See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/250163494

Petrogenesis of the Neoproterozoic Bikilal-Ghimbi gabbro, Western Ethiopia

Article  in  Journal of Mineralogical and Petrological Sciences · February 2008

DOI: 10.2465/jmps.070401

CITATIONS READS

5 317

2 authors:

Binyam Woldemichael Jun-Ichi Kimura

2 PUBLICATIONS   28 CITATIONS   
Japan Agency for Marine-Earth Science Technology
306 PUBLICATIONS   5,315 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Genesis of ultra-high-Ni olivine in high-Mg andesite lava triggered by seamount subduction View project

Chlorine isotope geochemistry View project

All content following this page was uploaded by Binyam Woldemichael on 30 May 2014.

The user has requested enhancement of the downloaded file.

 Petrogenesis and
Journal of Mineralogical of the Neoproterozoic
Petrological Bikilal
Sciences, − Ghimbi
Volume gabbro
103, page 23 ─ 46, 2008 23

Petrogenesis of the Neoproterozoic Bikilal-Ghimbi

gabbro, Western Ethiopia
Binyam W. WOLDEMICHAEL*, ** and Jun−Ichi KIMURA*
*
Department of Geoscience, Shimane University, Matsue 690 −8504, Japan
**
On leave from Geological Survey of Ethiopia, P. O. Box 2302, Addis Ababa, Ethiopia

The Western Ethiopian Shield is an exposed Neoproterozoic metamorphic belt and forms part of the Arabian−
Nubian Shield. The metamorphic belt consists of high−grade biotite gneisses, low−grade volcanogenic sedi-
ments, and mafic −ultramafic complexes. The Bikilal − Ghimbi gabbro is a mafic body surrounded by these
gneissic rocks, and is located 440 km west of Addis Ababa. The gabbro is elliptical in shape and covers an area
of 350 km2. It consists of olivine gabbro in its center and hornblende gabbro and hornblendite at the perimeter.
The olivine gabbros are very fresh and undeformed, but hornblende−bearing suites have deformational tex-
tures. Each rock type can be divided into apatite−bearing and apatite−free subtypes. The major element geo-
chemistry shows that despite the differences between the olivine and the hornblende gabbros, there is no sys-
tematic chemical contrast between the lithotypes except for the fluid mobile elements, suggesting an origin
from a common parental magma. Only the perimeter is affected by metasomatism. An estimation of the paren-
tal magma composition using the trace element abundance in fresh clinopyroxenes and fresh olivine gabbro
bulk rock suggests an intraplate−type tholeiite. Crystallization model calculations using a tholeiitic parental
magma suggest that the gabbros crystallized in a manner where small amounts of interstitial melt were re-
tained. The apatite−bearing varieties are always associated with Mg−rich mineral phases, suggesting an origin
from the supercooling of replenished basalt into an evolved low temperature magma chamber. The supercool-
ing caused saturation of the apatite in the basalt melt, along with Mg−rich crystals, and these later mixed to-
gether with the more evolved crystals that had precipitated previously. The intraplate−type tholeiitic parental
magma suggests plume−type magmatism for the origin of the Bikilal− Ghimbi gabbro body.

Keywords: Bikilal− Ghimbi gabbro, Geochemistry, Neoproterozoic, Tholeiite, Parental magma, Petrology,
Ethiopia

INTRODUCTION conjunction with the partition coefficients and mineral

Studies on mafic intrusions, e.g., those covering their pa-
rental magma composition, their movements upward
through the crust, and their differentiations and modifi-
cations after being emplaced in the crust, are critical to
an understanding of the development of the Earth’s crust.
A common problem faced in igneous petrology is that
many intrusive rocks are not quenched liquids, but are
crystal segregations or cumulates (e.g., Wager and
Brown, 1968; Irvine, 1982; Hunter, 1996). The chemistry
of cumulates in plutonic rocks is principally controlled
by the modal proportions and composition of the miner-
als that have been accumulated and the interstitial melt
entrapped during crystallization (Bédard, 1994; Green
1994). Whole−rock trace element analyses can be used in
doi:10.2465/jmps.070401
B.W. Woldemichael, wbinyam@yahoo.com Corresponding author
modes to calculate the liquidus−temperature equilibrium
distribution of the trace elements among the constituent
minerals of the cumulate and plutonic rocks (Bédard,
1994; Green, 1994; Wood and Blundy, 1997). These data
can also provide information on the parental magma
types, and thus, the tectonic implications of gabbro bod-
ies (e.g., Bédard, 2001; Claeson and Meurer, 2004;
Kumar et al., 2007). This is particularly important in
studies on gabbros in old terrains.
The Western Ethiopia Shield is a Neoproterozoic
metamorphic belt that forms part of the Arabian−Nubian
shield (ANS, Abdelsalam and Stern, 1996). The meta-
morphic terrain is underlain by high−grade gneisses and
by low−grade metavolcanics and metasediments with as-
sociated mafic −ultramafic belts and syn − to post−tectonic
gabbroic to granitic intrusions. The Bikilal− Ghimbi gab-
bro is elliptical in shape, and has a concentric structure,
 24 B.W. Woldemichael and J.−I. Kimura
with olivine gabbro in its center and hornblende gabbro
and hornblendite at its periphery. The Bikilal− Ghimbi
gabbro is regarded as a syn− to post−tectonic intrusion,
based on its relationship with the surrounding country
rocks (Abraham, 1989; Alemu and Abebe, 2000; Besha-
wered, 2001; Allen and Tadesse, 2003). Previous studies
have regarded the gabbro as being two separate bodies,
namely a syn−tectonic hornblende gabbro and a post−tec-
tonic olivine gabbro. Several geological mapping and
mineral exploration programs on different scales have
been conducted in the Bikilal− Ghimbi area that have at-
tempted to find economic deposits of iron and phospho-
rus. However, no petrological and geochemical investiga-
tions that could constrain the petrogenesis of the gabbroic
body have yet been carried out. This work aimed to inves-
tigate the petrology and geochemistry of the Bikilal−
Ghimbi gabbro to describe the mineralogical and geo-
chemical variations within it, and to clarify the petro-
genesis of the intrusion, including the nature of the paren-
tal magma and apatite ore genesis. The results show that:
(1) the crystallization sequence includes fractional crys-
tallization of olivine, plagioclase, and clinopyroxene and
primary basalt replenishment to form apatite ores within
the Mg−rich layers; (2) the hornblende gabbro and horn-
blendites are metasomatized equivalents of the olivine
gabbro; and (3) the major and trace element analyses of
fresh clinopyroxenes and bulk olivine gabbros suggests a
tholeiitic parental magma in composition. These results
shed light on the igneous and tectonic processes in the
West Ethiopian Shield and in the ANS.
GEOLOGICAL SETTING OF THE BIKILAL-
GHIMBI GABBRO
Regional geology
The term “Pan−African” (Kennedy, 1964) originally re-
ferred to a sequence of tectonothermal events at 500 ±
100 Ma within Africa and Gondwana. However, Kröner
(1984) included orogenic events in the period 950− 450
Ma, and identified the Arabian −Nubian Shield (ANS) as
one of the major orogenic belts formed during Neoprot-
erozoic time. The “Pan−African” assembly of Gondwana
is thought to be the cause of the East African Orogen,
which included a complex intraoceanic and continental
margin magmatism and a protracted sequence of tecto-
nothermal events (Stern, 1994). Two major terrains have
been identified in the East African Orogen. The first is
the juvenile (i.e., mantle derived) ANS in the north,
where the ANS is dominated by low−grade volcano−sedi-
mentary rocks associated with plutons and ophiolitic
remnants (Fig. 1a). The second is a tract of older remobi-
lized crust to the south of the ANS, known as the pre−
Neoproterozoic Mozambique Belt (MB, Fig. 1a). This is
dominated by high−grade gneisses and migmatites. The
East African Orogen stretches south from Israel and
Jordan through to Tanzania and into Antarctica (Stern,
1994). The transition between the juvenile ANS and the
MB is located within Ethiopia, Eritrea, Sudan, and
Somalia.
The Western Ethiopian Shield (WES) lies near the
transition between the ANS and the MB, and also lies ad-
jacent to the enigmatic “East Saharan Metacraton”, which
consists of older crust that was extensively remobilized
during Neoproterozoic time (Abdelsalam et al., 2002).
There has been a limited number of studies on the geo-
chronology and magmatic geochemistry of the WES, and
those that have been carried out have shown results that
are mostly in accordance with models derived for other
parts of the ANS. Three generations of plutonism are rec-
ognized in the WES: (1) “prekinematic” plutons yield em-
placement ages of 814−866 Ma (Ayalew et al., 1990;
Kebede et al., 2001; Grenne et al., 2003); (2) “synkine-
matic” plutons yield emplacement ages of 700−783 Ma
(Ayalew et al., 1990; Kebede et al., 2001); and (3) “late−
to postkinematic” plutons yield emplacement ages of 541−
625 Ma (Ayalew et al., 1990; Kebede et al., 2001). Ayalew
and Peccerillo (1998) suggested that the pre− to synkine-
matic plutons were subduction related, and were emplaced
in intraoceanic island−arc environments. Ayalew and
Peccerillo (1998) and Kebede et al. (1999) inferred that
“late− to postkinematic” plutons possessed geochemical
signatures of both subduction−related and intraplate com-
ponents.
Allen and Tadesse (2003) classified and described
the N−S trending Tuludimtu Belt, where the area of study
of this work lies, into five lithotectonic domains. They in-
terpreted the belts as being a collision of the orogenic belt
formed during the assembly of west Gondwana before the
final closure of the Mozambique Ocean. According to this
tectonic subdivision, the Bikilal− Ghimbi intrusion lies in
the Didesa Domain of these five lithotectonic domains
(Fig. 1b). The rocks within the Didesa Domain consist of
moderate grade paragneisses and orthogneisses intruded
by Neoproterozoic intrusive rocks. The paragneisses con-
sist of interlayered biotite amphibole gneiss, garnet−biotite
gneiss, quartzitic gneiss, and very coarse granitoid gneiss.
The orthogneisses are represented by a banded mafic
gneiss containing ultramafic bands, and could have been
derived from a layered mafic intrusive body. With the ex-
ception of the quartzofeldspathic gneiss, which is rela-
tively massive, all are strongly foliated and have abundant
refolded folds indicating two or more generations of fold-
ing. These rocks have undergone several later brittle de-
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro 25

