Journal of Petrology Advance Access originally published online on February 21, 2006
Journal of Petrology 2006 47(5):929-964; doi:10.1093/petrology/egi101
© The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org
Petrology and Geochemistry of Eclogite Xenoliths from the Colorado Plateau: Implications for the Evolution of Subducted Oceanic Crust
TOMOHIRO USUI1,*,
EIZO NAKAMURA1 and
HERWART HELMSTAEDT2
1 PHEASANT MEMORIAL LABORATORY FOR GEOCHEMISTRY AND COSMOCHEMISTRY, INSTITUTE FOR STUDY OF THE EARTH'S INTERIOR, OKAYAMA UNIVERSITY AT MISASA, TOTTORI, 682-0193, JAPAN
2 DEPARTMENT OF GEOLOGICAL SCIENCES AND GEOLOGICAL ENGINEERING, QUEEN'S UNIVERSITY, KINGSTON, ONT., K7L 3N6, CANADA
RECEIVED
DECEMBER 9, 2004;
ACCEPTED
DECEMBER 19, 2005
 |
ABSTRACT
|
|---|
Eclogite xenoliths from the Colorado Plateau, interpreted as
fragments of the subducted Farallon plate, are used to constrain
the trace element and SrNdPb isotopic compositions
of oceanic crust subducted into the upper mantle. The xenoliths
consist of almandine-rich garnet, Na-clinopyroxene, lawsonite
and zoisite with minor amounts of phengite, rutile, pyrite and
zircon. They have essentially basaltic bulk-rock major element
compositions; their Na
2O contents are significantly elevated,
but K
2O contents are similar to those of unaltered mid-ocean
ridge basalt (MORB). These alkali element characteristics are
explained by spilitization or albitization processes on the
sea floor and during subduction-zone metasomatism in the fore-arc
region. The whole-rock trace element abundances of the xenoliths
are variable relative to sea-floor-altered MORB, except for
the restricted Zr/Hf ratios (36·937·6).
Whole-rock mass balances for two Colorado Plateau eclogite xenoliths
are examined for 22 trace elements, Rb, Cs, Sr, Ba, Y, rare
earth elements, Pb, Th and U. Mass balance considerations and
mineralogical observations indicate that the whole-rock chemistries
of the xenoliths were modified by near-surface processes after
emplacement and limited interaction with their host rock, a
serpentinized ultramafic microbreccia. To avoid these secondary
effects, the Sr, Nd and Pb isotopic compositions of minerals
separated from the xenoliths were measured, yielding 0·704530·70590
for
87Sr/
86Sr, 3·1 to 0·5 for

