Journal of Petrology Volume 43 Number 1 Pages 17-43 2002
© Oxford University Press 2002
Genesis of Pyroxenite-rich Peridotite at Cabo Ortegal (NW Spain): Geochemical and Pb–Sr–Nd Isotope Data
J. F. SANTOS1,*,
U. SCHÄRER2,
J. I. GIL IBARGUCHI1 and
J. GIRARDEAU3
1DEPARTAMENTO DE MINERALOGÍA–PETROLOGÍA, UNIVERSIDAD DEL PAÍS VASCO, APTDO. 644, 48080 BILBAO, SPAIN
2LABORATOIRE DE GÉOCHRONOLOGIE, UNIVERSITÉ PARIS 7-IPGP, 2 PLACE JUSSIEU, 75251 PARIS CEDEX 05, FRANCE
3LABORATOIRE DE PLANÉTOLOGIE ET GÉODYNAMIQUE, UNIVERSITÉ DE NANTES, 2 RUE DE LA HOUSSINIÈRE, BP 92208, 44322 NANTES CEDEX 3, FRANCE
Received
December 10, 1999;
Revised typescript accepted
June 25, 2001
 |
ABSTRACT
|
|---|
Petrographic and field data indicate the existence of four main
rock types within the allochthonous Cabo Ortegal ultramafic
units: (1) harzburgites; (2) dunites; (3) massive, occasionally
garnet-bearing, pyroxenites; (4) less abundant mafic rocks with
variable amounts of garnet-rich pyroxenite. The major and trace
element compositions of the analysed ultramafic rocks define
well-delimited fields in binary variation diagrams. Normalized
trace element patterns, however, exhibit large ion lithophile
element (LILE) and light rare earth element (LREE) enrichment
that do not correlate with the main rock types distinguished.
NiO contents and fo-number of olivine in the harzburgites match
those of the mantle olivine array, whereas a fractional crystallization
trend is observed from dunites to pyroxenites. Spinel and olivine
in the harzburgites have residual characteristics comparable
with those of abyssal peridotites or peridotites from arc settings,
whereas in most of the dunites and pyroxenites the range of
fo-number and Cr/(Cr + Al) ratio suggests crystallization from
primitive subduction-related magmas. Whole-rock major and trace
element and Pb–Sr–Nd isotope data suggest that regional-scale
massive pyroxenites from Cabo Ortegal originated from relatively
homogeneous parental melts. Fractional crystallization processes,
coeval with intense deformation, might result in the formation
of cumulate layers (clinopyroxene, orthopyroxene, olivine, chromite,
etc.). Some less abundant mafic rocks and associated pyroxenites
are also homogeneous but have different chemical and isotopic
signatures suggesting a different parental melt from that of
the massive pyroxenites. Although some differences exist in
the major element and isotopic composition of the clinopyroxenes,
their initial isotopic ratios (
206Pb/
204Pb = 17·845–18·305,
207Pb/
206Pb = 15·433–15·634;
87Sr/
86Sr =
0·70330–0·70476;
143Nd/
144Nd = 0·512539–0·512916)
suggest involvement of an enriched component in their mantle
source, which may be related to the subduction of terrigenous
sediments (i.e. EMI). The new data obtained confirm that ultramafic
units of Cabo Ortegal experienced a complex tectonothermal history
similar to that of other units of the same area and allow us
to distinguish at least two different events. Sm–Nd whole-rock–clinopyroxene
ages suggest formation of the ultramafic units at
500 Ma, an
age similar to that of formation of the protoliths of associated
HP/HT units. Internal Sm–Nd isochrons (Cpx–Grt–whole
rock) from two pyroxenites indicate ages of
390 Ma for garnet
crystallization in these rocks, which is consistent with previous
U–Pb dating of HP/HT recrystallization in Cabo Ortegal.
Pyroxenite–dunite-rich ultramafic massifs such as Cabo
Ortegal might have originated within the lithospheric mantle
above a subduction zone, variably modified by fluid or melt
infiltration from the subducted oceanic crust.
KEY WORDS: geochemistry; isotopes; Ortegal; pyroxenites; mantle
|
INTRODUCTION
|
|---|
Pyroxenite layers parallel or subparallel to the dominant foliation
of the host peridotite generally constitute <10% of the volume
of ultramafic massifs (e.g. Fabriès
et al., 1991
). They
are normally centimetres to metres thick but larger pyroxenite
bodies within peridotite have been reported from a few localities,
e.g. Cabo Ortegal, Kohistan and Tonsina (Girardeau
et al., 1989
;
DeBari & Sleep, 1991
; Burg
et al., 1998
, and references
therein). To explain the origin of pyroxenite, either in-folding
of subducted oceanic crust through convection into the mantle
(Allègre & Turcotte, 1986
) or crystallization from
a melt derived from a peridotitic mantle source (e.g. Suen &
Frey, 1987
) has been proposed. Recent work on various pyroxenite-bearing
peridotite massifs, however, has left open the question of the
origin of the magmas that formed the pyroxenites, and a combination
of both processes, including strong interaction with surrounding
peridotite, has been considered (Kelemen & Ghiorso, 1986
;
Downes
et al., 1991
; Becker, 1996
; Zanetti
et al., 1999
).