Figure 1. (a) A generalized map of the ANS showing the regional pre−Neoproterozoic crust, Neoproterozoic crust (juvenile and remobilized),
and possible suture zones/ophiolites. A part of the MB is indicated (modified after Abdelsalam and Stern, 1996; Worku and Schandelmeier,
1996). (b) A geological map of part of the Tulu−Dimtu orogenic belt in the Ghimbi region (modified after Allen and Tadesse, 2003). Three
of the five litho −tectonic domains are indicated. (c) A geological map of the Bikilal− Ghimbi gabbroic intrusion and the location of the sam-
pling sites and the cross−section A−B (modified after Beshawered, 2001).

formation episodes, manifested by complex vein arrays
filling fracture networks. The gneisses appear to be in-
truded by a series of weakly deformed igneous rocks, in-
cluding gabbro, granodiorite, and granite, all of which
display a steep N−S foliation, which is concordant with
the extension direction of the belt.
Few ages are available for the WES gneisses. John-
son et al. (2004) dated 830−785 Ma for the emplacement/
crystallization of the igneous protoliths to the orthog-
neisses of the WES. The age and affinity of the poly-
 26 B.W. Woldemichael and J.−I. Kimura
deformed gneisses in western Ethiopia are still am-
biguous, and at present, no chronological data constrain
the ages of the mafic intrusions in western Ethiopia.
The Bikilal-Ghimbi gabbroic intrusion
Several gabbroic intrusions occur within the WES. Based
on their occurrence, these are classified into syn −, late−,
and post−tectonic varieties. The Bikilal− Ghimbi gabbroic
intrusion is one of the largest intrusions in the region,
and is considered to be a syn− to post−tectonic type,
based on its apparently intrusive relationship with the
surrounding gneiss (Abraham, 1989; Alemu and Abebe,
2000; Beshawered, 2001; Allen and Tadesse, 2003). The
intrusion underlies an area of about 350 km2, and is
about 40 km long, with a maximum width of about 14
km in its central part and a minimum width of 2 km at
its southern tip (Figs. 1b and 1c). An area of about 70
km2 is underlain by olivine gabbro, whereas the remain-
der mainly consists of hornblende gabbro with layers of
hornblendite. The gabbro defines an elliptical area that is
elongated parallel to the N−S foliation of the surrounding
gneissic terrain. The contact with the gneissic terrain is
thought to be an intrusive type, although no chilled mar-
gins or dikes of gabbro in the gneiss are observed.
Layering is a common feature of gabbroic intrusions.
Individual layers or layered sequences can vary greatly in
thickness, texture, shape, and in their mineralogical and
chemical composition (Naslund and McBirney, 1996); the
Bikilal− Ghimbi intrusion is no exception. The olivine
gabbro body has no clear layering. However, an intensive
layering is only limited in the periphery, which mostly
consists of hornblendites and hornblende gabbros,
whereas the olivine gabbro cores are apparently massive.
Layering/banding features occur in the hornblende gab-
bro. The olivine gabbro shows some variation, especially
on pyroxene grain size level. Pyroxene crystals as large as
2 to 3 cm characterize the coarser varieties, whereas the
equivalent finer varieties contain relatively fine −grained
pyroxene (1−5 mm). This textural variation can be consid-
ered as textural layering, although clear layering boundar-
ies are not noticeable.
Rhythmic layering is common in the northern part
of the intrusion. Macrorhythmic layering (1 m to > 100 m)
is manifested by hornblendite occurring as separate units
within the hornblende gabbro. Microrhythmic layering, a
few centimeters thick, is seen in some hornblende gab-
bros. The microrhythmic mafic and felsic layers (bands)
contain concentrations of hornblende and plagioclase, re-
spectively. The hornblende gabbro also varies from
leuco−, meso− to melanocratic, defining a range of light
(plagioclase rich) to dark (hornblende rich) bands. The ig-
neous layering in the western part of the hornblende gab-
bro dips 45 to 75° toward the center of the intrusion. In
the eastern part, the layering follows the general strike of
the intrusion and dips steeply to the west. Hornblendite
layers/bodies occur frequently in the northern and north-
eastern parts, are less common in the south, and are rare
in the central, western, and northwestern parts of the
Bikilal− Ghimbi gabbro. Hornblende gabbro is also char-
acterized by an Fe−Ti and apatite mineralization. These
mineralizations also occur within the hornblendite layers.
SAMPLES AND ANALYTICAL METHODS
Samples
Eighty−nine unweathered fresh samples were analyzed.
Of these, 72 were collected from surface outcrops, 16
were drill core samples, and one additional surface sam-
ple was drawn from the core and outcrop archives of the
Geological Survey of Ethiopia. Of the samples analyzed,
32 were apatite−free olivine gabbros, 35 were hornblende
gabbros, and two were hornblendites. Apatite−bearing
equivalents were represented by three, six, and five sam-
ples, respectively. Three additional samples from the sur-
rounding gneisses were also analyzed, along with three
samples from the granitic intrusions. The outcrop sample
numbers (Fig. 1c) were prefixed by the abbreviation BG
(e.g., BG 16), except for the two samples preceded by the
codes TCH and POG. The drill core samples were pre-
ceded by either of the numbers 512 or 610. A 1:50000
scale geological map of the Ghimbi area was used for the
sampling (Fig. 1c), which was produced by an agro−min-
eral exploration project team that included the first author
of this work and was later compiled by Beshawered (2001).
Bulk rock analysis
The bulk rock major and trace element compositions of
all the samples were determined using X−ray fluorescence
(XRF) spectrometry employing glass disks. The samples
were crushed into small chips using a hammer, rinsed
with distilled water in an ultrasonic bath, and dried for a
period of 2 h at 100 °C. The dried chips were then ground
in an agate mortar. The resulting powders were ignited in
a muffle furnace for a period of 3 h at > 1000 °C to deter-
mine the loss on ignition. The ignited samples were then
used to prepare glass beads containing a mass of 1.8 g of
the powder and 3.6 g of an alkali flux, following the
method of Kimura and Yamada (1996). The alkali flux
used was a mixture of lithium tetraborate and lithium
metaborate in a ratio of 4:1. The glass beads were ana-
lyzed for the major elements and 11 trace elements using
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro 27

a Rigaku RIX 2000 XRF spectrometer located at Shima- Table 1. Modal analyses (vol%) of the different rock units
ne University, Japan.
The trace and ultratrace element analyses of 16 sam-
ples of the olivine and hornblende gabbros and their apa-
tite−bearing equivalents were carried out using solution−
based Inductively Coupled Plasma−Mass Spectrometry
(ICP−MS) employing the method described by Kimura
et al. (1995). The ICP−MS used was a Thermo Elemental
VG PQ3 located at Shimane University, Japan.