Nd and 18·92819·063
for
206Pb/
204Pb. These isotopic compositions are distinctly
more radiogenic for Sr and Pb and less radiogenic for Nd than
those of altered MORB. Our results suggest that the MORB-like
protolith of the xenoliths was metasomatized by a fluid equilibrated
with sediment in the fore-arc region of a subduction zone and
that this metasomatic fluid produced continental crust-like
isotopic compositions of the xenoliths.
KEY WORDS: Colorado Plateau; eclogite xenolith; geochemistry; subducted oceanic crust
 |
INTRODUCTION
|
|---|
Material recycling through subduction zones is one of the most
important processes controlling the chemical evolution of the
Earth, and materials from the Earth's surface are likely to
have been reintroduced into its interior throughout geological
history. Although most of the subducting plate is made of mantle
material returning to depths from which it originated, hydrated
and altered upper oceanic crust, as well as overlying sediments,
carry a record of low-temperature interactions with continents,
oceans and the atmosphere into the deep mantle. Subduction of
surface and near-surface components can change the volatile
contents, trace element abundances, amounts of heat-producing
elements and radiogenic isotope systematics in the mantle.
A number of geochemical models have been developed to estimate the integrated trace element and isotopic composition of a hypothetical slab-derived component in arc and ocean island basalt (OIB) lavas (e.g. Zindler & Hart, 1986
; McCulloch & Gamble, 1991
; Ishikawa & Nakamura, 1994
; Hofmann, 1997
; Taylor & Nesbitt, 1998
; Stracke et al., 2003
). However, contributions from individual sources are difficult to assess from lava compositions alone, because element fractionations during slab dehydration and mass fluxes from the various subducted lithologies are poorly constrained. Nevertheless, most models have used a geochemical component estimated from rock compositions prior to subduction (e.g. samples from ocean-floor drilling). Therefore, understanding the trace element and isotopic compositions of individual rock types subducted into the sub-arc mantle is indispensable for constraining geochemical models for subduction-zone material recycling.
One approach to constrain material cycling involves high-pressure experimental studies that simulate processes such as dehydration of the subducting slab under controlled conditions (e.g. Schmidt & Poli, 1998
; Okamoto & Maruyama, 1999
; Poli & Schmidt, 2002
). Such experiments provide important constraints on the water contents and metamorphic phase relations in the subducting slab as a function of temperature and pressure. Other experiments have investigated the solubility of elements in fluids at high pressures, as well as element partitioning between the fluids and the constituent minerals in potential slab materials such as eclogites, blueschists and serpentinites (e.g. Brenan et al., 1994
, 1995
, 1998
; Iizuka & Nakamura, 1995
; Ayers et al., 1997
; Kogiso et al., 1997
; Stalder et al., 1998
). However, microanalytical and mass balance studies of subduction-related metamorphic rocks have shown that many so-called trace elements are preferentially partitioned into accessory minerals such as allanite, rutile, apatite, and zircon, and that elemental partitioning commonly does not reach equilibrium because of the relatively low-temperature environment envisaged in the subducting oceanic lithosphere (e.g. Sorensen & Grossman, 1989
, 1993
; Sorensen, 1991
; Tribuzio et al., 1996
; Zack et al., 2002a
, 2002b
; Spandler et al., 2003
). This suggests that partition coefficient data for trace elements may not be appropriate for evaluating trace element behaviors in subducting oceanic crust.
Thermomechanical studies indicate that subducting oceanic crust in cold subduction zones passes through the blueschist to lawsonite eclogite transition, and that the most important dehydration process for inducing island arc volcanism occurs in the lawsonite eclogite facies (Peacock, 1993
). Thus, lawsonite eclogites are critical to investigating trace element and isotopic behavior in subducted oceanic crust. However, data for trace element distributions among the mineral assemblages in the lawsonite eclogite facies are limited, because lawsonite eclogite terranes are rare (Zack et al., 2004
), and commonly much overprinted by high-T/P retrogression.
Lawsonite-bearing eclogite xenoliths from the Colorado Plateau of the southwestern USA present an important opportunity to study the trace element and isotopic systematics of this lithology. Helmstaedt & Doig (1975)
showed that these xenoliths resemble eclogites of the Franciscan Complex of the California Coast Ranges, western USA, in many aspects of their petrography and geochemistry. The Franciscan Complex represents fragments of a fossil subduction-zone plate boundary from the Jurassic to the Paleogene (e.g. Ernst, 1970
; Maruyama & Liou, 1989
). Offshore magnetic anomalies demonstrate that thousands of kilometers of oceanic crust have been subducted beneath the western US continental margin since the Late Cretaceous (Hamilton, 1969
). Usui et al. (2003)
used ion microprobe techniques to determine the UPb ages of zircons from the Colorado Plateau eclogite xenoliths, which yielded concordant ages from 81 to 33 Ma. Those workers interpreted the two extremes in age to reflect zircon crystallization during subduction-related metamorphism and zircon recrystallization during intrusion of the host rock. Combining geotectonic reasoning with the zircon geochronology and petrographic similarities to subduction-related eclogites led Usui et al. (2003)
to conclude that the Colorado Plateau eclogite xenoliths originated as fragments of the subducted Farallon plate that had resided in the upper mantle since the Late Cretaceous. Owing to their rapid transport to the surface, the eclogite xenoliths show few effects of retrograde overprinting compared with crustal eclogites exhumed by tectonic processes. Thus, they may preserve important chemical information about their subduction-related metamorphic history.
Two problematic issues remain concerning the origin of the Colorado Plateau eclogite xenoliths. One is whether they represent fragments of relatively young Phanerozoic subducted oceanic crust (Helmstaedt & Doig, 1975
; Usui et al., 2003
) or remnants of the much older, Proterozoic basement of the Colorado Plateau (Roden et al., 1990
; Wendlandt et al., 1993
; Smith et al., 2004
). The contrasting hypotheses for their ages may be considered to result, in part, from the application of different geochronological methods to date them. UPb zircon ages by ion microprobe analysis have yielded Phanerozoic ages (Usui et al., 2003
), whereas SmNd isotopic data (Roden et al., 1990
; Wendlandt et al., 1993
) give Proterozoic model ages in the range of 15001800 Ma. In addition, UPb analyses of multigrain zircon fragments by the isotope dilution method with thermal ionization mass spectrometry (ID-TIMS) resulted in a linear discordant line that intersected the concordia at 1514 Ma (Smith et al., 2004
).
The other debated issue is whether the Colorado Plateau eclogite xenoliths were formed by prograde metamorphism during subduction of oceanic lithosphere or by magmatic recrystallization processes in the upper mantle. Based on mineral textures and compositional zoning of garnet and clinopyroxene, Helmstaedt & Schulze (1988
, 1991
) concluded the former. They estimated two metamorphic temperatures for the eclogite xenoliths, corresponding to crystallization of garnet cores and peak metamorphic conditions, respectively, and concluded that the garnet rims formed at
100°C higher temperatures than the garnet cores. However, these estimates are unlikely to be conclusive because: (1) they depend on calculating the Fe2+/Fe3+ of clinopyroxene, a parameter that cannot be determined by electron microprobe; (2) it is difficult to determine the clinopyroxene coregarnet core temperature by using clinopyroxene with an oscillatory zoned structure. Smith & Zientek (1979)
suggested that the oscillatory zoning of clinopyroxene may represent disequilibrium growth in the presence of fluid phases, and that the Colorado Plateau eclogite xenoliths formed coexisting with a fluid phase during cooling and metasomatism of basaltic dikes in a cool upper mantle.
In this paper, we report whole-rock and mineral data for major, minor, and trace elements, and Sr, Nd, and Pb isotopic data for a suite of Colorado Plateau eclogite xenoliths. The geochemical data are complemented by observations of the modes of constituent minerals, their textures and chemical zoning structures. We reassess the contrasting hypotheses for the origin of the Colorado Plateau eclogite xenoliths and demonstrate that these rocks can be used as proxies to understand the geochemical evolution of subducted oceanic crust. These data allow us to develop a model for the trace element and isotopic (Sr, Nd and Pb) composition of subducted oceanic crust in the sub-arc region.
 |
GEOLOGICAL SETTING
|
|---|
The Colorado Plateau is a broadly elevated region,

1500 m above
sea level, in the southwestern interior of the USA. Unmetamorphosed
and little deformed Paleozoic to Cenozoic sedimentary rocks
are well exposed on Precambrian basement (1·41·8
Ga; Condie, 1982

). The plateau is surrounded by the Basin and
Range, Rocky Mountains, and the Rio Grande Rift tectonic provinces,
all of which have experienced intense Cenozoic orogenic activity
and extensional tectonics (
Fig. 1a). The structure of the Colorado
Plateau consists of broad basins, uplifts and platforms, which
are locally bordered by monoclines. Tectonic reconstruction
of the history of the western margin of North America suggests
that during the Late Cretaceous and Early Tertiary, the Farallon
plate was subducting eastward (Atwater, 1970

). The Late Cretaceous
and Tertiary records of arc magmatism in the southwestern USA
constrain the slab geometry and its evolution, suggesting that
the migration of arc magmatism was probably caused by progressive
flattening of a subducting slab (Coney & Reynolds, 1977

).
Intra-plate magmatism, which occurred extensively around the
Colorado Plateau, may be associated with the formation of a
slab window and/or slab tearing (Dickinson, 1997

).

