their stratigraphical relationships, are compatible

with a tectonic model involving incremental batch

melting of progressively depleted and variably SSZ

LILE- and LREE-enriched mantle sources, which

were necessarily close to each other.

The occurrence of stratigraphical associations of

MOR- and SSZ-type rocks previously documented in

volcanic sequences from the Albanian ophiolites

(Bortolotti et al., 1996, 2002; Bebien et al., 2000;

Hoeck et al., 2002) indicates that undepleted MOR-

type and variably depleted (and variably further

enriched) SSZ-type mantle sources were contempora-

neously active in a relatively restricted sector of the

supra-subduction region at a regional scale, from

north Albania to Greece.

In accordance with the two-dimensional geophys-

ical model proposed by Bebien et al. (2000) and

Insergueix-Filippi et al. (2000), the stratigraphical

association of MORB and SSZ basalts in Albanian

ophiolites have been interpreted by Bortolotti et al.

(2002) and Hoeck et al. (2002) as having generated

from the partial melting of both undepleted and

variably depleted peridotites in a supra-subduction

mantle wedge, which was locally thermally perturbed

by a nearby active mid-ocean ridge. This interpreta-

tion can be effectively applied to the Pindos ophio-

lites, as follows.

During the Early Jurassic, the Pindos oceanic

basin was characterized by the development of

MOR-type oceanic crust. This early oceanic accretion

stage, as well as the former Triassic continental

rifting, is most likely recorded respectively by MORB

and WPB blocks preserved in the Avdella Melange,

rather than by the MOR-type sequences of the Lower

Ophiolitic Unit.

During the Middle Jurassic, an east-dipping sub-

duction was likely initiated near the spreading ridge.

At this stage, MOR-type magmas were still generated

for a short time after initiation of the subduction from

an undepleted MOR source. Soon after, the boninitic

massive lava flows and part of the dykes were

extruded almost contemporaneously with the depleted

MORBs forming the upper pillow series of the

Aspropotamos River section. The melt extraction,

which produced MOR magmatism left behind a more

depleted source, so that only increasingly refractory

melts could be further generated. According to the

model of incremental batch melting of progressively

depleted mantle sources described in a previous sec-

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253250

tion, the upper pillow series was thus generated by

partial melting of localized mantle portions that had

previously experienced moderate MORB melt extrac-

tion without significant enrichment in SSZ-derived

incompatible elements, whereas boninitic lavas were

generated through partial melting of depleted mantle

portions subsequently enriched by SSZ-derived fluids,

leaving more depleted harzburgites, similar in com-

position to those of the Upper Ophiolitic Unit, as a

refractory mantle residue. Insergueix-Filippi et al.

(2000) concluded that the initiation of subduction

near an active ridge axis produces an upwelling of

the asthenospheric flow in the same direction as the

subducting lithosphere. Consequently, melting of a

mantle peridotite, which previously experienced

MORB extraction, may occur as a consequence of

decompression of the ascending lithosphere, without

significant SSZ chemical imprinting. The melting of

localized mantle portions characterized by different

degrees of depletion/enrichment can simultaneously

generate a range of melt compositions, which produce

a spectrum of geochemically different parental liquids.

According to the model presented above, the Lower

Ophiolitic Unit likely generated from the latest stages

of mid-ocean ridge activity to the early stages of

immature arc development.

An alternative model for explaining the partial

melting of depleted sub-arc mantle peridotites, as well

as the MORB-SSZ lava alternations in Albanian

ophiolites, has recently been proposed by Beccaluva

et al. (in press), and can also be adequately applied to

the Pindos ophiolites. This model implies a roll-back

of a west-dipping slab, which allowed an astheno-

spheric diapirism toward the forearc region, with

consequent shallow partial melting of a depleted

sub-arc mantle and generation of IATs and boninites.

The roll-back of the descending slab also induced

neighbouring diapirism of undepleted mantle sources,

which produced MOR-type magmas. This model also

implies the progressive transition in magma compo-

sition from SSZ-related magmas to pure MORB.