Mineral analysis

Petrological descriptions and modal analysis were carried

out on 23 samples, including four hornblendites, eight horn-
blende gabbros, and 11 olivine gabbros. The major ele-
ment mineral chemistry of representative plagioclase, ol-
ivine, pyroxene, hornblende, and oxide minerals was deter-
mined using a JEOL JXA−8800M electron microprobe
located at Shimane University, Japan, using an accelerat-
ing voltage of 15 keV and a beam current of 2 × 10−8A.
Data correction followed the ZAF method (e.g., Reed,
1975). The diameters of the focused beam for the analy-
ses were 5 μm for the olivine, plagioclase, and the oxide
minerals, and 20 μm for the pyroxene.
The trace element abundance of several minerals
was analyzed using an in−house 193−nm excimer laser ab-
lation microprobe combined with the VG PQ3 ICP−MS
(Kimura et al., 2000). The mass peak occurring at 44Ca
was used as an internal standard. The analysis included
45 spots on the clinopyroxenes. Three adjacent points
were analyzed, and the resulting data were averaged. The
major element compositions of the analyzed grains were
determined using the EPMA before analysis using the
LA−ICP−MS were carried out. The instrument setting
used for the pyroxenes a pulsed laser with 200 mJ of laser
output power. The laser craters formed were ~ 60 μm in
diameter with depths of ~ 30 μm after an ablation time of
60 s.
PETROGRAPHY
Hornblendite
The hornblendite is fine − to medium−grained with phane-
ritic and massive texture. The hornblende grains are sub-
hedral to anhedral, and show slight schistosity. Uralitiza-
tion is prominent, with rare traces of unaltered pyroxene
cores. The hornblendite is subdivided into apatite−bearing
(modal apatite > 3%) and apatite−free groups (modal apa-
tite ≤ 3%). The “apatite−free groups” have no discernible
apatite. The modal composition of the hornblendite ranges
from 60 to 90% hornblende, 5 to 30% ilmenite−magne-
Ol Gb, olivine gabbro (apatite free); Ol Gb Ap, olivine gabbro
apatite bearing); Hb Gb, hornblende gabbro (apatite free); Hb Gb
Ap, hornblende gabbro (apatite bearing); HBT, hornblendite (apa-
tite free); HBT AP, hornblendite (apatite bearing); Ol, olivine;
Cpx, clinopyroxene; Hb, hornblende; Pl, plagioclase; Ox, ox-
ide minerals; Ap, apatite.
tite, 0 to 9% plagioclase, and 0 to 20% apatite (Table 1).
The hornblendite units are hosted within the hornblende
gabbro.
Hornblende gabbro
The hornblende gabbro is the outer unit and formed a
large portion of the Bikilal− Ghimbi intrusion. This unit is
characterized by melanocratic to leucocratic varieties, de-
pending on the plagioclase and hornblende content, and is
generally medium to coarse grained. This unit also hosts
hornblendite. The plagioclase and hornblende exhibit a
relict to granoblastic texture, with accessory ilmenite, ap-
atite, and chlorite. The hornblende gabbro is composed of
40 to 70% hornblende, 0.5 to 15% ilmenite−magnetite, 20
to 55% plagioclase, and 0 to 20% apatite (Table 1). This
unit is also subdivided into apatite−bearing and apatite−
free hornblende gabbro. The apatite−bearing gabbro com-
monly occurs within the hornblende gabbro mass, and lo-
cally, is regarded as disseminated apatite mineralization.
The plagioclase shows fine albite twining and a kink
structure had developed within the crystals, indicating
that the rocks had undergone deformation. In a few
places, the plagioclase shows a granular texture and is
 28 B.W. Woldemichael and J.−I. Kimura
Figure 2. Representative photographs of thin sections of the
Bikilal− Ghimbi gabbro: (a) fresh olivine gabbro and (b) horn-
blende gabbro with apatite and ilmenite−magnetite. Remnants of
pyroxene are visible, indicating the hornblende was originally
pyroxene. Pl, plagioclase; Cpx, clinopyroxene; Ol, olivine; Hbl,
hornblende; Ilm, ilmenite; Mgt, magnetite; Ap, apatite; Alt, al-
tered.
broken down to form fine grains, from which the original
twining is lost. The hornblende appears to be an altered/
metasomatized product of the pyroxene, as suggested by
some hornblendes that contain traces of pyroxene (Fig.
2b). These observations, combined with the gradational
contact with the olivine gabbros seen in some places, sug-
gest a hydrothermal alteration/metasomatism at late stage
magmatism may have played a role in the formation of
the hornblende gabbro.
Olivine gabbro
A large mass of olivine gabbro occupies the central por-
tion of the Bikilal− Ghimbi intrusion. This massive olivine
gabbro is surprisingly fresh and is unaltered, except for
minor veins and weathering. Deformation textures are
rare, and all the crystals are fresh and retained their origi-
nal igneous shape. The olivine gabbro is medium− to
coarse−grained holocrystalline, with a hypidiomorphic
granular gabbroic texture, and the modal compositions
range from 5 to 25% olivine, 10 to 35% clinopyroxene, 30
to 55% plagioclase, 0 to 16% apatite, and 1 to 15% ilmen-
ite−magnetite (Table 1). Most of the crystals are subhedral
to anhedral, but most of the apatite is euhedral. The oliv-
ine is interstitial with ophitic to subophitic plagioclase in
the clinopyroxene oikocrysts (Fig. 2a). All the crystals are
weakly zoned. This unit is further classified into apatite −
bearing and apatite−free olivine gabbro. The olivine gab-
bro is essentially massive, and may contain coarser pyrox-
ene crystals (2 to 3 cm in size).
The fresh nature of the olivine gabbro allows us to
investigate the origin of the Bikilal− Ghimbi gabbro fur-
ther using geochemistry. Therefore, the geochemical in-
terpretations described in later sections in this paper are
based on data from the unaltered olivine gabbro.
GEOCHEMISTRY
Major element bulk rock chemistry
The major element geochemistry of the Bikilal− Ghimbi
gabbro is based on analyses of 36 olivine gabbros (includ-
ing a subset of three apatite−bearing samples), 45 horn-
blende gabbros (seven apatite−bearing), and seven horn-
blendites (five apatite −bearing). Hereafter, we will apply
the above descriptors to the three rock units. Represen-
tative bulk rock major and trace XRF data are shown in
Table 2, and all the analyzed samples are tabulated in
Appendix 1.
In the Na2O + K2O versus SiO2 classification dia-
gram of Cox et al. (1979), as modified for plutonic rocks
by Wilson (1989), the Bikilal− Ghimbi gabbro plots almost
entirely in the subalkalic magma series field, and has a
gabbroic composition (Fig. 3). The silica content of the
apatite−free gabbros and hornblendites shows a wide
range from 45 to 55 wt%. The apatite−bearing subsets all
plot out of range on this diagram due to their low silica
content (25− 45 wt%, not shown in Fig. 3). This is due to
the effect of the apatite concentration, which results in an
abundance of P2O5 of up to 8 wt%. Although rock types
range from fresh olivine gabbro to uralitized hornblende
gabbro to hornblendite, overall, the Bikilal− Ghimbi gab-
bro samples fall in a narrow compositional range.
Significant variations in FeO, TiO2, and P2O5 are ob-
served in the Bikilal− Ghimbi gabbro. The AFM triangu-
lar plot shows an Fe enrichment trend for the apatite−
bearing subsets. The boundary proposed by Kuno (1968)
reveals both tholeiitic and calc−alkaline trends (Fig. 4),
whereas only a tholeiitic trend is observed according to
the boundary proposed by Irvine and Baragar (1971).
 Table 2. Representative whole−rock analyses from the Bikilal− Ghimbi gabbroic intrusion
Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro

Ol Gb, olivine gabbro (apatite free); Ol Gb Ap, olivine gabbro (apatite bearing); Hb Gb, hornblende gabbro (apatite free); Hb Gb Ap, hornblende gabbro (apatite bearing); HBT, hornblendite (ap-
atite free); HBT AP, hornblendite (apatite bearing).
29
 30 B.W. Woldemichael and J.−I. Kimura
Figure 3. A total alkali versus silica diagram (Cox et al., 1979, ad-
opted for plutons by Wilson, 1989). Only apatite−free samples
are plotted, as apatite−bearing samples are poor in both silica
and alkali, and are plotted out of scale.
The MgO content of the Bikilal− Ghimbi gabbro
ranges from 2 wt% to 13 wt%. The MgO abundance over-
laps between rock suites, irrespective of the apatite con-
tent (4−12 wt% for the olivine gabbro, 2−12 wt% for the
hornblende gabbro, and 8−13 wt% for the hornblendites)
(see Fig. 5). Although there is little difference in the range
of MgO in the olivine and hornblende gabbros, most of
the olivine gabbros are characterized as having a higher
MgO content (8−13 wt%), whereas the majority of the
hornblende gabbros are low in MgO (2−8 wt%) for both
the apatite−bearing and −free subsets. A plot of MgO ver-
sus P2O5 highly discriminates between the apatite−bear-
ing and apatite−free groups (Fig. 5). The apatite−bearing
subsets are enriched in P2O5 (2−8 wt%), whereas the apa-
tite−free subsets contain very little P2O5 (0.1− 0.3 wt%).
The magnesium numbers also range from highly evolved,
with values as low as Mg# = 15, to slightly evolved (Mg#
= 55). Most olivine gabbros have Mg# = 50−55. The
modal enrichment of the Fe−Ti oxides in the gabbro led to
a lower Mg#.
MgO correlates positively with FeO, TiO2, MnO, and
P2O5 in the apatite−bearing subsets, and correlates nega-
tively with Al2O3, SiO2, K2O, and Na2O. Both the apatite−
free and the apatite−bearing olivine gabbros show strong
negative correlations of Na2O and K2O with MgO. A
slightly negative correlation is observed for Al2O3 and
SiO2 in the apatite−free lithotypes. FeO, TiO2, and MnO
in the apatite−free olivine gabbros increase slightly with
increasing MgO up to ~ 7 wt%, and then they decreased
with increasing MgO. The chemical composition of the
apatite−free hornblende gabbros and hornblendites over-
Figure 4. An AFM classification diagram for the Bikilal − Ghimbi
gabbro showing both calc−alkaline and tholeiitic trends accord-
ing to the boundary between the two fields, as proposed by
Kuno (1968), showing the tholeiitic trend proposed by Irvine
and Baragar (1971).
lapped extensively with that of the apatite−free olivine
gabbros, but the data are more scattered. This is particu-
larly true for K2O. The K2O abundance in the hornblende
gabbros is up to about five times greater than that in the
other lithotypes (Fig. 5), suggesting a metasomatic addi-
tion of potassium. Most of the hornblende gabbros with
the highest K2O contents lie either close to the contact
with the gneiss (e.g., Sample BG 20) or along the margins
of the regional lineaments/faults (e.g., Samples BG 76,
BG 77, and BG 82; Fig. 1c). The apatite−bearing subsets
have greater modal apatite and iron oxide contents (18−30
modal percent in the olivine and hornblende gabbros, and
30− 47 modal percent in the hornblendites; see Table 1).
The difference in modal composition is also reflected in
the low SiO2, Al2O3 content, and the higher MnO abun-
dance in the apatite−bearing samples.
Major element mineral chemistry
The mineral chemistry of the olivines, pyroxenes, plagio-
clases, hornblendes, and the oxide minerals of the Bikilal−
Ghimbi gabbro was examined based on 542 spot analy-
ses. Most of the analyses were carried out on the olivine
gabbro, because it was the freshest lithotype and was vir-
tually unaltered. Representative results are shown in
Tables 3−7, and all the data from the analyzed spots are
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro
Figure 5. Major element−MgO variation diagrams for the Bikilal− Ghimbi gabbro. The end member was estimated using multiple least square
regression calculations.
tabulated in Appendix 2.
Olivine. Olivine constitutes 5−25% of the olivine gabbros.
The olivine has Mg numbers [Mg# = 100 Mg/(Mg + Fe)]
ranging from 45 to 75 (Fo45−75). Representative olivine
analyses are listed in Table 3. The olivine is normally ho-
mogeneous from core to rim with no obvious zoning. The
Fo content varies little within individual samples (e.g.,
Sample BG 54; Fo 70−74), whereas a higher variation was
observed between different samples (e.g., Samples BG11
and TCH 23, Fo60−76). The olivine composition showed a
trimodal distribution, with a median Fo content at Fo49.2,
Fo61.1 and Fo 72.4 corresponding to a host bulk rock MgO
content of 6 wt%, 6−9 wt%, and 10−11 wt%, respectively.
The olivine exhibits a variable Fo content, especially with
evolved compositions in the apatite−bearing samples
(Fo40−60). This shows that the span of olivine crystalliza-
31
tion was prolonged. Given that the Ni content of the oliv-
ines is very low (< 0.1 wt%) and the Fo content is low to
moderate (Fo40−70), it appears that the Bikilal− Ghimbi
gabbro crystallized from an evolved magma.
Pyroxene. Clinopyroxene is present throughout the oliv-
ine gabbros and constitutes 25 to 35% of the modal com-
position. Representative clinopyroxene analyses are
shown in Table 4. Compositionally, the clinopyroxenes
are mainly augite (34−52 mol% En, 10−27 mol% Fs, and
32− 45 mol% Wo), with little diopside present (42− 48
mol% En, 10−19 mol% Fs, and 45− 47 mol% Wo). The py-
roxenes exhibit a limited Fe enrichment relative to oliv-
ine, with Mg−numbers in the range 69−80 (Mg# = 45−75
in olivine). The grains are homogeneous from core to
rim, with no identifiable zoning. The value of the Mg#
varies considerably within individual samples (e.g.,
 32 B.W. Woldemichael and J.−I. Kimura
Table 3. Representative olivine analyses
Ol Gb, olivine gabbro (apatite free); Ol Gb Ap, olivine gabbro
(apatite bearing).
Sample BG 11 with Mg# = 77−73), but not within grains.
However, the variation in Mg# was even higher between
different samples (e.g., Samples 6104 and TCH 23, with
Mg# = 69−80). The value of the Mg# of the clinopyrox-
enes and the range in silicon content (1.82−1.95 pfu) is
typical of subalkaline gabbro (e.g., Tepper, 1996).
The concentration of the minor elements, such as
NiO, is very low (< 0.07 wt%), whereas the Cr2O3 content
showed more variation. Some clinopyroxenes from both
apatite−free and apatite−bearing suites have very low
Cr2O3 content (< 0.06 wt%). Other clinopyroxenes mostly
from the apatite−free suites contain 0.1− 0.5 wt% of Cr2O3.
These Cr−rich clinopyroxenes may have crystallized at an
early stage before the parental magma was depleted in
Cr2O3.
Plagioclase. Plagioclase occurs in both the olivine gabbro
and the hornblende gabbro. Representative plagioclase
Table 4. Representative clinopyroxene analyses
Ol Gb, olivine gabbro (apatite free); Ol Gb Ap, olivine gabbro (ap-
atite bearing).
analyses are given in Table 5. The plagioclase from the
olivine gabbro ranges in composition from An44−72. The
plagioclase grains are generally homogeneous from core
to rim, with no evident zoning. Thus, the compositional
variation within each sample is relatively small (e.g.,
Samples BG 54; An60−62) compared to that between sam-
ples (e.g., Samples BG11 and TCH 23, An45−70). The pla-
gioclase composition shows a trimodal distribution, with
medians at An46.7, An55, and An63. Most of the plagioclase
in the apatite−bearing samples (e.g., Samples 6104 and
5125) is less than in An46.7 and is more sodic than that
from the apatite−free varieties.
Although the number of plagioclase analyses carried
out on the hornblende gabbro was insufficient for statisti-
cal examination, they do show a slightly wider range of
An content than those from the olivine gabbro (e.g.,
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro 33