View larger version (17K):
[in this window]
[in a new window]
|
Fig. 1. (a) Sketch map of the southwestern interior of the USA and outline of the Colorado Plateau (dotted line). (b) Sketch map of the Navajo Volcanic Field in the Colorado Plateau near the Four Corners shown as the gray shaded area in (a), modified after Smith & Levy (1976) . , minettes; , serpentinized ultramafic microbreccia diatremes. Major monoclines with dip directions are also shown. ME, Mule Ear; MR, Moses Rock; CN, Cane Valley; GR, Garnet Ridge; RM, Red Mesa; BP, Buell Park; GK, Green Knobs; SF, San Francisco; LA, Los Angeles.
|
|
Eight diatremes of serpentinized ultramafic microbreccia and
numerous minette diatremes, the products of intra-plate magmatism,
are exposed in the Navajo Volcanic Field of the Colorado Plateau
(
Fig. 1b). The microbreccia diatremes occur as two structurally
controlled clusters (McGetchin
et al., 1977

). One, related to
the Comb Ridge monocline on the eastern margin of Monument uplift,
includes the Mule Ear, Moses Rock, Cane Valley, and Garnet Ridge
diatremes. The other, at the eastern margin of the Defiance
uplift, includes the Buell Park and Green Knobs diatremes. A
single diatreme, Red Mesa, lies between the two fields and is
located about 30 km east of the Comb Ridge monocline.
The diatreme rocks that host the eclogite xenoliths lack primary igneous textures and consist of finely crushed fragments of strongly serpentinized ultramafic rocks. Although initially described as kimberlites (e.g. McGetchin et al., 1977
), these rocks are not true kimberlites (e.g. Clement et al., 1984
) but serpentinized ultramafic microbreccias (SUM; Roden, 1981
). The microbreccia diatremes were emplaced between 25 and 35 Ma (Naeser, 1971
; Helmstaedt & Doig, 1975
; Roden et al., 1979
), contemporaneous with the numerous minette diatremes of the Navajo Volcanic Field (Ehrenberg, 1982
). The diatremes contain several types of eclogite, as well as a broad spectrum of possible upper mantle rocks that include variously hydrated spinel- and garnet-bearing peridotites, pyroxenites and serpentinites. The eclogite xenoliths are found mainly in three microbreccia diatremes in the Comb Ridge cluster (Garnet Ridge, Moses Rock, Mule Ear), and are absent from the Green Knobs and Buell Park pipes near the Defiance uplift (McGetchin et al., 1977
; Selverstone et al., 1999
). The samples used for this study were collected from the Moses Rock and Garnet Ridge diatremes.
 |
ANALYTICAL METHODS
|
|---|
Major element compositions of minerals were determined using
a JEOL JXA-8800 electron microprobe at ISEI (Institute for Study
of the Earth's Interior). Analyses were carried out at 15 kV
and 20 nA sample current for all silicate minerals. Counting
times of 30 s were used, and matrix corrections were performed
by ZAF methods. The standards are synthetic oxides and natural
minerals. The modal abundance of zircon was determined by digital
image analysis of 25 randomly selected back-scattered electron
images (each 1000 µm) per sample.
Trace element analyses of garnet, clinopyroxene, lawsonite and zoisite were performed using a CAMECA ims-5f ion microprobe at the PML (Pheasant Memorial Laboratory) in the ISEI, using techniques described by Nakamura & Kushiro (1998)
. Clinopyroxene from mantle xenoliths and basaltic natural glass were used as standard minerals for trace element calibrations. Trace element concentrations in these standards were chemically determined by inductively coupled plasma mass spectrometry (ICP-MS; Makishima & Nakamura, 1997
). The homogeneity of these standards was confirmed by optical microscope, electron microprobe and ion microprobe analysis. Trace element contents of garnet and clinopyroxene were calibrated on the basis of the clinopyroxene standard, and those of lawsonite and zoisite were calibrated using the basaltic glass standards. Minerals in thin sections were sputtered with an O primary beam of 1020 nA intensity, resulting in 1020 µm beam diameter. Clinopyroxene and lawsonite inclusions <30 µm in diameter were measured by an O primary beam of
6 nA intensity, which produced a beam <10 µm in diameter. Positive secondary ions were collected by ion counting using an energy offset of 60 V from 4500 V acceleration with an energy bandpass of ±10 V. These operating conditions resulted in (11·5) x 105 c.p.s. for 30Si secondary ions. The secondary ion intensities of atomic masses of interest were normalized to 30Si. For each analysis spot, 15 trace elements were determined in a run that took
40 min, including presputtering. The analytical reproducibility (relative standard deviation % for n = 10) for trace element measurements is generally <10% for the clinopyroxene standard and <5% for the basaltic glass standard, except for Er and Lu (710%).
To prepare whole-rock powders for analysis, eclogite samples were broken into chips
2 mm in diameter. Fresh pieces were hand-picked and washed with deionized water in an ultrasonic bath for 30 min. After drying at 60°C overnight, the chips were pulverized using a silicon nitride triturator to a grain size <400 mesh. To separate minerals, rock powders obtained using the silicon nitride triturator were sieved to capture grain sizes between 80 and 100 mesh (an average grain size of
200 µm). This sieved fraction was processed with an isodynamic separator, and the resulting mineral fractions were hand-picked.
Whole-rock major element compositions, along with Cr and Ni, were determined by X-ray fluorescence (XRF) spectrometry at the PML, on glass beads fluxed from mixing 100 mg of powdered sample and 5 g of lithium tetraborate (Takei, 2002
). Trace element compositions were measured by ICP-MS at the PML, based on the methods of Makishima & Nakamura (1997)
for Y, Cs, Ba, rare earth elements (REE), Pb, Th and U, those of Makishima et al. (1999)
for Zr, Hf and Nb, and of Moriguti et al. (2004)
for Li. Powdered samples (20 mg) for ICP-MS analysis were decomposed in 1·0 ml of HF and 0·1 ml of HNO3 in Teflon bombs at 210°C for 2 days, because acid-resistant minerals such as zircon and rutile are present in the eclogite samples. Analytical errors for trace element were generally 3% for Zr, Hf, and Li, and 5% for other elements.
The analytical procedures for mass spectrometry followed Yoshikawa & Nakamura (1993)
for Sr isotopic ratios and abundances of Rb and Sr, Makishima & Nakamura (1991)
for Nd isotopic ratios and abundances of Sm and Nd, Kuritani & Nakamura (2002
, 2003
) for Pb isotopic ratios and Pb abundances, and Yokoyama et al. (1999
, 2003
) for abundances of U and Th, employing TIMS in static multi-collection mode [modified Finnigan MAT261: Kiji (Nakano & Nakamura, 1998
), MAT262: Taro and MAT262: INU]. All of these elements (Rb, Sr, Sm, Nd, U, Th and Pb) were successively separated from each of the whole-rock and mineral samples, using a multi-ion exchange column chemistry approach modified after Nakamura et al. (2002)
.
Basaltic standard JB3 from the Geological Survey of Japan (GSJ) yields typical analytical reproducibility of 0·005%, 0·005% and 0·02% for Sr, Nd and Pb isotopic ratios, respectively, and <1% for Rb, Sr, Sm, Nd, U, Th and Pb abundances. Accuracies of the isotopic and concentration analyses were confirmed by repeated measurement of JB3. We obtained data for this standard consistent with the previously published data within the analytical reproducibility (see Appendix, Table A1). Isotopic fractionation during analysis was corrected using 86Sr/88Sr = 0·1194 and 146Nd/144Nd = 0·7219 as normalizing factors. Pb isotopic fractionation was corrected by the two-double spikes method with 205Pb204Pb and 207Pb204Pb enriched spikes following the technique of Kuritani & Nakamura (2003)
. The composition of the spike was calibrated by assuming that the 208Pb/206Pb ratio of NBS982 is 1·00016 (Catanzaro et al., 1968
). Instrumental mass discrimination of TIMS analyses was corrected by the following values of standards: 87Sr/86Sr = 0·71024 (NIST987), 143Nd/144Nd = 0·511839 (La Jolla), and 206Pb/204Pb = 16·9424, 207Pb/204Pb = 15·5003, 208Pb/204Pb = 36·7266 (NBS981).
 |
RESULTS
|
|---|
Mineral assemblage and classification of eclogite xenoliths
The Colorado Plateau eclogite xenoliths investigated in this
study were previously described and referred to as metabasic
eclogite and jadeite-clinopyroxenite, based not only on their
mineral assemblages and basaltic bulk compositions, but also
on their textural similarity to crustal examples of eclogites
(Helmstaedt & Schulze, 1988