Both the models presented above can effectively

explain the compositional variations in lava types in

the Pindos ophiolites. A clear definition of the geo-

dynamic scenario, in which the Pindos ophiolites

generated, is beyond the scope of this paper; none-

theless, some aspects of the general geological frame-

work of the Pindos ophiolites can be discussed.

Falloon and Danyushevsky (2000) suggested that

boninite genesis from strongly depleted peridotites

requires a high thermal regime in the mantle source.

A subduction located near a spreading ridge implies

relatively high temperatures in the supra_subduction

mantle wedge, which can account for boninite genesis

in the Pindos ophiolites. In fact, an excessively high

thermal regime, recorded in several metamorphic

soles in Albania, characterizes the emplacement of

the Subpelagonian ophiolites (Bebien et al., 2000).

However, the depleted mantle diapirism proposed by

Beccaluva et al. (in press) can also effectively explain

such boninite genesis.

A major difference between the magmatic associ-

ations from Pindos and the contiguous Vourinos

ophiolites (Fig. 1) consists of the presence of MORB,

IAT, and boninite in the former, and IAT and boninite

in the latter (Beccaluva et al., 1984). According to the

model of east-dipping subduction presented by Bor-

tolotti et al. (2002), the Pindos sequence formed in a

proto-forearc, close to an eastward-dipping subduc-

tion zone, while the Vourinos Complex was generated

further to the east, that is, in a more mature island arc

setting. This is also in agreement with the youngest

age of the Vourinos complex with respect to Pindos

(Clift and Dixon, 1998).

The harzburgites of the Upper Ophiolitic Unit

(Dramala Complex) represent a mantle section related

to a SSZ setting, and are very similar to those found in

Vourinos (Beccaluva et al., 1984). However, the Pindos

mantle sequence lacks abundant massive chromites

(Economou-Eliopoulos and Vacondios, 1995), while

the Vourinos counterpart contains abundant podiform

chromite (Economou-Eliopoulos, 1993). On this basis,

Economou-Eliopoulos and Vacondios (1995) sug-

gested that the Pindos Upper Unit represents the mantle

portion closest to the trench, while the Vourinos was

located directly above the subduction zone.

The variation of the age of both MOR- and SSZ-

type ophiolites (Bebien et al., 2000), from older ages in

the north (Mirdita) to younger ages in the south

(Greece), suggests that Pindos ophiolites most likely

originated above an oblique subduction zone. A sim-

ilar scenario has been inferred by Robinson and

Malpas (1990) for the Troodos ophiolites, as well as

for the Andaman Sea (Moores et al., 1984), which can

be considered as a modern analogue of the Pindos

ophiolites. According to Robinson and Malpas (1990),

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 251

highly oblique subduction may imply that plate con-

sumption would be relatively small, and this could be

reflected in the small volume of arc-type magma

produced during the onset of an oblique subduction.

In fact, in contrast to what has been observed in the

Vourinos Complex (Beccaluva et al., 1984), in the

Pindos ophiolites, intrusive rocks with SSZ geochem-

ical affinity are lacking, suggesting that the IAT and

boninitic magmatisms are probably incapable of con-

tributing significant volumes of liquid to the plutonic

section. Moreover, according to Kostopoulos (1988),

the original crustal thickness of the Lower Ophiolitic

Unit appears to be anomalously thin (about 2 km).

10. Conclusions

The Pindos ophiolitic massif can be considered as

the southern prolongation of the Albanian ophiolites,

sharing with these many tectonic and petrological

characteristics and, possibly, a similar tectono-mag-

matic setting of formation.

The Pindos ophiolites can be subdivided in two

tectonically distinct ophiolitic units: (1) a Lower

Ophiolitic Unit, represented by a cumulate section

showing MOR affinity, and a volcanic section includ-

ing both MOR and boninite volcanic rocks; and (2) an

Upper Ophiolitic Unit, mainly including mantle ultra-

mafics. Both units are cross-cut by boninitic dykes,

and are characterized by the occurrence of a meta-

morphic sole and a tectono-sedimentary melange at

their bases.