Table 5. Representative plagioclase analyses Table 6. Representative hornblende analyses

Hb Gb, hornblende gabbro (apatite free).

Table 7. Representative oxide minerals analyses

Ol Gb, olivine gabbro (apatite free); Ol Gb Ap, olivine gabbro
(apatite bearing); Hb Gb, hornblende gabbro (apatite free).

Samples BG57 and BG35, An42−75). Like the plagioclase

of the olivine gabbro, the plagioclase in the hornblende
gabbro do not show any zoning.

Amphibole. The uralitized hornblendes were difficult to

analyze using the microprobe because of their secondary
origin and their derivation from primary pyroxene. Des-
pite the difficulty in analyzing the uralitized hornblendes,
several amphibole analyses with acceptable stoichiometry
were acquired. Representative results are listed in Table 6.
The amphiboles in the Bikilal− Ghimbi hornblende gabbro
are calcic amphiboles, ranging from tschermakite and
Ol Gb, olivine gabbro (apatite free); Ol Gb Ap, olivine gabbro (ap-
magnesiohornblende to actinolite hornblende, according
atite bearing); Hb Gb, hornblende gabbro (apatite free); Mag,
the classification of Leake et al. (1997). Most are magne- magnetite; Ilm, ilmenite.
siohornblende.
The amphiboles were characterized by high Al2O3
contents (6−12 wt%). The XMg values [XMg = Mg/(Mg
 34 B.W. Woldemichael and J.−I. Kimura
Figure 6. Trace element−MgO variation diagrams for the Bikilal− Ghimbi gabbro.
+ Fe2+)] are > 0.5, and range from 0.6 to 0.8. Their Cr2O3
content ranges from < 0.01 wt% to 0.8 wt%, and a high
Cr2O3 abundance corresponds with a high XMg value.
This shows that Mg and Cr abundances in the olivine gab-
bro protoliths were variable, and this variability was in-
herited by the amphiboles.
The optical evidence suggests that the origin of the
hornblende is from the pyroxene (Fig. 2b), but the geo-
chemical process involved could not be constrained. Most
of the alteration seen is uralitization, and uralitization to
form hornblende can be associated with late stage hydro-
thermal recrystallization (Deer et al., 1992).
Oxide minerals. The oxide minerals in the Bikilal−
Ghimbi gabbro are ilmenite (92−97 mol% Ilm) and minor
magnetite (86−99 mol% Mgt) (Table 7). The Cr2O3 abun-
dance in the magnetite is generally very low, but several
analyses from the most magnesian olivine gabbros show
higher levels (2 to 3 wt% Cr2O3).
Trace element bulk rock chemistry
All the samples analyzed for major element analysis were
also analyzed for 11 trace elements using XRF (Table 2).
The Ba, Sr, and Ga contents correlate negatively with
MgO, and the chemical variation between the apatite−free
and apatite−bearing samples is similar, whereas in the
trends for other trace elements, the apatite−free and apa-
tite−bearing gabbros are more distinct (Fig. 6). The apa-
tite−free olivine gabbro shows clearer correlations than
the hornblende gabbro did, whose data tend to be scat-
tered. The incompatible trace elements of Ce and Y cor-
relate positively with MgO in the apatite−bearing subset,
and their abundance is elevated due to the concentration
of rare earth elements (REE) and yttrium in the apatite
(Henderson, 1984). Both Ni and Cr show positive correla-
 Table 8. Trace and REE element analyses of the Bikilal− Ghimbi intrusion
Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro

Ol Gb, olivine gabbro (apatite free); Ol Gb Ap, olivine gabbro (apatite bearing); Hb Gb, hornblende gabbro (apatite free); Hb Gb Ap, hornblende gabbro (apatite bearing).
35
 36 B.W. Woldemichael and J.−I. Kimura

Figure 7. Primitive mantle normalized (Sun and McDonough, 1989) trace element spider diagrams of the Bikilal− Ghimbi gabbro: (a) olivine
gabbro (apatite free); (b) hornblende gabbro (apatite free); (c) olivine gabbro (apatite bearing); and (d) hornblende gabbro (apatite bearing).
The shaded regions are olivine gabbro and hornblende gabbro (apatite free) from Figures 7a and 7c, respectively.