In the Cabo Ortegal complex of NW Spain, a concentration of metre-thick websterite layers separated by dunite, locally alternating with garnet-rich mafic rocks, forms a kilometre size pyroxenite body within the Herbeira peridotite massif (Fig. 1). Several explanations exist for this body, including high-pressure crystallization of melts that intruded the solid peridotite at 390 Ma (Girardeau & Gil Ibarguchi, 1991; Santos et al., 1996) or that it is part of a layered complex overlying tectonized mantle harzburgite (Moreno et al., 1999). Controversy exists about the origin of the magma that crystallized to form the pyroxenites. Models have included partial melting of a metasomatized mantle wedge (Gravestock, 1992), of primitive mantle in an oceanic setting (Laribi-Halimi, 1992) or of the mantle wedge underlying an arc root (Moreno et al., 2001). In an attempt to elucidate the origin of the pyroxenites and to investigate mantle heterogeneity on a regional scale, new U–Pb, Pb–Pb, Sm–Nd and Rb–Sr isotopic data have been obtained from mineral separates and whole rocks from selected pyroxenites and peridotites from the Cabo Ortegal complex.
|
GEOLOGICAL SETTING
|
|---|
The uppermost units of the Cabo Ortegal allochthonous complex
comprise ultramafic rocks, granulites, gneisses and eclogites
(Fig.
1; Arenas
et al., 1986
; Gil Ibarguchi
et al., 1990
). Ultramafic
rocks are exposed in the Limo, Herbeira and Uzal massifs and
in dispersed outcrops associated with high-pressure granulites.
They are mostly harzburgite, pyroxenite and dunite in decreasing
order of abundance. Pyroxenites occur as: (1) scattered metre-size
layers usually parallel to the main foliation of the host harzburgite;
(2) a composite multilayer pyroxenite–dunite unit
3 km
long and >500 m thick within the Herbeira peridotite massif.
This multilayer sequence may have originated through folding
and boudinage of a pyroxenite–dunite-rich fragment of
mantle peridotite during shearing related to emplacement of
the massif (Girardeau & Gil Ibarguchi, 1991
). More recently,
the pyroxenites have been viewed as the middle part of an essentially
tripartite dunite–pyroxenite–dunite ultramafic layered
complex overlying a basement of tectonized harzburgite (Moreno
et al., 2001
).
New mapping of the Cabo Ortegal complex (Abalos et al., 2000) has revealed that the Herbeira peridotite massif forms an upright asymmetric synform with a thinner gently dipping eastern limb (Fig. 1b). The thickness of ultramafic rocks (normal to the foliation) decreases from 1000–1500 m in the western limb to 750 m in the east. The thickness of the central part of the main pyroxenite–dunite body as seen along the western cliffs of the massif (Fig. 1b) changes from 450 m (S) to 150 m (N). This part is formed essentially of pyroxenite (80–90%) alternating with dunite and less abundant harzburgite. Towards the base and top, this pyroxenite-rich portion changes into a dunite-rich unit in which pyroxenites and harzburgite occur in greatly reduced amounts. The thickness of the lower dunite-rich unit is 100 m, whereas the upper dunite is >400 m thick. Towards the eastern limb of the massif, both dunite-rich sections appear to coalesce in a single outcropping dunite-rich unit 100 m thick (Fig. 1b). The rest of the Herbeira massif is composed of harzburgite up to 500 m in thickness below the basal dunite and 
200 m thick above the upper dunite. The wedge shape of the units described is not as spectacular as the thickness variations described seem to imply. The calculated angle between the top and the base of any of the dunite or pyroxenite units is 5°.
The central pyroxenite-rich section is composed of clinopyroxenite and websterite layers up to several metres thick, separated by dunite and less often by harzburgite. Garnet-bearing pyroxenites occur randomly throughout the pyroxenite–dunite section, whereas olivine websterite, orthopyroxenite and phlogopite-bearing varieties occur in minor amounts. Towards the top of the pyroxenite-rich unit, and also at one locality within the harzburgite (samples GRCP6, GZC7, GZA2; Fig. 1b), decimetre- to metre-thick layers of garnet–zoisite-rich mafic rocks and garnet–rutile-rich pyroxenite occur. Peridotites and pyroxenite layers are locally enriched in amphibole (to >30% modal) and cut by late garnet-rich and pyroxenite veins. Girardeau & Gil Ibarguchi (1991), on the basis of major element mineral data, proposed an affinity of the peridotites with residual oceanic tectonites and suggested that the pyroxenites correspond to the products of high-pressure crystal fractionation in dykes or sills intruded into the peridotite during subduction. The peridotite–pyroxenite association from the equivalent Bragança complex in north Portugal (Fig. 1) has also been interpreted as magma intrusion into a depleted mantle protolith in a supra-subduction zone setting (Bridges et al., 1995). In Cabo Ortegal the pyroxenites, peridotites and mafic rocks have a common tectonothermal evolution with nearby granulites and eclogites (Girardeau & Gil Ibarguchi, 1991; Abalos et al., 1996). They underwent, however, a more complex tectonothermal history involving a first episode of shear deformation at high temperature (>1000°C) and relatively low deviatoric stress (300–600 bars), followed by late- to syn-kinematic development of garnet after spinel in pyroxenites (including olivine websterite) and mafic rocks at 800°C, 1·65 GPa. Amphibole was probably produced during subsequent decompression at relatively high temperature (>750°C). A comprehensive account of the petrography and structure of these rocks has been given by Girardeau & Gil Ibarguchi (1991).