The volcanic and subvolcanic sequences of the

lower unit can be geochemically subdivided into three

groups of rocks: (1) basalts and basaltic andesites

showing a clear high-Ti affinity; (2) basaltic andesites

with high-Ti affinity, but showing more depleted

geochemical features with respect to the first group;

(3) basaltic and basaltic andesitic lava flows and

dykes showing very low-Ti (boninitic) affinity.

In this paper, a petrogenetic model for explaining

the formation of the different primary lava groups in

the Pindos ophiolites is proposed according to an

incremental batch melting model. In this model, the

lower pillow sequence MORBs are generated by 20%

partial melting of an undepleted lherzolitic mantle

source leaving, as a residue, depleted lherzolitic to

harzburgitic mantle compositions. These, in turn, may

represent the probable mantle source for the upper

pillow sequence MORBs (10% partial melting), and

possibly for some boninitic lavas and dykes (>20%

partial melting). The final mantle residue is repre-

sented by harzburgites, which can be considered as

the source for the more depleted boninites, cor-

responding to 12–17% partial melting.

In order to explain the coexistence of these geo-

chemically different magma groups, two petrogenetic

models formerly proposed for the neighbouring Al-

banian ophiolites are discussed in this paper. One

model implies the initiation of an east-dipping sub-

duction process close to a mid-ocean ridge (Bortolotti

et al., 1996, 2002; Bebien et al., 2000), whereas the

other implies a roll-back of a west-dipping descending

slab (Beccaluva et al., in press).

Acknowledgements

This research was financially supported by a

M.I.U.R.-COFIN grant (project 2000). We thank L.

Beccaluva for his comments on the manuscript, as well

as R. Tassinari for the chemical analyses. We are very

grateful to J.H. Bedard and V. Hoeck whose critical

reviews greatly improved the quality of this paper.

References

Ballantyne, P., 1992. Petrology and geochemistry of the plutonic

rocks of the Halmahera ophiolite, Eastern Indonesia, an ana-

logue of modern oceanic forearcs. In: Parson, L.M., Murton,

B.J., Browning, P. (Eds.), Ophiolites and their Modern Oceanic

Analogues. Spec. Publ.-Geol. Soc. Lond., vol. 60. Blackwell,

Oxford, pp. 179–198.

Bebien, J., Shallo, M., Manika, K., Gega, D., 1998. The Shebenik

Massif (Albania): a link between MOR- and SSZ-type ophio-

lites? Ofioliti 23, 7–15.

Bebien, J., Dimo-Lahitte, A., Vergely, P., Insergueix-Filippi, D.,

Dupeyrat, L., 2000. Albanian ophiolites: I. Magmatic and meta-

morphic processes associated with the initiation of a subduction.

Ofioliti 25, 47–53.

Beccaluva, L., Serri, G., 1988. Boninitic and low-Ti subduction-

related lavas from intraoceanic arc-backarc systems and low-Ti

ophiolites: a reappraisal of their petrogenesis and original tec-

tonic setting. Tectonophysics 146, 291–315.

Beccaluva, L., Ohnenstetter, D., Ohnenstetter, M., 1979. Geochem-

ical discrimination between ocean-floor and island-arc tholei-

ites—application to some ophiolites. Can. J. Earth Sci. 16,

1874–1882.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253252

Beccaluva, L., Girolamo, P.D., Macciotta, G., Morra, V., 1983.

Magma affinities and fractionation trends in ophiolites. Ofioliti

8, 307–324.

Beccaluva, L., Ohnenstetter, D., Ohnenstetter, M., Paupy, A., 1984.

Two magmatic series with island arc affinities within the vour-

inos ophiolite. Contrib. Mineral. Petrol. 85, 253–271.

Beccaluva, L., Macciotta, G., Piccardo, G.B., Zeda, O., 1989. Cli-

nopyroxene compositions of ophiolite basalts as petrogenetic

indicator. Chem. Geol. 77, 165–182.