tions with MgO in the apatite−free olivine gabbro and
hornblende gabbro, and are almost constant over the en-
tire MgO range in the apatite−bearing samples. In con-
trast, the V content is high, and it correlates with the MgO
content in the apatite−bearing samples, whereas the V
content is low and varies similar to the TiO2 and FeO con-
tent (see Fig. 5) in the apatite−free olivine gabbro. The de-
crease in Ni content with decreasing MgO content sug-
gests an olivine fractionation. The same pattern for Cr
suggests a clinopyroxene fractionation. The low Cr and Ni
contents of the apatite−bearing samples cannot be ac-
counted for by apatite and Fe−Ti oxide concentration
alone, suggesting a depletion of these elements in the
magma. A similar behavior between V, FeO, and TiO2 is
suggestive of a fractionation or accumulation of Fe−Ti ox-
ides in the apatite−bearing samples. The zirconium abun-
dance is very low, with only a few samples’ concentra-
tions exceeding 50 ppm. This could be due to its incom-
patibility in olivine, plagioclase, and apatite. The scattered
and high abundance of Ba in both the apatite−free and the
apatite−bearing hornblende gabbros implies that a substi-
tution of K by Ba had occurred in the hornblende during
metasomatism.
Ultratrace element and REE analyses were carried
out on 16 samples, including 11 olivine gabbros (two apa-
tite−bearing) and five hornblende gabbros (two apatite −
bearing; see Table 8), and the data were plotted on a
primitive mantle−normalized trace element spidergram
(Fig. 7). The spidergram patterns for the apatite−free oliv-
ine and hornblende gabbros are similar, except for ele-
vated incompatible element abundances, particularly Cs,
Rb, Pb, and Li, in the hornblende gabbro (Figs. 7a and
7b). The similarity in patterns, except for these slight en-
richments, further supports the petrological findings that
the hornblende gabbro suites are hydrothermally altered
products of the olivine gabbro. The patterns show positive
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro
Table 9. Representative trace and REE analyses of clinopyroxenes from the olivine gabbro of the Bikilal− Ghimbi intrusion, including
analysis of some apatite and oxide minerals
Alt., altered.
spikes at Cs, Ba, K, Pb, Sr, and Eu, and negative spikes at
Li, Zr, Hf, Nb, Ta, U, and Th. The samples are slightly
enriched in the light REE (LREE), but the concentration
level is as low as that in the primitive mantle. The apatite−
bearing samples are enriched in almost all the elements
relative to the apatite−free samples. Higher concentrations
of Cs, Rb, and Li are also observed in the apatite−bearing
hornblende gabbro.
The positive spikes in Ba, Pb, Sr, and Eu may be due
to a plagioclase accumulation. Low concentrations of Zr,
Hf, U, Th, Ta, and Nb could reflect depletion in the source
magma, but are more likely the result of a significant pla-
gioclase accumulation due to the very low partition coef-
ficients of these elements (Rollinson, 1993; Bindeman et
al., 1998; Kimura et al., 2002). The relatively elevated
patterns in the apatite−bearing samples (Figs. 7c and 7d)
may be a product of apatite accumulation, as indicated by
37
the XRF major and trace element data. The unusual pat-
terns shown by the Bikilal− Ghimbi gabbros do not show
the melt composition, suggesting that the trace element
abundance is controlled by the crystallization of specific
minerals.
Trace element mineral chemistry
Forty−seven LA−ICP−MS spot analyses of the clinopyrox-
enes from the MgO−rich (9−11 wt%) olivine gabbro were
carried out. (See Table 9 for representative data and Ap-
pendix 3 for all the analyzed spot data). The pyroxenes
were analyzed from the core to the rim to determine if
chemical zoning was present. The results suggest that the
trace element composition of the clinopyroxenes is homo-
geneous within single grains, as observed for the major
elements. With the exception of one grain, the average
 38 B.W. Woldemichael and J.−I. Kimura
Figure 8. A primitive mantle normalized (Sun and McDonough,
1989) trace element plot of the pyroxenes from Bikilal− Ghimbi
olivine gabbro. The average values of the analyzed spots were
used. The numbers in brackets represent bulk MgO in wt% and
the asterisk denotes an apatite−bearing subset. Altered pyroxene
is denoted by the dashed line.
composition of the clinopyroxenes from four samples
shows a limited variation, as observed by their near paral-
lel patterns (Fig. 8). The REE abundance is similar, but a
larger variation is observed for the high field strength ele-
ments (HFSEs; Th, U, Nb, and Ta) and the large ion litho-
phile elements (LILEs; Pb, Ba, and Rb). Overall, the
clinopyroxene spidergram patterns show a strong deple-
tion in LILEs and convex−upward shapes. This is similar
to the behavior of clinopyroxene partition coefficients (e.g.,
Green, 1994). One exceptional clinopyroxene grain in
Sample TCH 23 is enriched in the incompatible trace ele-
ments Rb, Ba, Th, U, and Pb (Fig. 8). Although not evi-
dent optically, the enrichment in this grain is likely to be
due to trace alteration (uralitization), given that the ele-
ments above are all fluid mobile.
DISCUSSION
Layered structures and their origin
A variety of layer−forming mechanisms can produce lay-
ering in igneous intrusions. The layering seen in the
northern part of the Bikilal− Ghimbi intrusion can be ex-
plained by the combined mechanisms of convection, as
seen by modal layering, and nucleation, manifested by the
texture.
Rhythmically banded layers with differing modal
composition can result from convective circulation. In
shallow magma chambers, heat is lost mainly through the
roof, while crystals accumulate on the floor. This creates
buoyancy fluxes that can promote convection in the
magma (Naslund and McBirney, 1996). Modally banded
rhythmic layers in the northern perimeter of the Bikilal−
Ghimbi hornblende gabbro may be related to such mag-
matic convection. Although the hornblende gabbro was
slightly deformed and metasomatized, contrasts in the
modal abundances of the unaltered plagioclase between
leucocratic to melanocratic bands can be attributed to
crystallization from convecting magma, as seen in the
Skaergaard intrusion (Naslund and McBirney, 1996).
Apatite−bearing bands/layers are common in the horn-
blende gabbros, which also exhibit modal banding. The
apatite zones are more or less structurally concordant
with the layering, and hence, these would also originate
from a direct precipitation from the parental magma.
Bikilal− Ghimbi olivine gabbro can be characterized
by textural layering, as represented by the coarser and
finer varieties of the pyroxene crystals. Textural layering
is based on differences in textural elements (e.g., mode
and grain size/shape distribution) and individual layers
showing considerable internal variation. These textural
variations can indicate nucleation (Hunter, 1996). The
general correlation between the bulk rock MgO abun-
dance and the Mg# of the mafic minerals and An content
of the plagioclase of the gabbro suggests that overall, the
gabbro was generated from the melt at different fraction-
ation levels in the magma chamber. A differentiation in
the olivine gabbro is reflected by the chemical variability
found in the olivine (Fo45−75), plagioclase (An44−72), and in
the clinopyroxene (Mg# = 69−80), which suggests cryptic
layering. However, based on our data, the olivine gabbro
body showed no systematic cryptic layering, such as a
concentric structure or layered patterns. However, nucle-
ation may have played a major role in the formation of the
olivine gabbro, as implied by the texture.
Crystallization sequence
Olivine gabbro. As discussed above, the hornblende gab-
bro and hornblendite represent altered equivalents of the
olivine gabbro rather than being a separate intrusion from
different parental magmas. The apparent similarity of the
igneous texture of the plagioclase and its close chemistry
with that of the olivine gabbro indicates that substantial
igneous differentiation occurred before metasomatism.
This is consistent with the interpretation from the major
element bulk rock chemistry that the hornblende gabbros
were produced by a metasomatism of the olivine gabbro.
The olivine gabbros are surprisingly well preserved and
show clear igneous textures, and hence, are the best sam-
ples to examine the crystallization sequence of the
Bikilal− Ghimbi body.
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro
Figure 9. K2O content, mode abundance, and mineral chemistry
of the olivine gabbro versus the bulk rock MgO content (in
wt%). Pl, plagioclase; Cpx, clinopyroxene; Ol, olivine; Ilm, il-
menite; Mgt, magnetite.
Taking into account the holocrystalline hypidiomor-
phic gabbroic texture observed under the microscope, the
order of crystallization of the major phases is olivine, fol-
lowed by plagioclase and clinopyroxene. The olivine ex-
hibits an intercumulus texture that may reflect a liquid −
solid reaction, followed by the crystallization of the clino-
pyroxene. The plagioclase also crystallized before the
clinopyroxene, as it occurs as oikocrysts in the clinopy-
roxene. However, the crystallization of these three major
phases should have been contemporaneous, based on the
correspondence between various mineral compositions
and the bulk rock chemistry. The crystallization of the
major phases was followed by crystallization of Fe−Ti ox-
ides, as these mostly occur in volume only in low MgO
content samples.
The olivine gabbros show linear geochemical trends
(as shown by the MgO versus SiO2, Al2O3, CaO, Na2O,
and K2O contents), and the inflected trends for TiO2 and
FeO contents (Fig. 5), suggestive of a crystallization se-
quence formed along with a fractionation of the source
magma. This has been tested by examination of the
modal and mineral composition of the representative oliv-
ine gabbros in the MgO content range of 6−12 wt% (Fig.
9). The selected samples show a very strong correlation
between the MgO and K2O contents (Fig. 9a). However,
the maximum K2O content is less than 0.2 wt% K2O, an
amount that could be hosted almost entirely in the plagio-
clase. The modal olivine and plagioclase contents do not
vary significantly, whereas the clinopyroxene tends to de-
crease, and the Fe−Ti oxide content clearly increases with
decreasing bulk MgO content (Fig. 9b). The Fo content of
39
Figure 10. Plagioclase anorthite (An) content versus coexisting ol-
ivine forsterite (Fo) content for establishing the tectonic setting
according to the classification diagram of Beard (1986). Types I,
II, and III represent different varieties of arc cumulate gabbros.
the olivine and the An content of the plagioclase de-
creases almost linearly as the bulk MgO content de-
creases (Figs. 9c and 9d). These features indicate that the
bulk rock composition is controlled by the mineral chem-
istry and modal abundances. The restricted chemical
variations of each mineral, together with their euhedral to
subhedral habits, strongly suggest that the olivine gabbro
does not represent the melt composition.
Other elements provided additional tests of the pre-
ferred crystallization sequence. The FeO, MnO, TiO2, and
V abundances initially increase with decreasing MgO
content. Below an MgO content of ~6 wt%, the abun-
dance decreases as the MgO content decreases (Figs. 5
and 6). The Fe−Ti oxide minerals hosting these elements
and phases are modally the most important in the MgO−
poor samples. Thus, the inflected trends suggest a deple-
tion of these elements from the parental magma due to a
separation of the Fe−Ti oxides. The abundance of Ni and
Cr in the olivine gabbros decline steeply with decreasing
MgO content (Fig. 10). This suggests separation of Fo−
rich olivine (for Ni) and of Mg−rich clinopyroxene (for Cr)
from the parental magma. Such a sequence is typical in
melt compositions in fractional crystallization. Hence, the
chemical evolution of the parental magma and accumula-
tions of the precipitated minerals are the prime controls
of the chemical variation in the Bikilal− Ghimbi olivine
gabbro samples.
Apatite-bearing rocks. The apatite−bearing hornblen-
dite, hornblende gabbro and olivine gabbro subsets have
various MgO contents. As shown by the MgO−SiO2, MgO−
 40 B.W. Woldemichael and J.−I. Kimura
Al2O3, MgO−FeO, MgO−TiO2, and MgO−P2O5 plots, the
apatite−bearing suites plot along a line that converges
with the apatite−free suite trend at low MgO contents (Fig.
5). This suggests the existence of an end member phase
with low Si and Al contents and high Ti, Fe, Mn, P, and
Mg contents (see the high Mg end member compositions
in Fig. 5). Compositional control by apatite plus Fe−Ti ox-
ides alone cannot entirely explain this variation. However,
a high Mg end member composition can be achieved by
mixing olivine, apatite, and Fe−Ti oxides. Thus, mixing
between the evolved low Mg olivine gabbro and the high
MgO end member can generate a linear trend in the apa-
tite−bearing rocks. The extreme high−Mg end member
would have a modal composition of olivine : clinopyrox-
ene : plagioclase : magnetite : ilmenite : apatite = 29 :19 :7:
17:14 :14, assuming that the most primitive minerals ob-
served in the olivine gabbro were the constituents. This
estimate was derived by solving multiple least square re-
gression calculations with the high Mg end member and
mineral compositions. The observed percentages of apa-
tite (17%), Fe−Ti oxide (29%), and plagioclase (3%) in the
most MgO−rich apatite−bearing hornblendites (Table 1)
are almost identical to those in the estimated mineral
mode. Given that the uralitized hornblende is an altered
product of olivine, clinopyroxene, and plagioclase, the
MgO−rich hornblendite represents the metasomatized end
member of the apatite−bearing olivine gabbro.
A comprehensive mineral chemistry is not available
for the apatite−bearing subsets. If the proposed mixing
model is correct, then the mineral chemistry of the apa-
tite−bearing subsets should be variable. As shown by the
limited data, the apatite−bearing olivine gabbros (Samples
6104 and 5125), olivine (Fo44−60), and plagioclase (An44−48)
are highly evolved, whereas the clinopyroxene (Mg# =
70−77) is more magnesian than the olivine. This suggests
that the apatite−bearing suite rocks are a mixture between
the MgO−rich and MgO−poor end members, consistent
with the above mixing model. This mixing will be dis-
cussed further in a later section, along with the apatite
saturation conditions.
Parental magma and tectonic significance
Mineralogical constraints. The composition of the coex-
isting plagioclase and olivine in the arc gabbros are
unique to the arc setting, despite the significant variations
in mineralogy and mineral chemistry present in the arc
cumulate suites. Calcic (An85−100) plagioclase and moder-
ately Fe−rich (Fo60−80) olivine commonly occur together
in the arc cumulate gabbros, but do not occur together in
cumulate gabbros from mid− oceanic ridges, oceanic is-
lands, or from tholeiitic layered intrusions (Fig. 10, Beard,
1986). The An versus Fo plot implies that the Bikilal−
Ghimbi olivine gabbro is unlikely to be from an arc or
mid− oceanic ridge setting; rather, the gabbros may have
formed in oceanic or continental intraplate settings,
shown by oceanic island and tholeiitic layered intrusion
fields in Figure 10.
Trace element constraints. The spidergrams of the
Bikilal− Ghimbi olivine gabbro (Fig. 7) do not show typi-
cal melt patterns. The spidergrams show very strong neg-
ative anomalies for the HFSEs, Th and U, with striking
positive peaks for Ba, Sr, and Pb. These features may
largely be due to plagioclase, which dominates the mode
(Rollinson, 1993; Bindeman et al., 1998; Kimura et al.,
2002). Therefore, coexisting phases are the dominant fac-
tor determining the trace and REE element chemistry of
the gabbroic rocks, rather than the trapped interstitial
melt.
The parental magma of the crystallized end member
could be further constrained by estimation of the parental
melt composition. As discussed earlier, the hornblende
gabbro is an altered product of the olivine gabbro, and
shows a slight enrichment in fluid −mobile trace elements.
Therefore, we used the fresh olivine gabbro to estimate
the parental melt composition. The trace element distribu-
tion coefficients are essential for determining a magma
composition from the mineral composition (e.g., Bédard,
1994; Green, 1994). The distribution coefficient (D) of a
given element i is used to calculate the concentration of
that element in the equilibrium liquid. The bulk value of
D for element i in mineral j is calculated using the follow-
ing equation:
,
where Φ is the modal proportion of a mineral phase. The
liquid composition of element i, Liq i, can then be calcu-
lated by dividing the whole−rock concentration WRi by
the bulk value of D:
.
If the trapped melt fraction is neglected, then in the
case of cumulates, the calculations overestimate the abun-
dance of incompatible trace elements in the minerals and
coexisting liquids (Bédard, 1994). In nature, the trapped
melt fraction crystallizes in the interstices between the
accumulated crystals. The trapped melt crystallizes either
as adcumulate overgrowths, which are difficult to recog-
nize optically, or as discrete interstitial or reaction phases
(Bédard, 1994).
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro 41