). In this paper, jadeitic-clinopyroxenite
is used instead of jadeite-clinopyroxenite, because clinopyroxene
in this xenolith type is not pure jadeite, but omphacite or
impure jadeite as described later. The mineral assemblages of
the xenoliths and their modal proportions are given in
Tables 1 and
2.
Metabasic eclogites consist mainly of various proportions of
almandine-rich garnet and omphacitic clinopyroxene with some
lawsonite and zoisite (
Fig. 2). They are either lawsonite-eclogite
or zoisite-eclogite, according to the major constituent mineral
assemblages (
Table 1). Zoisite-eclogite is distinguished from
lawsonite-eclogite only by its lack of lawsonite, although trace
amounts of lawsonite crystals occur as small (

10 µm) inclusions
in garnet in zoisite-eclogites (
Fig. 3c). With a decrease in
garnet content (approximately <10 vol. %) and an increase
of the jadeite component in the clinopyroxene (approximately
>70 mol %), the metabasic eclogites grade continuously into
jadeitic-clinopyroxenites. Other minerals that occur in trace
amounts in both xenolith types are phengite, pyrite, rutile,
zircon and apatite. Rare chlorite, albite and barite occur as
<5 µm crystals around garnet and clinopyroxene. In
both eclogite types, coesite occurs as <20 µm inclusions
in garnet crystals, but not in the matrix (
Fig. 3a; Usui
et al., 2002

, 2003

).

View larger version (113K):
[in this window]
[in a new window]
|
Fig. 2. (a), (b) Back-scattered electron (BSE) image of lawsonite-eclogite GR1. (c) BSE image of included clinopyroxene and lawsonite in GR1. A crack in the lower part of the garnet connects the lawsonite pseudomorphs and an included clinopyroxene to the exterior of the garnet. (d) Photomicrograph (plane-polarized light) of GR1 showing curved trails of inclusions (rutile and coesite) in garnet. Abbreviations are as in Table 1.
|
|

View larger version (65K):
[in this window]
[in a new window]
|
Fig. 3. (a) BSE image of zoisite-eclogite MR7 showing that garnet includes various kinds of inclusions, such as clinopyroxene, coesite, rutile and zircon. (b) High-contrast BSE image of annealed cracks (shown by small arrows) in garnet in zoisite-eclogite MR4. Major element compositions of annealed garnet are similar to that of the reaction rim (see text for details). (c) BSE image of zoisite-eclogite MR7 showing that lawsonite can still survive as inclusions in garnet. (d) BSE image of jadeitic-clinopyroxenite MR3B. Garnet (shown in Fig. 5) includes lawsonite pseudomorphs that are replaced by fibrous zoisite aggregates along with rutile. Cracks connect these lawsonite pseudomorphs. Ion-microprobe pits are indicated by small arrows. Abbreviations are as in Table 1.
|
|
Occurrence, textures and compositions of minerals
Garnet
Garnet commonly occurs in small clusters or in thin layers accompanied
by lawsonite and zoisite; these generally have euhedral to subhedral
outlines with maximum diameters of

0·11·0
mm (
Figs 2a and
3a). They are commonly color-zoned, from pink
cores to reddish brown rims under plane-polarized light (
Fig. 2d).
Some grains exhibit rotation textures and contain fine-grained
curved inclusion trails of coesite and rutile, indicating that
deformation continued after the development of a planar tectonic
fabric (
Fig. 2d) (see also Helmstaedt & Schulze, 1991

).
Garnets have compositions of Alm (Almandine)5070, Sps (Spessartine)<1, Pyr (Pyrope)1030, Grs (Grossular)20, except for locally developed pyrope-rich rims (Table 3 and Fig. 4). These pyrope-rich rims have distinctively different compositions from the other parts of the garnet, indicating a sharp compositional boundary (Fig. 2a). The compositions of these pyrope-rich rims are about Alm40Sps<1Pyr50Grs10 for the lawsonite-eclogite and Alm40Sps<1Pyr40Grs20 for the zoisite-eclogite. The rims may have formed under higher temperature conditions when the xenoliths were entrained in the microbreccia host rock, as will be discussed later, and are called reaction rims, to distinguish them from normal rims the composition of which is almost identical to those of the core and mantle (Table 4).