|
PREVIOUS GEOCHRONOLOGICAL AND GEOCHEMICAL DATA
|
|---|
The age of the Cabo Ortegal peridotite unit was poorly constrained
at 477 ± 122 Ma by Van Calsteren
et al. (1979
; Rb–Sr
in whole rock). Rutile from pyroxenite and magmatic zircon from
pyroxenite and peridotite yielded more precise U–Pb ages
of
383–395 Ma (Peucat
et al., 1990
; Santos
et al., 1996
;
Ordóñez Casado
et al., 2001
), and a granitic pegmatite
intrusive rock in the Uzal peridotite yielded
385–395
Ma (U–Pb monazite and zircon ages and Rb–Sr isochron;
Santos
et al., 1996
). This substantiates a magmatic event around
390 ± 10 Ma nearly coeval with HP/HT metamorphism dated
at
382–395 Ma on adjacent eclogites, granulites and HP
gneisses (U–Pb zircon, titanite and rutile ages, Santos
et al., 1996
; Valverde Vaquero & Fernández, 1996
;
Ordóñez Casado
et al., 2001
). Phlogopite–
and amphibole–whole-rock pairs from harzburgites have
been dated at 386 ± 10 and 337 ± 10 Ma, respectively
(Rb–Sr age, Van Calsteren
et al., 1979
).
Previous geochemical data for Cabo Ortegal ultramafic rocks have been reported by Vogel (1967), Maaskant (1970), Van Calsteren et al. (1979), Peucat et al. (1990), Gravestock (1992) and Laribi-Halimi (1992). On the basis of major and limited trace element and Sr isotope data, and the field association with high-temperature metabasites and gneisses, the origin of the ultramafic rocks was considered by Van Calsteren et al. (1979) to be related to Lower Palaeozoic mantle plume activity within an intra-continental setting. For those workers, the peridotites correspond to slightly depleted spinel lherzolites whose partial melting products, represented in part by the pyroxenites, would have crystallized in place during ascent and emplacement of a mantle diapir at the base of the continental crust. In contrast, later results, including Nd isotope data (Peucat et al., 1990) were interpreted in terms of an oceanic mantle setting modified with crustal components for the peridotite–pyroxenite association.
Gravestock (1992) used Pb, Sr and Nd isotope data together with modelling of HFSE and REE compositions to propose that the Cabo Ortegal pyroxenites represent crystal cumulates. However, they could not be related to parent magmas by a simple igneous fractionation process and Gravestock suggested a depleted source in a mantle-wedge environment metasomatized by a low-viscosity carbonatite melt as a likely scenario. Using mainly major and trace element whole-rock data, Laribi-Halimi (1992) proposed variable degrees of partial melting of primitive mantle in relation to mid-ocean ridge basalt (MORB) magma generation for the origin of the peridotites, and fractional crystallization of tholeiitic magmas for the pyroxenites. Finally, on the basis of the platinum-group element (PGE) contents of the ultramafic units, Moreno et al. (2001) considered the existence of two batches of mantle-derived magmas from which the pyroxenite–dunite association crystallized at the crust–mantle interface of an arc-root region.
|
SAMPLE DESCRIPTION
|
|---|
Harzburgite
Intensely deformed and serpentinized (up to >70% serpentine)
amphibole-bearing harzburgite (samples 432, H3, Table
1) constitutes
most of the ultramafic outcrops. These rocks were previously
described as lherzolite and the amphibole was interpreted as
resulting from transformation of primary clinopyroxene during
retrograde amphibolite-facies metamorphism (Vogel, 1967
; Maaskant,
1970
; Van Calsteren, 1978
). Indeed, they are comparable with
mantle lherzolites worldwide, in their high contents of CaO
(1·7–4·3%) and Al
2O
3 (2·2–4·1%;
Table
1). However, they are clearly harzburgites in terms of
their modal composition (Table
1). Harzburgites contain sub-spherical,
5–10 mm, orthopyroxene porphyroclasts or aggregates in
a matrix of olivine and amphibole, minor clinopyroxene as isolated
crystals or as part of discontinuous microlayers of millimetre
thickness, and green to brown spinel. Orthopyroxene accounts
for 10–20% of the rock volume and generally contains exsolution
lamellae of clinopyroxene parallel to (100). More rarely, orthopyroxene
contains exsolution of Al-rich spinel towards the grain margins,
which may result in the formation of minute spinel grains along
crystal boundaries. Olivine forms 1–3 mm porphyroclasts
and <0·5 mm neoblasts. Clinopyroxene (<2% modal)
forms anhedral porphyroclasts rich in fine exsolution lamellae
of orthopyroxene. Green and brown subidiomorphic spinel up to
5 mm in diameter forms 1–3% of the rock volume. It is
usually disseminated and more rarely forms discontinuous aggregates
parallel to the main foliation.
Harzburgite may be rich (10–30 modal %) in subidiomorphic, unzoned, pale green magnesio-hornblende and more rarely pargasitic to edenitic hornblende in textural equilibrium with olivine and pyroxene. Amphibole may contain inclusions of brown–green spinel and often forms aggregates or thin microlayers parallel to the main foliation. Within fine-grained, mylonitized and strongly deformed rocks, amphibole has grown syn- to post-kinematically around or in pressure shadows of orthopyroxene porphyroclasts. More rarely, hornblende has the features of a phase formed from exsolution lamellae in orthopyroxene. Tremolitic amphibole occurs replacing hornblende and orthopyroxene along grain margins. Phlogopite in textural equilibrium with olivine, pyroxene and amphibole is a rare accessory. Rare magnetite may occur associated with Ni and Fe sulphides, and millimetre-size chlorite, often associated with magnetite, is locally present defining a late schistosity.