Table 10. Partition coefficients between minerals and a melt,
and the modal compositions used in the model calculations
Pl, plagioclase; Cpx, clinopyroxene; Opx, orthopyroxene; Ol, ol-
ivine; Ox, oxides; Sp, spinel.
Based on the textural and major element data, the
Bikilal− Ghimbi olivine gabbro does not contain a signifi-
cant volume of trapped melt, and consequently, the trace
element calculations discussed above can be applied.
Published values of D for olivine, clinopyroxene, and pla-
gioclase (Rollinson, 1993; Green, 1994; Bindeman et al.,
1998; Kimura et al., 2002; see Table 10 for the values of
D used in the calculations), and the modal composition of
the olivine gabbros (Table 1) were used in our calcula-
tions. Magnetite and apatite were ignored, because they
are rare in the MgO−rich gabbros.
The bulk rock inversion is evaluated first by compar-
ing the recalculated composition of the clinopyroxene. At
pressures less than 2.5 GPa, the clinopyroxene is the prin-
cipal host for the REE and, therefore, can be used to cal-
culate the melt REE composition from the clinopyroxene
composition in cumulate and plutonic rocks, where the
melt is not preserved (e.g., Wood and Blundy, 1997). The
spidergram patterns calculated for the clinopyroxenes
(from Samples TCH 23 and BG 28) are similar to those
of intraplate−type tholeiites, as represented by the
Hawaiian Mauna Loa and Kilauea (Hilina) primitive tho-
leiitic to transitional basalts (Kimura et al., 2006), and
clearly differ from N− and E−MORBs or oceanic island
alkali basalts (Fig. 11a). The results show a good agree-
ment for the REE and Rb, but the abundances of Th, U,
Ta, and Nb are slightly elevated (Fig. 11a). As the D val-
ues for the Th to Nb region are very low in plagioclase
compared to other minerals, these elements could be se-
lectively concentrated in the interstitial melts. This would
eventually increase the concentrations of these elements

Figure 11. A primitive mantle (Sun and McDonough, 1989) normalized plot of: (a) the calculated clinopyroxene (Cpx) composition of the liq-
uids; the results are averaged from unaltered clinopyroxene analysis from olivine gabbro (TCH 23 and BG 28), and (b) the calculated melt
composition that produces the olivine gabbros (TCH 23 and BG 28). The N−MORB, E−MORB (Sun and McDonough, 1989), and OIT
fields, including the Hilina (tholeiitic and transitional) and Moana Loa (tholeiitic) after Kimura et al. (2006), are plotted for comparison.
 42 B.W. Woldemichael and J.−I. Kimura

Figure 12. A comparison of the recalculated solid composition of the OIT simulated using the COMAGMAT software package and the ob-
served composition of the Bikilal− Ghimbi gabbro. The low−pressure sequence of fractional (Fc) or equilibrium (Ec) crystallization fits well
(Figs. 12a−12e). Figure 12f shows a solidification temperature of 1100 −1160 °C, as indicted by the box. The different symbols denote differ-
ent pressures (1, 5, and 10 kbar), which can be reproduced from the initial tholeiite magma.