View larger version (26K):
[in this window]
[in a new window]
|
Fig. 4. Composition of garnets in (a) lawsonite-eclogite GR1, (b) zoisite-eclogite MR4 and (c) jadeitic-clinopyroxenite MR3B plotted in the system (Alm + Sps)PyrGrs. The shaded part of the full compositional triangle (d) is enlarged in (a), (b) and (c). Stippled pattern indicates garnets in eclogite xenoliths from diamondiferous kimberlites; cross-hatched pattern, garnets in eclogites from subduction-related metamorphic suites.
|
|
In some particularly zoisite-rich eclogites (e.g. MR4, see
Table 2),
wide reaction rims are present around most garnet crystals,
and also penetrate into the garnet cores along cracks (
Fig. 3b).
The chemical compositions of these garnets in the zoisite-rich
eclogites trend toward those of the reaction rims in the (Alm
+ Sps)PyrGrs diagram (
Fig. 4b). In contrast, no
reaction rims were observed in jadeitic-clinopyroxenites, which
have little or no lawsonite and zoisite, although some garnets
have slightly pyrope-rich compositions (
Figs 4c and
5a).

View larger version (82K):
[in this window]
[in a new window]
|
Fig. 5. (a) BSE image of garnet in jadeitic-clinopyroxenite MR3B. The garnet does not have any reaction rims. Lawsonite pseudomorphs preserve sharp and euhedral crystal boundaries of the original lawsonite, which are connected by cracks. (b) Electron microprobe X-ray map for Mn in the garnet shown in (a). White and red colors indicate higher concentration than green and blue.
|
|
The major element compositions of the garnet, except for the
reaction rims, are more similar to those from eclogites in subduction-related
metamorphic terranes (Low-Temperature eclogite) than to those
from diamondiferous kimberlite pipes (High-Temperature eclogite)
(Carswell, 1990

). The garnets show faint zoning in their Alm,
Pyr and Grs components but strong and euhedral zoning in their
Sps component, which is much lower than the other three components
(
Fig. 5). The core is characteristically enriched in Mn relative
to the rim.
Clinopyroxene
Clinopyroxene is generally prismatic, and its orientation defines the rock fabric. In thin section it varies in color from colorless to pale green under plane-polarized light. The grain size varies as a result of deformation and synkinematic recrystallization (Helmstaedt & Schulze, 1988
). Clinopyroxene grains in the eclogites have compositions near Jd (Jadeite)4050, Aug (Augite)5060, Ac (Acmite)08, falling into the range of omphacite compositions (Table 4, Fig. 6a and b). On the other hand, clinopyroxenes in the jadeitic-clinopyroxenites have compositions of Jd65Aug20Ac15 to Jd75Aug20Ac5. This variation suggests relatively linear jadeite to acmite substitution (Fig. 6c). Clinopyroxene occurs both as inclusions in garnet (referred as included clinopyroxene)and in the matrix (referred to as matrix clinopyroxene; Figs 2c and 3a). Included clinopyroxene grains are richer in Ac and poorer in Jd components than matrix clinopyroxene grains, and their composition varies linearly between Jd35Aug50Ac15 and Jd40Aug55Ac5 (Fig. 6a and b).This trend is also a jadeite to acmite substitution but it differs from the compositional variations of the matrix clinopyroxene.
Lawsonite and zoisite
Lawsonite is transparent in plane-polarized light. Except for
minute inclusions in garnet crystals in the zoisite-eclogite
(
Fig. 3c), lawsonite is restricted to lawsonite-eclogite. It
is invariably rimmed by fine-grained zoisite aggregates (
Fig. 2b).
Lawsonite grains display relatively uniform compositions close
to the ideal formula of CaAlSi
2O
7(OH)
2·H
2O, with limited
substitution of Fe
3+ for Al and (Na + K) for Ca (
Table 5; e.g.
Moore & Liou, 1979

). Like clinopyroxene, lawsonite occurs
both as inclusions in garnet (referred to as included
lawsonite) and as matrix grains (referred to as matrix
lawsonite) (
Fig. 2b and c). Included lawsonite grains
in unfractured garnet crystals lack any rims of zoisite aggregates
(
Fig. 3c). Most included lawsonites have higher Fe
2O
3 and lower
Al
2O
3 contents than matrix lawsonite (
Table 5 and
Fig. 7).