Beccaluva, L., Coltorti, M., Premti, I., Saccani, E., Siena, F., Zeda,

O., 1994. Mid-ocean ridge and suprasubduction affinities in the

ophiolitic belts from Albania. In: Beccaluva, L. (Ed.), Albanian

Ophiolites: State of the Art and Perspectives. Ofioliti, vol. 19,

pp. 77–96.

Beccaluva, L., Coltorti, M., Giunta, G., Siena, F., 2004. Tethyan vs.

Cordilleran ophiolites: a reappraisal of distinctive tectono-mag-

matic features of supra-subduction complexes in relation to the

subduction mode. Tectonophysics (in press).

Bedard, J.H., 1999. Petrogenesis of boninites from the betts cove

ophiolite, Newfoundland, Canada: identification of subducted

source components. J. Petrol. 40, 1853–1889.

Bonatti, E., Michael, P.J., 1989. Mantle peridotites from continental

rifted to ocean basin to subduction zones. Earth Planet. Sci. Lett.

91, 297–311.

Bortolotti, V., Kodra, A., Marroni, M., Mustafa, F., Pandolfi, L.,

Principi, G., Saccani, E., 1996. Geology and petrology of ophio-

litic sequences in the central Mirdita region (northern Albania).

Ofioliti 21, 3–20.

Bortolotti, V., Marroni, M., Pandolfi, L., Principi, G., Saccani, E.,

2002. Interaction between mid-ocean ridge and subduction mag-

matism in Albanian ophiolites. J. Geol. 110, 561–576.

Capedri, S., Venturelli, G., Bocchi, G., Dostal, J., Garuti, G., Rossi,

A., 1980. The geochemistry and petrogenesis of an ophiolitic

sequence from Pindos Greece. Contrib. Mineral. Petrol. 74,

189–200.

Capedri, S., Venturelli, G., Bebien, J., Toscani, L., 1981. Low- and

high-Ti ophiolites in Northern Pindos: petrological and geolog-

ical constraints. Bull. Volcanol. 44, 439–449.

Clift, P.D., Dixon, J.E., 1998. Jurassic ridge collapse, subduction

initiation and ophiolite obduction in the southern Greek Tethys.

Eclogae Geol. Helv. 91, 128–138.

Crawford, A.J., Falloon, A., Green, D.H., 1989. Classification, pet-

rogenesis and tectonic setting of boninites. In: Crawford, A.J.

(Ed.), Boninites and Related Rocks. Unwin Hyman, London,

pp. 1–49.

Duncan, R.A., Green, D.H., 1980. Role of multistage melting in the

formation of oceanic crust. Geology 8, 22–26.

Economou-Eliopoulos, M., 1993. Platinum-group element (PGE)

distribution in chromite ores from ophiolite complexes of Greece:

implications for chromite exploration. Ofioliti 18, 83–98.

Economou-Eliopoulos, M., Vacondios, I., 1995. Geochemistry of

chromitites and host rocks from the Pindos ophiolite complex,

northwestern Greece. Earth Planet. Sci. Lett. 122, 99–108.

Falloon, T.J., Crawford, A.J., 1991. The petrogenesis of high-cal-

cium boninite lavas dredged from the northern Tonga ridge.

Earth Planet. Sci. Lett. 102, 375–394.

Falloon, T.J., Danyushevsky, L.V., 2000. Melting of refractory man-

tle at 1.5, 2 and 2.5 Gpa under anydrous and H2O-undersaturat-

ed conditions: implication for the petrogenesis of high-Ca

boninites and the influence of subduction components on mantle

melting. J. Petrol. 41, 257–283.

Franzini, M., Leoni, L., Saitta, M., 1972. A simple method to valu-

tate the matrix effect in X-ray fluorescence analysis. X-Ray

Spectrom. 1, 151–154.