in the clinopyroxene. However, this process is not well
constrained because of a lack of information on the inter-
stitial melts in the gabbro. The calculated melt composi-
tions from the bulk rock composition and the bulk values
of D were normalized against those of a primitive mantle
(Fig. 11b). The estimated incompatible element patterns
of the calculated melt composition are also similar to
those of intraplate−type tholeiite to transitional basalts.
The olivine gabbros (Samples TCH 23 and BG 28) have a
convex upward pattern in their primitive mantle normal-
ized diagram, and also clearly different from E− and N−
MORB patterns (Fig. 11b). This further supports the re-
sults from the clinopyroxene. The Bikilal− Ghimbi paren-
tal basalt composition is identical to a typical intraplate−
type tholeiite overall, except for the elevated Th to Nb
contents (Fig. 11b). This may reflect a slight fluid metaso-
matism or an alteration of the clinopyroxene, as detected
in the altered clinopyroxenes in Sample TCH 23 (Fig. 8).
The olivine gabbros are apparently fresh, but traces of
alteration/metasomatism may be present. However, the
overall similarity between the estimated melt composi-
tions and the recalculated clinopyroxene compositions
strongly support an intraplate tholeiitic nature for the pa-
rental magma of the Bikilal− Ghimbi gabbro.
The above evidence suggests that the Bikilal− Ghimbi
gabbro was perhaps formed by the introduction of a man-
tle plume to generate intraplate−type tholeiitic magma
during the Neoproterozoic time in a continental or oce-
anic setting. Alternatively, it could also represent an intru-
sion in a marginal ocean basin between Gondwana frag-
ments during the Neoproterozoic time. This gabbro body
was subsequently accreted to a suture zone, or a plume
source magma intruded an accretionary suture. Reliable
age dating will provide a more conclusive argument con-
cerning the continental or oceanic setting, and will clar-
ify the pre−, syn−, or post−tectonic nature of the intrusion.
This is beyond the scope of this paper, and SHRIMP zir-
con age− dating is now in progress. The results will be
presented in a separate paper.
Crystallization conditions and magma replenishment
Mathematical simulations of the crystallization sequence
of the basalt magma using thermodynamic models have
recently become available. The computer software pro-
grams that can be used include MELTS (Ghiorso and
Sack, 1995) and COMAGMAT (Ariskin and Barmina,
2003). As MELTS does not accommodate pyroxene sta-
bility for the modeled intraplate−type tholeiitic magma,
the COMAGMAT program was used to calculate the
crystallization sequence. The starting melt composition
was set as a typical intraplate tholeiitic basalt, as repre-
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro
Figure 13. (a) Mineral mode simulation results for the fractional
crystallization of the OIT using the COMAGMAT software
package compared to the Bikilal− Ghimbi gabbroic intrusion
data for fractional (Fc) and equilibrium (Ec) crystallization at 1
kbar. (b) Mineral chemistry simulation results for fractional
crystallization of the OIT using the COMAGMAT software
package compared to the gabbroic intrusion data for fractional
(Fc) and equilibrium (Ec) crystallization at 1 kbar. The different
symbols denote different pressures (1, 5, and 10 kbar), which can
be reproduced from the initial tholeiite magma.
sented by the Hawaiian tholeiite (Wilson, 1989, Table 9.6,
Kilauea and Mauna Loa tholeiites).
The crystallization sequences differed according to
changes in pressure and oxygen fugacity. The best solu-
tion was achieved with the conditions of: (1) low pressure
(1 kbar); (2) high oxygen fugacity (~ QFM + 1); and (3)
two contrasting crystallization conditions, fractional crys-
tallization or equilibrium crystallization. The resulting
model calculations reproduced the mineralogical modes,
mineral compositions, and bulk composition of the
Bikilal− Ghimbi gabbro (Figs. 12 and 13). The COMAG-
MAT modal compositions were 14−18% olivine, 22−35%
clinopyroxene, and 54−58% plagioclase, which are similar
to the observed values of the Bikilal− Ghimbi gabbro
(Table 1). The calculated mineral compositions were Fo53−61
for olivine, Mg# = 75−70 for clinopyroxene, and An55−62
for plagioclase, all of which are also in good agreement
with the observed compositions of the Bikilal− Ghimbi
gabbro (Tables 3− 6). The calculated bulk solid composi-
tion is almost identical for TiO2, Al2O3, FeO, CaO, and
43
Figure 14. Apatite saturation conditions for the Bikilal− Ghimbi
gabbro. The source OIB with 1.00 wt% P2O5 was used in the
COMAGMAT software simulations. Apatite saturation can oc-
cur at 1095 °C and with a degree of crystallization of 80−90.
The precipitation temperature of apatite is lower than that of ol-
ivine, clinopyroxene, and plagioclase.
the alkali metals. However, some low MgO gabbros are
not identical, and might have different parameters due to
their evolved nature (e.g., temperature, pressure, and oxy-
gen fugacity) (see Fig. 12). These results support our con-
clusion that the Bikilal− Ghimbi olivine gabbro crystal-
lized from a composition that lay between a total fractional
and an equilibrium crystallization. The key intensive vari-
ables of the crystallization conditions are a low pressure
(1 kbar) and a temperature of 1100−1160 °C (Fig. 12f).
Apatite does not host alkali metals, FeO, or MgO.
Therefore, the Fe enrichment of the apatite−bearing sub-
sets is not an effect of the apatite concentration, but may
reflect the different activity of oxygen and water during
the shallow level magma chamber fractionation rather
than any fundamental difference in the chemistry of the
parental magma (Wilson, 1989). As noted above, judging
from the binary plot diagrams, the apatite crystallization
may have been caused by a mixing between a low−Mg
content gabbro from an evolved melt and a high−Mg pre-
cipitate from a mafic melt (Figs. 5 and 6). Such a high −
MgO end member could be derived from a less− evolved
magma that had a higher temperature (> 1160 °C; see Fig.
12).
It is known that the SiO2 and P2O5 contents and the
temperature are the controlling factors for apatite precipi-
tation (Watson, 1979). The SiO2 content of the fractionat-
ing melt remained at ~ 50 wt% during the tholeiitic crys-
tallization (COMAGMAT results not shown). Conse-
 44 B.W. Woldemichael and J.−I. Kimura
quently, the apatite saturation is a function of the P2O5
content of the melt and the temperature. The P2O5 content
of the mafic melt increases with increasing fractionation
(Fig. 14). From our model, P2O5 saturation occurs at
~ 1100 °C, and then apatite crystallization occurs at a
very late stage of crystallization from an evolved melt.
However, this ignores the role of the high−Mg high−P end
member deduced from the major elements (Fig. 5). In
contrast, if a high−temperature/high−MgO melt is forced
to cool, then apatite saturation occurs even when the P2O5
content in the melt is low, due to the downward shift of
the apatite saturation isotherm (see where P2O5 = ~ 1 wt%
at 1000 °C, Fig. 14). This mechanism accounts for the
mixing line observed in the apatite−bearing subsets. If an
unfractionated basaltic melt is input into a low tempera-
ture evolved melt, then the crystallization of apatite, along
with a high−MgO olivine, clinopyroxene, and perhaps a
calcic plagioclase would take place from the basalt melt
due to supercooling. The formation of an immiscible liq-
uid is a possible origin of the large amount of apatite
(> 20 modal percent) in the apatite−rich layers. Von
Gruenwaldt (1994) proposed that an immiscible Fe−Ti−
Ca−P liquid was periodically developed in the Bushveld
complex, and suggested that the development of the min-
eralized zones at the bottom of the body is the result of
the formation of an immiscible liquid by replenishment
and magma mixing. Crystallization of apatite and Fe−Ti
oxides in the Bikilal− Ghimbi gabbro is associated with
Mg−rich mafic minerals that cannot be accomplished by
an immiscible liquid alone. We believe that the supercool-
ing of the replenished unfractionated basalt melt is crucial
for the origin of the layering with a periodic formation of
an immiscible liquid. The combination of fractional crys-
tallization and magma replenishment has been reported
from many layered intrusions (e.g., Clark, 2004; Zhou et.
al, 2005), and evidence of supercooling of basaltic mag-
mas in layered intrusions has also been documented (e.g.,
Ballhaus and Glikson, 1989; Tegner et al., 1993).
CONCLUSIONS
The Bikilal− Ghimbi gabbro consists of apatite−free oliv-
ine gabbro, hornblende gabbro, and hornblendite suites
and their apatite−bearing equivalents. Despite the modal
differences between the olivine and the hornblende gab-
bros, there is no systematic chemical difference in the
major and trace element compositions, except for enrich-
ment in the fluid −mobile elements in the hornblende−
bearing suites. This suggests that the gabbros were gen-
erated from the same parental magma, unlike previous
interpretations, where the hornblende gabbro suites and
the olivine gabbros were separate intrusions. The horn-
blende−bearing suites contain uralitized hornblende and
plagioclase with kink bands. These facies occur only at
the perimeters of the gabbro body, and are structurally
concordant with the host gneiss.
Mineralogical constraints from the olivine and pla-
gioclase suggest an intraplate−type tholeiitic composition.
The estimation of the parental magma from the trace ele-
ment compositions in the clinopyroxene and bulk rock
also supports an intraplate tholeiitic nature of the parental
magma of the gabbros. Therefore, we conclude that the
Bikilal− Ghimbi gabbro formed by the introduction of a
mantle plume to generate a tholeiitic magma during Neo-
proterozoic time. Although the mineralogical data suggest
an intraplate setting, it cannot provide an unequivocal
conclusion given the apparent pre−, syn−, or post−tectonic
relationship of the intrusion with the surrounding gneissic
terrain.
Crystallization model calculations using a tholeiitic
parental magma suggest that the gabbros crystallized
from a basaltic magma, in which no interstitial melt re-
mains. The apatite−bearing subsets are likely to have orig-
inated from the supercooling of the replenished basalt that
precipitated both apatite and high−Mg mafic minerals.
Our petrological and geochemical examination of