Dunite
Dunite within the pyroxenite–dunite section of Herbeira is found as (1) centimetre- to decimetre- thick bands alternating with pyroxenite and, less often, with harzburgite bands of similar thickness, in the lower and middle parts of the section, and (2) outcrops of several metres in size that include scattered pyroxenite layers in the upper portions of the section. Less commonly, dunite forms metre- to decametre-size pods with diffuse contacts within harzburgite. Pyroxenites are generally more abundant (up to 30%) in orthopyroxene- and amphibole-bearing varieties of dunite. Massive dunite shows a foliation defined by elongated crystals of spinel and is generally less serpentinized than harzburgite. Massive dunite and dunite bands contain millimetre-size olivine porphyroclasts and <0·5 mm neoblasts as in harzburgite, sporadic, small (1 mm), interstitial orthopyroxene and large (
3 mm) interstitial clinopyroxene grains. Subidiomorphic millimetre-size chromian spinel occasionally associated with platinum-group minerals may concentrate in pods or layers >10 cm thick (Monterrubio et al., 1992). In this way, Moreno et al. (2001) have distinguished the PGE-poor lower dunites, which are considered genetically related to the massive pyroxenites, from the PGE-rich upper dunites. Unzoned edenitic- to magnesio-hornblende occurs in minor amounts. Phlogopite is an accessory in some samples. Secondary serpentine, chlorite and magnetite are common. Dunite was not analysed for whole-rock composition but only for mineral composition in this study.
Massive pyroxenites and pyroxenite bands
Pyroxenites, which generally form homogeneous layers with sharp contacts to peridotite, mainly comprise medium-grained, orthopyroxene-rich, websterite and clinopyroxenite in equal amounts. Pegmatoid facies containing large orthopyroxene crystals (>5 cm) occur sporadically, and coarse-grained (2–5 mm) orthopyroxenite locally forms massive bands up to 30 cm thick. Websterites may show a poor layering marked by discontinuous thin orthopyroxenite layers and are occasionally zoned with olivine-rich rims and finer-grained, clinopyroxene-rich central parts. Garnet pyroxenites occur randomly throughout the pyroxenite–dunite section. Samples of orthopyroxenite (OP10), websterite (136, 198a), garnet websterite (GW1, GW7692), olivine websterite (196c), and clinopyroxenite (430a, CP11) have been studied (Table 1).
Pyroxenites are usually very fresh rocks and exhibit poorly oriented granoblastic textures. Orthopyroxene is often strained with evidence of dynamic recrystallization at grain boundaries. Clinopyroxene appears undeformed, except for local undulose extinction and mechanical twins. Olivine and idiomorphic green spinel in disseminated grains, more rarely forming elongated aggregates, are not abundant; spinel exsolution in pyroxene is exceedingly rare. Optically homogeneous amphibole may constitute >30% by volume in some samples, forming generally large (>1 cm) subidiomorphic crystals without deformation. This amphibole is in textural equilibrium with the pyroxenes although, occasionally, it may have developed after clinopyroxene. Garnet often occurs in coronas around spinel and/or ilmenite and may contain sporadic inclusions of corundum. Garnet may also form aggregates of subidiomorphic crystals with or without spinel relicts. Carbonates, close to pure dolomite and calcite, are rare accessories in apparently textural equilibrium with pyroxene and garnet. Interstitial phlogopite associated with ilmenite is also a rare accessory except in pyroxenites near late tectonic contacts, where it may be abundant. Garnet websterite GW7692 was originally interpreted as a vein cross-cutting the foliation at Uzal (sample 7692, Peucat et al., 1990). Actually, this sample comes from a strongly deformed portion of the Uzal peridotite massif where structural relations are obscured. As it is identical to garnet pyroxenites elsewhere within the pyroxenite–dunite section and pyroxenites parallel to the foliation in harzburgite, and very different from the late pyroxenite veins cross-cutting the foliation (see below), it has been included in the present study as belonging to the common pyroxenite unit.
Mafic rocks and related pyroxenites
Rare mafic rocks with variable amounts of associated garnet-rich pyroxenites occur at two localities near the top of the pyroxenite-rich section lying parallel to the pyroxenite–dunite banding. Garnet- and zoisite-rich metre-thick outcrops are composed of bands of 5–20 cm thickness containing either more clinopyroxene than amphibole (288a, 288b, GZC7), or more amphibole and practically no clinopyroxene (288, 431a, 431b, GZA2; Table 1). These rocks exhibit oriented granoblastic textures with millimetre-size garnet, often with inclusions of Cr-rich spinel and pink corundum, forming up to 70% of the rock volume. Subidiomorphic millimetre- to centimetre-size zoisite in textural equilibrium with pyroxene and amphibole is present in all cases and may constitute up to 40% of the rock volume (GZA2; Table 1). Unzoned amphibole in textural equilibrium is magnesio-hornblende and more rarely tschermakitic hornblende. K-feldspar and plagioclase have been found as accessories in one sample. Kyanite inclusions in subidiomorphic zoisite were also found in one sample. A gradual transition exists between the garnet–zoisite rocks and garnet–rutile clinopyroxenite (GRCP6; Table 1). This is composed essentially of clinopyroxene and garnet in textural equilibrium (60 and 30%, respectively), subidiomorphic, prismatic to rounded rutile and lesser amounts of ilmenite. Amphibole, locally accompanied by clinozoisite, is present in small amounts interstitially and along fractures within clinopyroxene and garnet, whereas rutile is occasionally replaced by titanite.