View larger version (17K):
[in this window]
[in a new window]
|
Fig. 7. XFe3+ of lawsonite in lawsonite-eclogite GR1 plotted against XAl. XFe3+ and XAl are the cation numbers of Fe3+ and Al in lawsonite, respectively, when the oxygen number is eight.
|
|
Zoisite invariably occurs as fine-grained, fibrous and radiating
crystal aggregates in eclogite and some jadeitic-clinopyroxenites
(
Figs 2 and
3). Whereas the zoisite crystals themselves are
transparent, the aggregates are dark brown in plane-polarized
light. The grain size of zoisite is generally <10 µm
along the major axis and <1 µm along the minor axis,
and is much smaller than that of the other constituent minerals,
garnet, clinopyroxene, lawsonite, phengite and rutile. It contains
39·740·2 wt % SiO
2, 32·632·3
wt % Al
2O
3, 23·624·2 wt % CaO, and trace
amounts of Fe
2O
3, MgO, TiO
2, Na
2O and Cr
2O
3 (
Table 5). The textures
and major element composition of zoisite are almost identical
in lawsonite-eclogite, zoisite-eclogite and jadeitic-clinopyroxenite.
Zoisite aggregates also occur as inclusions in fractured garnet
preserving the sharp crystal boundaries of euhedral lawsonite
pseudomorphs (
Fig. 3d), suggesting that the zoisite aggregates
originated by replacement of lawsonite.
Other minerals
Phengite occurs in one eclogite (GR1) and two jadeitic-clinopyroxenite (MR21 and MR3B) samples. Grains are platy and up to 0·5 mm in diameter, have relatively homogeneous major element compositions and are characterized by Si values of
7·5 p.f.u. (per formula unit, O = 22; Table 6). In jadeitic-clinopyroxenite samples, albite occurs as a secondary phase along clinopyroxene grain boundaries and rarely along cleavages. The mode of albite is <0·1 vol. %, and it contains <1 % of Or and An components. Orthoclase is present only as inclusions in the core of a garnet from eclogite MR7. Its composition is close to the ideal formula of KAlSi3O8 (Table 6). Amphibole does not occur in any of our metabasic eclogite xenoliths.
Trace element compositions of minerals
Garnets from eclogites and jadeitic-clinopyroxenites are characterized
by extreme LREE (light rare earth element) depletion and HREE
(heavy rare earth element) enrichment, exceeding a chondrite-normalized
Yb/Ce ratio of 10
4 (
Fig. 8). Garnets have REE patterns that
appear to increase progressively from La to Dy and are convex-upward
from Dy to Lu, except for the reaction rim. The REE patterns
of the reaction rims differ in shape from those in other parts
of the garnet; they are characteristically convex-upward from
La to Dy and concave-downward from Dy to Lu. LREE (La, Ce, Pr
and Nd) and MREE (middle rare earth elements: Sm, Eu, Gd and
Dy) contents are mostly uniform throughout the crystal, whereas
HREE (Er, Yb and Lu) contents are markedly zoned (
Fig. 8). The
HREE are more abundant in the core relative to the rim, and
show a strong correlation with Mn contents (
Fig. 9). Garnets
in all of the xenoliths are rich in Y, which decreases from
core to rim, similar to the HREE (
Table 7). Other trace element
data for garnet measured by ICP-MS are listed in
Table 10.

View larger version (26K):
[in this window]
[in a new window]
|
Fig. 8. Chondrite-normalized rare earth element patterns of garnet, clinopyroxene, lawsonite and zoisite in eclogites (GR1, MR4 and MR7) and jadeitic-clinopyroxenites (M26 and MR47). The rare earth element data are normalized to the C1 chondrite value of Sun & McDonough (1989) .
|
|

View larger version (17K):
[in this window]
[in a new window]
|
Fig. 9. Zoning profiles along traverse XY in Fig. 2a for HREE, Y and Mn (XSps) in the garnet. Concentrations of Dy, Er, Yb, Lu and Y were analyzed simultaneously by ion microprobe, and are normalized to the primitive mantle value of Sun & McDonough (1989) . The Mn content of the garnet next to the ion microprobe pits was analyzed by electron microprobe.
|
|
The clinopyroxenes show two groups of trace element compositions
(
Fig. 8). The LREE, MREE and Sr contents of clinopyroxenes are
higher in eclogite than jadeitic-clinopyroxenite. On the other
hand, the HREE and Y contents of clinopyroxenes in eclogite
are slightly lower than those of counterparts in jadeitic-clinopyroxenite
(
Table 8 and
Fig. 8). Clinopyroxene in eclogite has REE patterns
that are convex-upward from La to Dy and concave-downward from
Dy to Lu. Such HREE characteristics are only weakly developed
in clinopyroxene from the jadeitic-clinopyroxenites. In particular,
no HREE depletion is observed in jadeitic-clinopyroxenite MR47
that contains <0·1 vol. % of garnet. Clinopyroxenes
in both eclogite and jadeitic-clinopyroxenite are poor in HFSE
(Zr <1 ppm and Nb <0·001 ppm;
Table 8). Other trace
element data for clinopyroxene measured by ICP-MS are listed
in Table 10.
Lawsonite and zoisite are rich in incompatible trace elements,
especially U, Th, Pb, LREE and Sr. They are poor in high field
strength elements (HFSE), similar to clinopyroxene (
Tables 9 and
10). Matrix lawsonite from the lawsonite-eclogite sample
has trace element abundances similar to those of matrix zoisite
(
Fig. 8e). The matrix lawsonites appear unzoned in trace elements,
and the trace element abundances are almost identical among
the matrix lawsonite grains in each sample. However, matrix
and included lawsonites display different REE patterns. Matrix
lawsonites exhibit gently convex-upward LREE-enriched patterns
from La to Er and relatively unfractionated patterns from Er
to Lu. Included lawsonite grains have fairly flat REE patterns
that decease slightly from La to Lu (
Fig. 8). The trace element
patterns of zoisite grains in eclogite and jadeitic-clinopyroxenite
vary from sample to sample. For example, zoisite from apatite-bearing
zoisite-eclogite MR7 contains less LREE than other zoisites
(
Fig. 8). This may reflect equilibration with apatite that is
REE-rich (Usui, 2004

). Other trace element data for zoisite
measured by ICP-MS are listed in
Table 10.
Whole-rock chemistry
Major element whole-rock compositions (
Table 11) are plotted
against SiO
2 contents in
Fig. 10. Compositions of fresh and
altered mid-ocean ridge basalt (MORB) from ocean-floor drilling
holes 504B and 417D (Tual
et al., 1985

; Zuleger
et al., 1995

),
and subduction-related metamorphic rocks thought to originate
from MORB-type oceanic crust (Becker
et al., 1999

; Bröcker
& Enders, 2001

; Gao & Klemd, 2001

), are plotted for
comparison. The whole-rock major element compositions of the
eclogite and jadeitic-clinopyroxenite xenoliths do not lie in
the ranges of either altered or fresh MORB. Instead, they are
within the ranges of the subduction-related metamorphic rocks.
In particular, the xenoliths are characterized by distinctively
higher Na
2O contents with increasing SiO
2 contents than those
of altered MORB, resulting in a continuous and near-linear trend
on the SiO
2 vs Na
2O Harker diagram (
Fig. 10b). In contrast,
K
2O, FeO
*, Al
2O
3, and TiO
2 in both eclogites and jadeitic-clinopyroxenites
are almost identical to those of altered MORB. Such major element
characteristics are commonly observed in metabasic rocks from
subduction-related metamorphic terranes, and they have been
explained by spilitization during high-temperature hydrothermal
alteration at mid-ocean ridges and subduction-related metasomatism
at fore-arc depths (e.g. Zack
et al., 2003