Hickey, R.L., Frey, A.F., 1982. Geochemical characteristics of

boninite series volcanics: implications for their source. Geo-

chim. Cosmochim. Acta 46, 2099–2115.

Hoeck, V., Koller, F., Meisel, T., Onuzi, K., Kneringer, E., 2002.

The jurassic south Albanian ophiolites: MOR- vs. SSZ-type

ophiolites. Lithos 65, 143–164.

Insergueix-Filippi, D., Dupeyrat, L., Dimo-Lahitte, A., Vergely, P.,

Bebien, J., 2000. Albanian ophiolites: II. Model of subduction

zone infancy at a Mid-ocean ridge. Ofioliti 25, 47–53.

Jones, G., Robertson, A.H.F., 1991. Tectono-stratigraphic evolution

of the Mesozoic Pindos ophiolite and related units, northwestern

Greece. J. Geol. Soc. (Lond.) 148, 267–288.

Jones, G., Robertson, A.H.F., Cann, J.R., 1991. Genesis and em-

placement of the supra-subduction zone Pindos ophiolite, north-

western Greece. In: Peters, Tj., Nicolas, A., Coleman, R.G.

(Eds.), Ophiolite Genesis and Evolution of the Oceanic Litho-

sphere. Ministry of Petroleum and Minerals, Sultanate of Oman.

Kluwer Academic Publishing, Dordrecht, pp. 771–799.

Kemp, A.E.S., McCaig, A., 1984. Origin and significance of rocks

in an imbricate thrust zone beneath the Pindos ophiolite, north-

western Greece. In: Robertson, A.H.F., Dixon, J.E. (Eds.), The

Geological Evolution of the Eastern Mediterranean. Spec. Publ.-

Geol. Soc. Lond., vol. 17. Blackwell, Oxford, pp. 569–580.

Klein, E.M., Kastens, J.L., 1995. Ocean-ridge basalts with conver-

gent-margin geochemical affinities from the Chile Ridge. Nature

374, 52–57.

Kostopoulos, D., 1988. Geochemistry and tectonic setting of the

Pindos ophiolite, northwestern Greece. PhD thesis, University

of Newcastle-upon-Tyne, 498 p.

Leterrier, J., Maury, R.C., Thonon, P., Girard, D., Marchal, M.,

1982. Clinopyroxene composition as a method of identification

of the magmatic affinities of paleo-volcanic series. Earth Planet.

Sci. Lett. 59, 139–154.

Montigny, R., Bougault, H., Bottinga, Y., Allegre, C.J., 1973. Trace

element geochemistry and genesis of the Pindos ophiolite suite.

Geochim. Cosmochim. Acta 37, 2135–2147.

Moores, E.M., Robinson, P.T., Malpas, J., Xenophontos, C., 1984.

Model for the origin of the Troodos massif, Cyprus, and other

mideast ophiolites. Geology 12, 500–503.

Murton, B.J., 1989. Tectonic controls on boninite genesis. In:

Saunders, A.D., Norry, M.J. (Eds.), Magmatism in the Ocean

Basins. Spec. Publ.-Geol. Soc. Lond., vol. 42. Blackwell, Ox-

ford, pp. 347–377.

Pearce, J.A., 1983. Role of the sub-continental lithosphere in mag-

ma genesis at active continental margin. In: Hawkesworth, C.J.,

Norry, M.J. (Eds.), Continental Basalts and Mantle Xenoliths.

Shiva, Nantwich, pp. 230–249.

Pearce, J.A., Norry, M.J., 1979. Petrogenetic implications of Ti, Zr,

Y, and Nb variations in volcanic rocks. Contrib. Mineral. Petrol.

69, 33–47.

E. Saccani, A. Photiades / Lithos 73 (2004) 229–253 253

Pedersen, R.B., Hertogen, J., 1990. Magmatic evolution of the

Karmoy ophiolite complex, SW Norway: relationships between

MORB-IAT-boninitic-cal-alkaline and alkaline magmatism.

Contrib. Mineral. Petrol. 104, 277–293.