Other rock types
Late granitic intrusive rocks, pyroxenite dykes and garnet-rich dykes are less abundant rock types. The main granitic intrusive type is a pegmatite granite (OGS23) several metres thick and >30 m long within mylonitic harzburgite of the Uzal massif at the Cartés beach locality (Santos et al., 1996). The granite pegmatite is of the same age as the pyroxenites (see above). It contains decimetre- to centimetre-size blocks of peridotite with rims of chlorite and actinolite. The pegmatite exhibits a porphyroclastic texture with a foliation parallel to that in the host peridotite and is composed of centimetre-size K-feldspar, muscovite and less abundant plagioclase (An15), surrounded by a fine-grained matrix composed of the same minerals, phlogopite and quartz. Allanite, zircon, apatite and monazite are accessories. Decimetre-size granitic pods also occur at several places within the Herbeira peridotite massif. These are formed of plagioclase, quartz and biotite, with reaction rims of Mg-arfvedsonite and wollastonite at the contact with the peridotite; allanite, zircon and apatite are accessories.
Sparse mafic dykes occur cutting through massive pyroxenites and surrounding peridotites. These dykes, generally <10 cm thick, are medium grained (3–5 mm), may display orthopyroxene-rich outer zones and clinopyroxene-rich cores and, when included in harzburgite, dunitic rims. Hornblende, olivine and phlogopite occur in minor amounts. The garnet-rich dykes, sometimes filled only by garnet with spinel cores, are thinner and generally zoned, displaying thin clinopyroxene- or amphibole-rich rims and garnet-rich central parts. Amphibole and chlorite are abundant; carbonates, phlogopite, titanian spinel and ilmenite are rare accessories. Preliminary geochemical data by Gravestock (1992), including mineral isotope compositions, suggest that these rocks are genetically unrelated to the main peridotite–pyroxenite unit.
|
MAJOR ELEMENT MINERAL CHEMISTRY
|
|---|
Mineral compositions were determined by electron microprobe
methods using Camebax Micro and SX50 instruments at Clermont-Ferrand
and Paris-VII universities, respectively. Mineral analyses with
details of analytical procedures, structural formulae and end-member
calculations have been presented elsewhere by Girardeau &
Gil Ibarguchi (1991)
. An interpretation of the analytical data
in terms of the
P–
T evolution undergone by the ultramafic
rocks, stressing the residual character of the mineralogy of
peridotites, was presented in that study. A refined interpretation
aimed at the origin of peridotites and pyroxenites is presented
here, including new analytical data. Minerals from 46 samples
of pyroxenite, eight dunites, one harzburgite and two mafic
rocks spanning the whole 500 m thickness of the pyroxenite–dunite
section, and 16 harzburgites, eight pyroxenites and one dunite
from the Herbeira, Limo and Uzal peridotite massifs have been
analysed. The complete dataset is available from the
Journal of Petrology Web site at
http:www.petrology.oupjournals.org.
Olivines from harzburgite, dunite and pyroxenite are unzoned and have fairly constant low Ca, Al, Cr and Ti contents across the pyroxenite–dunite section of Herbeira. Olivine in pyroxenites has distinctly lower average fo-number and NiO contents (0·85 and 0·19% NiO, respectively) than olivine in dunite and harzburgite. Olivine in the harzburgite has similar fo-number (0·9) but higher Ni content (0·37% NiO, as average) than olivine in dunite (0·29% NiO, as average) and it overlaps the ‘mantle olivine array’ (Fig. 2a). In turn, olivines from dunites and pyroxenites deviate from the ‘mantle olivine array’ and define broader fields than harzburgite olivine, involving a rapid decrease of NiO content and fo-number from dunite to pyroxenite (Fig. 2b).
Spinel compositions are generally Al-rich in harzburgite, Cr-rich in dunite and range from Al- to Cr-rich in pyroxenites. Limited variations in spinel composition may be due in part to late re-equilibration processes as suggested by textural and colour changes of this mineral (Girardeau & Gil Ibarguchi, 1991). In a plot of Cr/(Cr + Al) of spinel and fo-number of olivine, dunites and harzburgites plot in separate fields and do not follow the ‘olivine–spinel mantle array’ (Fig. 2c). Olivines in the pyroxenites are characterized by lower Fo contents than those in harzburgites and dunites over a wide range in spinel Cr/(Cr + Al) plotting outside the olivine–spinel array (Fig. 2c).
Clinopyroxene from peridotites and pyroxenites is a Cr-, Ti- and Na-poor diopside. Despite an overall similarity, there exist some differences in composition among clinopyroxenes from different rock types. In a plot of Cr vs Na (Fig. 3a), clinopyroxenes from massive pyroxenites and dunites define a broad trend with an increase in Cr and Na towards dunite. Clinopyroxenes from harzburgites have variable Cr and Na contents generally overlapping the lowest values from pyroxenites and dunites, and clinopyroxenes from mafic rocks and related pyroxenites are Cr-poor with variable, generally higher, Na contents. In a plot of AlZ vs TiO2, the whole set of clinopyroxene analyses shows a limited compositional range with clinopyroxenes from mafic rocks and related pyroxenites exhibiting generally higher AlZ contents (Fig. 3b).
Amphibole from harzburgites, dunites and pyroxenites is unzoned magnesio-hornblende and, more rarely, tschermakitic-hornblende, with low contents of Ti, Fe and Mn, and moderate contents of K, Na and Cr. Amphiboles from dunites are generally richer in Na (1·6–2·4% Na2O) and Cr (1–1·6% Cr2O3) than those of pyroxenites (1·1–2·1% Na2O, 0·3–1·6% Cr2O3), whereas amphiboles from harzburgites have intermediate Na and Cr contents (1·2–2% Na2O, 0·5–1·2% Cr2O3). All these amphiboles are characterized by high Al[4]/(Na + K)A ratios suggesting significant pargasite (edenite + tschermakite) substitution.
|
WHOLE-ROCK MAJOR AND TRACE ELEMENT COMPOSITIONS
|
|---|
Three rock groups appear well characterized in variation diagrams
of MgO vs other elements (Fig.