). In particular,
high Na
2O contents that are decoupled from K
2O contents are
best explained by spilitizationalbitization (
Fig. 11).
The jadeitic-clinopyroxenites are richer in Na
2O and SiO
2 than
the eclogites, although there appears to be a continuous transition
and overlap between the two groups. This reflects the differences
in mineralogical composition and modal proportions of the major
constituent minerals: Jd components in the clinopyroxenes are
richer in the jadeitic-clinopyroxenites than those in the eclogites,
and the jadeitic-clinopyroxenites contain more clinopyroxene
but less garnet, lawsonite and zoisite than the eclogites (compare
Fig. 6 and
Table 2 with
Fig. 10).

View larger version (20K):
[in this window]
[in a new window]
|
Fig. 11. Whole-rock Na2O and K2O contents (wt %) in eclogite and jadeitic-clinopyroxenite xenoliths. The field of fresh MORB (data sources as in Fig. 10) and trends produced by spilitizationalbitization and low-T seawater alteration (see text for explanation) are also shown.
|
|
Whole-rock chondrite-normalized REE patterns of eclogite and
jadeitic-clinopyroxenite xenoliths are shown in
Fig. 12; altered
MORB from ocean-floor drilling hole 504B is shown as a shaded
field for comparison. Eclogites have LREE-enriched patterns
that decrease progressively from La to Lu, resulting in chondrite-normalized
abundance ratios [La/Lu]
N = 6·011 and [La/Sm]
N = 2·53·3 that are distinctively greater
than those of altered MORB (
Table 11). In addition to the higher
abundance ratios, LREE concentrations are higher in the eclogite
than in altered MORB. Jadeitic-clinopyroxenites display V-shaped
REE patterns with the lowest normalized abundances at Eu. Although
jadeitic-clinopyroxenites exhibit variable LREE contents, which
are lower than those of the eclogites, their chondrite-normalized
[La/Sm]
N ratios are relatively constant and similar to those
of the eclogite xenoliths. The REE patterns of eclogite and
jadeitic-clinopyroxenite xenoliths are probably controlled by
the modes of garnet and zoisite (or lawsonite), which have extremely
different REE characteristics (
Fig. 8). HFSE abundances are
almost identical between eclogite and jadeitic-clinopyroxenite.
The Zr/Hf and Nb/Ta ratios of the xenoliths vary from 36·9
to 37·6 and from 10·9 to 15·0, respectively.
Both fall in the range of altered MORB, although the contents
of these elements in the xenoliths are slightly higher than
those of altered MORB (David
et al., 2000

). The abundances of
other trace elements, especially Pb, Sr and U, vary considerably;
they are generally more abundant in eclogite than jadeitic-clinopyroxenite
(
Table 11).
SrNdPb isotopic compositions
Whole-rock Sr, Nd and Pb isotopic data were age-corrected to
30 Ma (
Table 12), the emplacement age of the host rock at Moses
Rocks and Garnet Ridge (Naeser, 1971

; Helmstaedt & Doig,
1975

; Usui
et al., 2003

), and are plotted on
87Sr/
86Sr

Nd,

Nd
206Pb/
204Pb,
206Pb/
204Pb
207Pb/
204Pb and
206Pb/
204Pb
208Pb/
204 diagrams (
Fig. 13). The isotopic compositions of Juan de Fuca
MORB and Phanerozoic Cordilleran crust, which represent Farallon
MORB and its sedimentary cover, respectively, are shown for
comparison.
The Sr isotopic ratios of the eclogites and jadeitic-clinopyroxenites
vary (
87Sr/
86Sr = 0·70500·7084) and are
greater than those of MORB. The jadeitic-clinopyroxenites have
much higher Sr isotopic ratios than the eclogites. In contrast,
the Nd isotopic compositions are similar in both eclogites and
jadeitic-clinopyroxenites;

Nd values vary from 3·1
to 1·2, much less than those of MORB, but similar
to those of crustally derived sediments. Eclogites and jadeitic-clinopyroxenites
have Pb isotopic compositions (
206Pb/
204Pb = 18·9119·38,
207Pb/
204Pb = 15·6215·68, and
206Pb/
204Pb
= 38·6838·92) that fall in the compositional
field of sedimentary rocks in
206Pb/
204Pb
207Pb/
204Pb
and
206Pb/
204Pb
208Pb/
204Pb diagrams (
Fig. 13).
In addition to the whole-rock data, RbSr, SmNd and UThPb isotopic studies were also performed on garnet, clinopyroxene and zoisite mineral separates from the xenoliths (Table 13). The Pb isotopic compositions of garnet were not measured because of its extremely low Pb contents (0·0380·087 ppm; see Table 10). Reference isochrons at 30 Ma were calculated to pass through each mineral in the 87Sr/86Sr87Rb/86Sr, 143Nd/144Nd147Sm/144Nd and 206Pb/204Pb238U/204Pb isotopic systems to distinguish isotopic equilibrium and disequilibrium at 30 Ma (Fig. 14). If one mineral was isotopically in equilibrium with another when the host diatremes were emplaced, the reference isochrons for the two minerals should be coincident within error. The slope of the reference isochron for the RbSr system is almost horizontal because of the small variation of Rb/Sr ratio.