Rassios, A., 1990. Geology and copper of the Vrinea area, East

Othris ophiolite, Greece. Ofioliti 15, 287–304.

Reagan, M.K., Meijer, A., 1984. Geology and geochemistry of

early arc-volcanic rocks from Guam. Geol. Soc. Amer. bull.

95, 701–713.

Robertson, A.H.F., 1994. Role of the tectonic facies concept in

orogenic analysis and its application to Tethys in the Eastern

Mediterranean region. Earth-Sci. Rev. 37, 139–213.

Robertson, A.H.F., Shallo, M., 2000. Mesozoic– tertiary tectonic

evolution of Albania in its regional Eastern Mediterranean con-

text. Tectonophysics 316, 197–254.

Robinson, P.T., Malpas, J., 1990. The Troodos ophiolites of Cyprus:

new perspectives on its origin and emplacement. In: Malpas, J.,

Moores, E.M., Panayiotou, A., Xenophontos, C. (Eds.), Ophio-

lites, Oceanic Crustal Analogues, Proceedings of the Sympo-

sium ‘‘Troodos 1987’’. Published by the Geological Survey

and Natural Resources Nicosia, Cyprus, pp. 13–26.

Ross, J.V., Zimmerman, J., 1996. Comparison of evolution and

tectonic significance of the Pindos and Vourinos ophiolite suites,

northern Greece. Tectonophysics 256, 1–15.

Serri, G., 1981. The petrochemistry of ophiolite gabbroic com-

plexes: a key for the classification of ophiolites into low-Ti

and high-Ti types. Earth Planet. Sci. Lett. 52, 203–212.

Shallo, M., 1994. Outline of the Albanian ophiolites. In: Beccaluva,

L. (Ed.), Albanian Ophiolites: State of the Art and Perspectives.

Ofioliti, vol. 19, pp. 57–75.

Shervais, J.W., 1982. Ti–V plots and the petrogenesis of modern

ophiolitic lavas. Earth Planet. Sci. Lett. 59, 101–118.

Smith, A.G., 1993. Tectonic significance of the Hellenic–Dinaric

ophiolites. In: Prichard, H.M., Alabaster, T., Harris, N.B.W.,

Neary, C.R. (Eds.), Magmatic Processes and Plate Tectonics.

Spec. Publ.-Geol. Soc. Lond., vol. 76. Blackwell, Oxford,

pp. 213–243.

Spray, J.G., Bebien, J., Rex, D.C., Roddick, J.C., 1984. Age

constraints on the igneous and metamorphic evolution of the

Hellenic –Dinaric ophiolites. In: Dixon, J.E., Robertson,

A.H.F. (Eds.), The Geological Evolution of Eastern Mediterra-

nean. Spec. Publ.-Geol. Soc. Lond., vol. 17. Blackwell, Ox-

ford, pp. 619–627.

Sun, S.S., 1982. Chemical composition and origin of the earth’s

primitive mantle. Geochim. Cosmochim. Acta 46, 179–192.

Sun, S.S., McDonough, W.F., 1989. Chemical and isotopic system-

atics of ocean basalts: implications for mantle composition and

processes. In: Saunders, A.D., Norry, M.J. (Eds.), Magmatism

in the Ocean Basins. Spec. Publ.-Geol. Soc. Lond., vol. 42.

Blackwell, Oxford, pp. 313–346.

Tatsumi, Y., Eggins, S., 1995. Subduction Zone Magmatism. Black-

well, London. 211 pp.

Thuizat, R., Whitechurch, H., Montigny, R., Juteau, T., 1981.

K–Ar dating of some infra-ophiolitic metamorphic soles

from the eastern Mediterranean: new evidence for oceanic

thrusting before obduction. Earth Planet. Sci. Lett. 52,

302–310.

Yumul, G.P., 1996. Varying mantle source of supra-subduction zone

ophiolites: REE evidence from the Zambales Ophiolite Com-

plex, Luzon, Philippines. Tectonophysics 262, 243–262.