4, Table
1). Harzburgite and
dunite have very similar compositions and plot at the Mg-rich
end (>30% MgO), massive pyroxenites show limited overlap
with peridotites and lower MgO values (down to 16·7%),
and mafic rocks and associated pyroxenites have MgO < 16%
(Table
1). Compared with pyroxenites from ultramafic massifs
elsewhere, e.g. Horoman, Ronda, Beni Boussera, Alps or Pyrenees,
massive pyroxenites from Herbeira show very different behaviour
with generally higher SiO
2, CaO, Cr and Sc contents and lower
Al
2O
3, TiO
2 and Ni contents. Only the D1 subtype of websterites
from the Ronda massif are found to be relatively similar, although,
as in the other cases, CaO/Al
2O
3 ratios in Cabo Ortegal are
significantly higher (Fig.
4).
View larger version (48K):
[in this window]
[in a new window]
|
Fig. 4. Selected major and trace element variation in bulk rock vs weight percent MgO for ultramafic and related rocks of Cabo Ortegal (data from this study, and from Vogel, 1967; Maaskant, 1970; Van Calsteren et al., 1979; Peucat et al., 1990; Laribi-Halimi, 1992); vertically and horizontally ruled areas indicate pyroxenites and peridotites, respectively, from Gravestock (1992). Other fields include data from Beni Boussera (Pearson et al., 1993; Kornprobst, 1969), Horoman (Yoshida & Takahashi, 1997; Takazawa et al., 1999, 2000), Pyrenees (Fabriès & Conquéré, 1983; Conquéré & Fabriès, 1984), Alps (Capredi et al., 1977; Garuti et al., 1980; Rivalenti et al., 1980; Pognante et al., 1985) and Ronda (Obata et al., 1980; Garrido & Bodinier, 1999).
|
|
Normalization to a primitive mantle composition shows that all the analysed rocks have slight LILE enrichment, with values intermediate between those of N-MORB and E-MORB, and heavy REE (HREE) depletion, except GRCP6 (Fig. 5a; Tables 2 and 3). Ti exhibits a variable character, samples with a negative Ti anomaly being richer in La to Sr than other samples and corresponding to clinopyroxene-rich rocks as previously shown by Gravestock (1992), whereas orthopyroxene-rich rocks show the opposite trend with a positive Ti anomaly, in agreement with the results by Rampone et al. (1991) on Ti partition in orthopyroxene and clinopyroxene. Other significant features are the generalized positive anomalies for Zr, Sr and U and negative anomaly for K, the higher contents of U respect to Th, and the enrichment in Nb with respect to mantle values (Fig. 5a).
REE contents (Table 3) normalized to chondrite values show generally variable enrichment in LREE [(La/Sm)n between 1·45 and 20] with small positive Eu anomalies (Eu/Eu* from 1·08 to 2·26) and nearly flat HREE patterns [(Gd/Lu)n 0·94–1·49], a feature noted previously by Peucat et al. (1990) and Gravestock (1992). In detail, samples H3, GW1 and GRCP6 have relatively flat patterns with variable Eu anomalies and concentrations (Fig. 5b), whereas samples OP10, CP11 and GZC7 present distinct LREE-enriched patterns and positive Eu anomalies (Fig. 5c). The latter patterns are similar to those of pyroxenites analysed by Gravestock (1992) from the Herbeira massif and to the analysis of harzburgite also from Herbeira reported by Peucat et al. (1990).
Trace element contents of the Cabo Ortegal rocks thus appear unique and different from those of other pyroxenites in ultramafic massifs, including the D1 subtype from Ronda, although locally there may be pyroxenites of composition resembling those from Ortegal (e.g. the low-MgO pyroxenites from Lower Austria; Becker, 1996). Almost all other pyroxenites in ultramafic massifs are LREE depleted and have much higher HREE contents (e.g. Downes et al., 1991). Cabo Ortegal pyroxenites have very low HREE contents, LREE enrichment and higher contents of Zr, Nb and U. Although whole-rock equivalents can hardly be found, a close similarity may be noted with the composition of REE-enriched clinopyroxenes from peridotite xenoliths of Hungary and the French Massif Central (Downes & Dupuy, 1987; Downes et al., 1992).
|
ISOTOPE ANALYTICAL METHODS
|
|---|
Mineral separates were obtained by processing >10 kg samples
through crushing, disc-mill, Wilfley table, Frantz magnetic
separator and heavy liquids. Each mineral fraction was selected
grain by grain, including crack- and inclusion-free grains to
the highest transparency possible. Before dissolution all fractions
were leached following a two-step process (modified after Polvé,
1983
), using first weak HF (2%) at room temperature (5 min),
then ultrasound (5 min), HCl (2N) at 100–120°C (30
min), ultrasound (5 min), and finally triple distilled water
at 100–120°C (30 min). Samples were dissolved in HF
(48%) and, if necessary, the residue was treated in Teflon bombs
at 220°C, before processing with HCl. U–Pb and Rb–Sr
isotope analyses were performed at the Laboratory of Geochronology
of the University of Paris-VII with a Thompson 206 mass spectrometer
with
87Sr/
86Sr = 0·71025 ± 4 (2
,
n = 33) for the
Sr standard NBS-987, whereas Sm–Nd analyses were performed
at the University of the Basque Country using a Finnigan MAT262
system with
143Nd/
144Nd = 0·5118435 ± 13 (2
,
n = 18) for La Jolla standard.