View larger version (17K):
[in this window]
[in a new window]
|
Fig. 14. RbSr, SmNd and UPb isochron diagrams for constituent minerals in eclogite MR19 (a, c, and e) and jadeitic-clinopyroxenite MR26 (b, d and f) xenoliths. The lines marked in each figure are 30 Ma reference isochrons, which were calculated to pass through the three constituent minerals: G, garnet; C, clinopyroxene; Z, zoisite; W, whole-rock. Continuous line, zoisite; dotted line, clinopyroxene; dashed line, garnet. Vertical bar shows standard deviation (2 ) in Table A1.
|
|
Figure 14a and b shows RbSr isochron diagrams for representative
eclogite sample MR19 and jadeitic-clinopyroxenite sample MR26;
the diagrams include both mineral and whole-rock data. Both
whole-rock values of
87Sr/
86Sr are greater than those of the
constituent minerals, although the minerals in the eclogite
xenolith have almost identical Sr isotopic compositions. Individual
minerals in the jadeitic-clinopyroxenite have different isotopic
compositions, suggesting that they were not in equilibrium with
each other at 30 Ma.
The reference SmNd isochrons for all minerals and the whole-rock of eclogite are coincident within error (Fig. 14c), but garnet from the jadeitic-clinopyroxenite has a different Nd isotopic composition from the other minerals and the whole-rock (Fig. 14d). Zoisites display almost identical Sm/Nd abundance ratios and Nd isotopic compositions to those of the whole-rocks because they are the dominant reservoir for LREE in both eclogite and jadeitic-clinopyroxenite (as discussed later; see Fig. 15).

View larger version (37K):
[in this window]
[in a new window]
|
Fig. 15. Mass balance calculations for (a) eclogite MR19 and (b) jadeitic-clinopyroxenite MR26. Density-weighted modal abundances of minerals are calculated by multiplying the mode in vol. % in Table 2 by densities. The densities are interpolated from values of Deer et al. (1992) : 4·1 for Alm70Pyr30 garnet; 3·3 for omphacitic clinopyroxene; 3·15 for zoisite. Bar length corresponds to the amount of the trace element in each phase in relative proportion to the whole-rock concentration. If the total amount of the relative proportions of the minerals falls in the range 80120%, mass balances are judged to be achieved, considering uncertainties in modal proportions of minerals and heterogeneity in their trace element concentrations; outside this range, either the uptake of trace elements in an unanalyzed phase or overestimation of modal proportions is suspected.
|
|
The Pb isotopic compositions of clinopyroxene in the eclogite
xenoliths are almost in equilibrium with zoisite at 30 Ma (
Fig. 14e).
In contrast, the Pb isotopic compositions of clinopyroxene and
zoisite in jadeitic-clinopyroxenite are not in equilibrium with
each other at 30 Ma. Such features are weakly observed in the
other U-series isochron diagrams (
207Pb/
204Pb
235U/
204Pb
and
208Pb/
204Pb
232Th/
204Pb). Although zoisite is the
dominant host of U and Pb, it has distinctively different Pb
isotopic compositions and U/Pb abundance ratios from those of
the whole-rocks. This might be explained by the presence of
zircons with extremely high
238U/
204Pb (>10
6) and
206Pb/
204Pb
(>10
3) (Usui, 2004

).
 |
DISCUSSION
|
|---|
Evidence for high-P/T prograde metamorphism
The compositional zoning structure of garnet and the inclusion
assemblages within garnet grains in eclogites and jadeitic-clinopyroxenites
preserve information about a certain range of the
P
T history of the xenoliths. The Sps component of garnet in both
xenolith types is markedly zoned and preserves a fine euhedral
structure (
Fig. 5). Mn zoning of this type is not observed in
other mantle-derived eclogite xenoliths; however, it is common
in garnet-bearing regional metamorphic rocks that have undergone
prograde metamorphism (e.g. Atherton & Edmunds, 1966

; Zack
et al., 2004

). The Mn zoning of the garnet in the xenoliths
is thus interpreted to reflect growth during prograde metamorphism.
Included clinopyroxenes are richer in Ac and poorer in Jd contents than matrix clinopyroxene (Fig. 6). This probably reflects jadeiteacmite substitution. Clinopyroxene generally becomes more jadeitic and less acmitic with increasing pressure and temperature in the upper blueschist to eclogite facies (Maruyama & Liou, 1988
). Most included lawsonite grains have higher Fe2O3 and lower Al2O3 contents than matrix lawsonite (Fig. 7). Maruyama & Liu (1988)
reported negative correlations between Fe3+ and Al molar abundances in lawsonite during progressive metamorphism at blueschist facies: Fe3+ content decreases with increasing pressure and temperature. Thus, included clinopyroxene and lawsonite probably formed in a lower pressure and temperature metamorphic environment than matrix clinopyroxene and lawsonite.
Garnet grains from the Colorado Plateau eclogites contain curved trails of fine-grained coesite and rutile inclusions (Fig. 2d), indicating the former presence of a planar tectonic fabric (see also Helmstaedt & Schulze, 1988
, 1991
). Such textures do not commonly form by magmatic processes, but are commonly found in regional metamorphic suites, such as those of Corsica and the Franciscan metamorphic complex of the western USA, which are known for their occurrences of high-pressure and low-temperature eclogites (Caron & Pèquignot, 1986
; Maruyama & Liou, 1988
). Thus, the inclusions indicate that the immediate protolith of the eclogite xenoliths was a fine-grained foliated rock, probably a blueschist.
In summary, garnet crystals grew during prograde metamorphism, and successively captured minerals such as clinopyroxene and lawsonite, reaching peak metamorphic conditions ranging from 560 to 700°C at 3 GPa and to 600 to 760°C at 5 GPa (Usui et al., 2003
). Such high-Plow-T environments, which are colder than typical mantle geotherms, can be achieved only in subducted oceanic lithosphere. Therefore, we propose that the Colorado Plateau eclogite xenoliths formed by prograde metamorphic recrystallization of subducted oceanic crust.
Retrogression of eclogite xenoliths during emplacement
The Colorado Plateau eclogites were entrained within and rapidly transported to the surface by the host ultramafic microbreccia diatremes during the Tertiary (
30 Ma). The eclogite xenoliths exhibit less retrograde overprinting than subduction-related eclogites from high-pressure metamorphic terranes, as indicated by the absence of amphibole. However, some modifications resulting from xenolithhost-rock interaction are observed. In the lawsonite-eclogite xenoliths, the lawsonite grains are invariably rimmed by zoisite, and, in the zoisite-eclogite xenoliths, the lawsonite is totally replaced by zoisite. Although similar replacement reactions of