For U and Pb analysis, isotope dilution methods (208Pb–235U–233U mixed isotope tracer) through aliquot solutions were utilized following a procedure modified after Manhès et al. (1987). Pb and U were loaded together on the same Re filament using Si gel and H3PO4 (Schärer & Gower, 1988). Mass discrimination for U and Pb was 0·1% per a.m.u. on both Faraday and secondary electron multiplier systems, and total blanks were 15 pg for Pb and <1 pg for U. Rb and Sr analysis was carried out with an 87Rb–84Sr mixed isotope tracer in aliquot solutions. Chemical procedures for Rb–Sr extraction and separate loading with a Ta mixture on Re filaments were described by Birck (1986). Isotope ratios were measured with a secondary electron multiplier for Rb and Sr concentrations, and a double Faraday collector for Sr composition. Blanks for Rb and Sr are negligible. Sm and Nd isotope analysis was carried out using a 150Nd/149Sm mixed isotope tracer. Chemical procedures for Sm and Nd extraction were described by Pin et al. (1994) and Pin & Santos Zalduegui (1997). Nd and Sm were loaded over Ta filaments with H3PO4–HNO3 (Platzner, 1997) and measured with a Faraday collector in multicollection mode. Blanks for Sm and Nd were negligible. The isochrons were regressed using the Isoplot/Ex program of Ludwig (1999).
|
Pb–Pb AND U–Pb ISOTOPE RESULTS
|
|---|
Three groups can be distinguished on the basis of measured Pb
isotopic composition (Table
4, Fig.
6a):
View this table:
[in this window]
[in a new window]
|
Table 4: Pb/Pb and U–Pb data for whole rocks and mineral separates of pyroxenite, peridotite and mafic rocks from Cabo Ortegal
|
|
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 6. (a) Measured Pb isotopic compositions of minerals and whole rocks from Cabo Ortegal compared with models of Pb growth (Stacey & Kramers, 1975; Zartman & Doe, 1981). UC, upper crust; AC, average crust; M, mantle. Fields: MORB, OIB, EMII and HIMU (Zindler & Hart, 1986; Hart, 1988); Cabo Ortegal pyroxenites (Gravestock, 1992); feldspars from pegmatitic granite in peridotite of Cabo Ortegal (Santos et al., 1996); Mediterranean ophiolites (Hamelin et al., 1984). Clinopyroxenes from pyroxenite GRCP6 are drawn separately to show apparent age correlation (see text for details). (b) Pbi isotopic compositions of minerals from Cabo Ortegal corrected for in situ decay after 390 Ma. Fields: feldspars from pegmatitic granite in peridotite of Cabo Ortegal and Mediterranean ophiolite group as in (a). Arrows for plume (HIMU), depleted mantle and ancient sediments components from Downes (2001).
|
|
- a radiogenic Pb-rich group represented by clinopyroxenes from samples GW7692 and GRCP6, whose relatively high µ values (>15) are in accord with the presence of U-rich phases such as zircon and rutile, respectively, in the rock (Peucat et al., 1990; Santos et al., 1996), and orthopyroxenes from H3. The measured Pb isotopic composition of these minerals is more radiogenic than EMII values and falls within the ocean island basalt (OIB) array. The anomalously high radiogenic Pb isotopic composition of fraction Grt-2 from sample GRCP6 (Table 4) could be due to the presence of clinopyroxene or minute rutile inclusions in garnet.
- Less radiogenic clinopyroxenes, garnet, amphibole and whole-rock data with µ values generally of 2–6·5 (Table 4). These are equivalent to previous results by Gravestock (1992) for analogous rocks of Cabo Ortegal. The data are similar to those of circum-Mediterranean ophiolites (Hamelin et al., 1984) or those of less radiogenic OIB and are intermediate between MORB and EMII values (Fig. 6a).
- Zoisite analyses yielded µ values of 2–5·7 but significantly lower ratios of 207Pb/204Pb than other minerals analysed, having a composition close to present-day MORB (Table 4, Fig. 6a).
Initial Pb isotopic compositions calculated at 390 Ma (U–Pb rutile ages from sample GRCP6, Santos et al., 1996) produced significant modifications only in the minerals with higher µ values, which tend to approach a common initial value for all minerals (except zoisite, Table 4, Fig. 6b). In this way, Pb isotopic compositions of garnets might correspond to initial values, as the measured U concentration was below detection limit. However, the difference from the initial Pb isotope ratios of clinopyroxenes (Table 4, Fig. 6b) suggests either that there was some U, or that the garnet initial composition was different from that of clinopyroxene within the same rock, which may be related to a metamorphic origin of the garnet. Orthopyroxenes from harzburgites have the highest initial Pb isotopic ratios (206Pb/204Pb = 18·286–18·426), clinopyroxene and amphibole of mafic rocks and associated pyroxenites show more scatter and have the lowest initial Pb isotopic ratios (206Pb/204Pb = 17·91–18·17), whereas clinopyroxene and amphibole of massive pyroxenites have intermediate values (206Pb/204Pb = 18·05–18·3) (Fig. 6b).
An attempt to derive geochronological information from the Pb–Pb and U–Pb isotope data provides two types of results. Clinopyroxenes from GRCP6, excluding the strongly deviating Cpx-6 fraction of slightly turbid grains, are the only samples to define a trend on a 206Pb/204Pb vs 207Pb/204Pb diagram (Fig. 6a). The trend corresponds to an age range of 366 Ma (Cpx1 to Cpx5) to 497 Ma (Cpx1 to Cpx4) Ma; however, errors in excess of 150 my render the calculated age meaningless. On the other hand, regression of data with known µ values yields considerably more consistent results in the 206Pb/204Pb vs 238U/204Pb diagram. A best-fit line through all the data points (20 mineral analyses) results in a calculated age of 400 ± 26 Ma [mean square weighted deviation (MSWD) = 3·4, Fig. 7a]. This result is barely modified if the three zoisites from mafic rocks, which show some evidence of Pb isotope disequilibrium, or the two values for clinopyroxenes of the Uzal massif, are excluded (394 ± 29 Ma). These ages are consistent with previous U–Pb data for the study area (Peucat et al., 1990; Santos et al., 1996; Ordóñez Casado et al., 2001) and the Sm–Nd ages obtained on the same samples (see next section).
|
Rb–Sr AND Sm–Nd ISOTOPE RESULTS
|
|---|
Rb–Sr and Sm–Nd elemental and isotopic compositions
from whole rock and separated minerals are presented in Table
5. The Sr content of clinopyroxene from massive pyroxenites
(except GW1 with the lowest values) is different from that of
clinopyroxenes from the mafic rocks and associated pyroxenites
(142–177 and 67–81 ppm, respectively). After correction
to 390 Ma, the initial
87Sr/
86Sr ratio is also, as in the case
of Pb isotopic data, different for both types of rock: 0·7044–0·7048
and 0·7040, respectively. Nevertheless, in both cases
87Sr/
86Sr
i ratios are higher than those of present-day N-MORB
(0·7020–0·7030) suggesting the participation
of a component with higher
87Sr/
86Sr ratio compared with mantle
values in the genesis of these rocks.
Nd values for whole rocks and clinopyroxene from clinopyroxenites, age corrected to 390 Ma, have a limited range of +2·4 to +4·9. Sm-poor orthopyroxenite OP10 yields an even lower value of -1·7, very different from the maximum value obtained of +6·8, corresponding to the harzburgite H3. A plot of Nd vs 87Sr/86Sr ratios for whole rocks and clinopyroxenes at 390 Ma shows a tendency towards higher 87Sr/86Sr ratios with respect to the ‘palaeo-mantle array’ (Fig. 8). These results are consistent with previous data for the ultramafic rocks of Cabo Ortegal (Peucat et al., 1990; Gravestock, 1992) represented in Fig. 8. A difference with respect to associated eclogites (interpreted as ancient N-MORB rocks) and a tendency towards EM (enriched mantle) characteristics, as inferred also from the Pb compositions (Fig. 6a), is also observed. Other minerals analysed exhibit even higher 87Sr/86Sr compositions (garnets of sample GRCP6) or lower Nd (amphibole and zoisite) compared with clinopyroxene and whole-rock values (Fig. 8). The new Rb–Sr results do not allow for calculation of a whole-rock isochron either for the pyroxenites or peridotites, even if previously published data are included. A best-fit line, highly dependent upon one Rb-rich phlogopite composition (Van Calsteren et al., 1979), is close to 390 Ma (Fig. 7b).
In the case of Sm–Nd, selected whole-rock data for pyroxenites excluding orthopyroxenite OP10 with the lowest Sm content, and adding data for similar rocks from Gravestock (1992), except one websterite with the greatest errors, result in an errorchron at 493 ± 130 Ma (Nd = +4, Fig. 9a). Sm–Nd isotope data for clinopyroxene from pyroxenites provide an equivalent age of 506 ± 91 Ma (Nd = +4·1), whereas a better result is obtained by combining whole-rock and clinopyroxene data yielding 506 ± 67 Ma (Nd = +4, Fig. 9b).
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 9. Sm–Nd isotopic correlation diagrams for minerals and whole rocks of Cabo Ortegal. (a) Whole-rock data. (b) The same as (a) including clinopyroxene data from this work. (c) Internal isochron (whole-rock–clinopyroxene–garnet) for sample GRCP6. (d) The same for sample GZC7. (e) Two-point line fits for sample GZA2. (f) Regression line for the three garnet fractions.
|
|
These results contrast with the more precise internal isochron (whole-rock–garnet–clinopyroxene) of 394 ± 22 Ma (Nd = +4·4, Fig. 9c) for clinopyroxenite sample GRCP6, which is in good agreement with the previous U–Pb age of 383 ± 1 Ma for rutile from the same sample (Santos et al., 1996). The spatially associated sample GZC7 provides a similar though less precise internal (whole-rock–garnet–clinopyroxene) age of 392 ± 69 Ma (Nd = +2·6, Fig. 9d). Amphibole-rich sample GZA2 gives 384 Ma for the garnet–amphibole pair and a distinctly younger age of 345 Ma for garnet and zoisite, in agreement with the previous suggestion of a late origin for zoisite (Fig. 9e). Finally, the three garnet fractions analysed yielded an errorchron at 413 ± 130 Ma with slightly lower Nd values of +1·8 (Fig. 9f).
|
DISCUSSION
|
|---|
Mineral data
The new major element mineral chemical data suggest an arc-related
origin for ultramafic units in Cabo Ortegal, reveal their similarity
with Bragança (Bridges
et al., 1995
) and exclude a purely
oceanic setting (Figs
2–
4). Moreover, the harzburgites
and dunite–pyroxenite units appear to have different origins.
In a plot of fo-number vs NiO, olivines from both dunites and pyroxenites deviate from the ‘mantle olivi
TOP
ABSTRACT
INTRODUCTION