this Neoproterozoic gabbro suggests that it is an intra-
plate−type tholeiite in composition. Isotope studies com-
bined with a reliable zircon SHRIMP age− dating tech-
nique will be important for clarifying the tectonic settings
of the gabbro, and this data will be presented in another
publication. A full reappraisal of the other gabbro bodies
in western Ethiopia is also necessary to reconstruct a reli-
able development history of the suture zone in the ANS.
ACKNOWLEDGMENTS
We greatly appreciate the support of the Geological
Survey of Ethiopia for logistical assistance in the field, for
provision of samples from the core and outcrop archives,
and for granting extended leave to B.W.W. for this study.
Dr. R.J. Stern of the University of Texas at Dallas, USA,
provided thoughtful comments on an early version of the
manuscript. This work was carried out at Shimane Univer-
sity with scholarship support from the Japanese Ministry
of Education, Culture, Sport, Science and Technology to
B.W.W. We also thank Drs. Y. Sawada and M. Akasaka of
Shimane University, Japan for free access to the XRF and
EPMA instruments and to Dr. BP Roser of Shimane
University, Japan for valuable comments on the draft man-
uscript. The authors are also grateful to an anonymous re-
viewer and Dr. Sakae Sano of Ehime University for their
valuable suggestions and critical comments regarding the
manuscript.
 Petrogenesis of the Neoproterozoic Bikilal− Ghimbi gabbro
SUPPLEMENTARY MATERIAL
Appendixes 1, 2 and 3 are available online from http://
www.jstage.jst.go.jp/browse/jmps.
REFERENCES
Abdelsalam, M.G. and Stern, R.J. (1996) Sutures and shear zones
in the Arabian Nubian Shield. Journal of African Earth
Sciences, 23, 289−310.
Abdelsalam, M.G., Liégeois, J.P. and Stern, R.J. (2002) The
Saharan Metacraton. Journal of African Earth Sciences, 34,
119−136.
Abraham, A. (1989) Tectonic history of the Pan−African low−
grade belt of western Ethiopia, Note 305. pp. 22, Ethiopian
Institute of Geological Survey, Addis Ababa, Ethiopia.
Alemu, T. and Abebe, T. (2000) Geology of the Ghimbi area,
Memoir 15. pp. 158, Geological Survey of Ethiopia, Addis
Ababa, Ethiopia.
Allen, A. and Tadesse, G. (2003) Geological setting and tectonic
subdivision of the Neoproterozoic orogenic belt of Tulu-
dimtu, western Ethiopia. Journal of African Earth Sciences,
36, 329−343.
Ariskin, A.A. and Barmina, G.S. (2004) COMAGMAT: Develop-
ment of a magma crystallization model and its petrological
applications. Geochemistry International, 42, 1−157.
Ayalew, T., Bell, K., Moore, J.M. and Parrish, R.R. (1990) U−Pb
and Rb−Sr geochronology of the Western Ethiopian Shield.
Geological Society of American Bulletin, 102, 1309−1316.
Ayalew, T. and Peccerillo, A. (1998) Petrology and geochemistry
of the Gore− Gambella plutonic rocks: implications for
magma genesis and the tectonic setting of the Pan−African
Orogenic Belt of western Ethiopia. Journal of African Earth
Sciences, 27, 397− 416.
Ballhaus, C.G. and Glikson, A.Y. (1989) Magma mixing and in-
traplutonic quenching in the Wingellina Hills intrusion,
Giles complex, central Australia. Journal of Petrology, 30,
1443−1469.
Beard, J. (1986) Characteristic mineralogy of arc related cumu-
late gabbros: implication for the tectonic setting of gabbroic
plutons and for andesite genesis. Geology, 14, 848−851.
Bédard, J.H. (1994) A procedure for the equilibrium distribution
of trace elements among the minerals of cumulate rocks and
the concentration of trace elements in coexisting liquids.
Chemical Geology, 118, 143−153.
Bédard, J.H. (2001) Parental magmas of the Nain Plutonic Suite
anorthosites and mafic cumulates: a trace element modeling
approach. Contributions to Mineralogy and Petrology, 141,
747−771.
Beshawered, E. (2001) Geology and phosphate mineralization of
Ghimbi−Bikilal area. AMEP Report 4, pp. 69, Geological
Survey of Ethiopia, Addis Ababa, Ethiopia.
Bindeman, I.N., Davis, A.M. and Drake, M.J. (1998) Ion micro-
probe study of plagioclase− basalt partition experiments at
natural concentration levels of trace elements. Geochimica
et Cosmochimica Acta, 62, 1175−1193.
Claeson, D.T. and Meurer, W.P. (2004) Fractional crystallization
of hydrous basaltic “arc−type” magmas and the formation of
amphibole−bearing gabbroic cumulates. Contributions to
Mineralogy and Petrology, 147, 288−304.
Clark, A.H. (2004) Fe−Ti−P Oxide melts generated through
45
magma mixing in the Antauta subvolcanic center, Peru: im-
plications for the origin of nelsonite and iron oxide− domi-
nated hydrothermal deposits. Economic Geology, 99, 377−
395.
Cox, K.G., Bell, J.D. and Pankhurst, R.J. (1979) The interpreta-
tion of igneous rocks. pp. 450, Allen and Unwin, London.
Deer, W.A., Howie, R.A. and Zussman, J. (1992) An introduction
to rock forming minerals. pp. 696, Pearson and Prentice
Hall, London.
Ghiorso, M.S. and Sack, R.O. (1995) Chemical mass−transfer in
magmatic processes IV. A revised and internally consistent
thermodynamic model for the interpolation and extrapola-
tion of liquid−solid equilibria in magmatic systems at ele-
vated temperatures and pressures. Contributions to Miner-
alogy and Petrology, 119, 197−212.
Green, T.H. (1994) Experimental studies of trace− element parti-
tioning applicable to igneous petrogenesis − Sedona 16 years
later. Chemical Geology, 117, 1−36.
Grenne, T., Pedersen, R.B., Bjerkgård, T., Braathen, A., Selassie,
M.G. and Worku, T. (2003) Neoproterozoic evolution of
western Ethiopia: igneous geochemistry, isotope systematics
and U−Pb ages. Geological Magazine, 140, 373−395.
Hunter, R.H. (1996) Texture developments in cumulate rocks: In
Developments in petrology 15, layered intrusions (Cawthorn,
R.G. Ed.). pp.531, Elsevier Science B.V., Amsterdam, 77−101.
Irvine, T.A. and Baragar, W.R.A. (1971) A guide to the chemical
classification of the common rocks. Canadian Journal of
Earth Sciences, 8, 523−548.
Irvine, T.N. (1982) Terminology for layered intrusions. Journal of
Petrology, 23, 127−162.
Johnson, T.E., Ayalew, T., Mogessie, A., Kruger, F.J. and Poujol,
M. (2004) Constraints on the tectonometamorphic evolution
of the Western Ethiopia Shield. Precambrian Research, 133,
305−327.
Kebede, T., Koeberl, C. and Koller, F. (1999) Geology, geochem-
istry and petrogenesis of intrusive rocks of the Wallagga
area, western Ethiopia. Journal of African Earth Sciences,
29, 715−734.
Kebede, T., Kloetzli, U.S. and Koeberl, C. (2001) U/Pb and Pb/Pb
zircon ages from granitoid rocks of Wallagga area: con-
straints on magmatic and tectonic evolution of Precambrian
rocks of western Ethiopia. Mineralogy and Petrology, 71,
251−271.
Kebede, T. and Koeberl, C. (2003) Petrogenesis of A−type granit-
oids from the Wallagga area, western Ethiopia: constraints
from mineralogy, bulk−rock chemistry, Nd and Sr isotopic
compositions. Precambrian Research, 121, 1−24.
Kennedy, W.Q. (1964) The structural differentiation of Africa in
the Pan−African ±500 m.y. tectonic episode, 8th annual re-
port on scientific results, pp. 50, University of Leeds, Leeds,
U.K.
Kimura, J.−I., Yoshida, T. and Takakau, Y. (1995) Igneous rock
analysis using ICP−MS with internal standardization, isobaric
ion overlap correction, and standard addition methods.
Science Report of Fukushima University, Fukushima, Japan,
56, 1−12.
Kimura, J.−I. and Yamada, Y. (1996) Evaluation of major and
trace element XRF analysis using a flux to sample ratio of
two to one glass beads. Journal of Mineralogy, Petrology
and Economic Geology, 91, 62 −72.
Kimura, J.−I., Danhara, T. and Iwano, H. (2000) A preliminary re-
port on trace element determination of zircon and apatite
 46 B.W. Woldemichael and J.−I. Kimura
crystals using Excimer Laser Ablation−Inductively Coupled
Plasma Mass Spectrometry ExLA−ICPMS. Fission Track
News Letter, 13, 11−20.
Kimura, J.−I., Yoshida, T. and Izumi, S. (2002) Origin of low−K
intermediate lavas at Nekoma volcano, NE Honshu arc,
Japan: geochemical constraints for lower − crustal melts. Jour-
nal of Petrology, 43, 631− 661.
Kimura, J.−I., Sisson, T.W., Nakano, N., Coombs, M.L. and
Lipman, P.W. (2006) Geochemistry of early Kilauea mag-
mas from the submarine Hilina bench: the nature of the
Hilina mantle component. Journal of Volcanology and Geo-
thermal Research, 151, 51−72.
Kröner, A. (1984) Late Precambrian plate tectonics and orogeny:
a need to redefine the term Pan −African. In African Geology
(Klerkx, J. and Michot, J. Eds.). Tervuren, Mus´ee R. L’,
Afrique Centrale, Belgium , 23−28.
Kumar, K.V., Frost, C.D., Frost, R. and Chamberlain, K.R. (2007)
The Chimakurti, Errakonda, and Uppalapadu plutons, East-
ern Ghats Belt, India: an unusual association of tholeiitic
and alkaline magmatism. Lithos, 97, 30−57.
Kuno, H. (1968) Differentiation of basalt magmas. In Basalts:
The Poldervaart treatise on rocks of basaltic composition 2
(Hess, A. Ed.). pp. 862, Inter science, New York, 623− 688.
Leake, B.E., Woolley, A.R., Leake, B.E., Woolley, A.R., Arps,
C.E.S., Birch, W.D., Gilbert, M.C., Grice, J.D., Hawthorne,
F.C., Kato, A., Kisch, H.J., Krivovichev, V.G., Linthout, K.,
Laird, J., Mandarino, J.A., Maresch, W.V., Nickel, E.H.,
Rock, N.M.S., Schumacher, J.C., Smith, D.C., Stephenson,
N.C.N., Ungaretti, L., Whittaker, E.J.W. and Youzhi, G.
(1997) Nomenclature of amphiboles: report of the subcom-
mittee on amphiboles of the International Mineralogical
Association, Commission on New Minerals and Mineral
Names. The Canadian Mineralogist, 35, 219−246.
Naslund, H.R. and McBirney, A.R. (1996) Mechanism of forma-
tion of igneous layering. In Developments in petrology 15,
layered intrusions (Cawthorn, R.G. Ed.). pp.531, Elsevier
Science B.V., Amsterdam, 1 − 43.
Reed, S.J.B. (1975) Electron microprobe analysis. pp. 400, Cam-
bridge University Press, New York.
Rollinson, H. (1993) Using geochemical data, evaluation, presen-
tation and interpretation. pp. 352, Wiley, New York.
Stern, R.J. (1994) Arc assembly and continental collision in the
Neoproterozoic East African Orogen: implications for the
View publication stats
consolidation of Gondwanaland. Annual Review of Earth
and Planetary Sciences, 22, 319−351.
Sun, S.S. and McDonough, W.F. (1989) Chemical and isotopic
systematic of oceanic basalts: implications for mantle com-
position and processes. In Magmatism in the ocean basins,
Geological Society special publication 42, (Saunders A.D.
and Norry M.J. Eds.). pp. 398, Blackwell Scientific Publish-
ing, London, 313−345.
Tegner, C., Wilson, J.R. and Brooks, C.K. (1993) Intraplutonic
quench zones in the Kap Edvard Holm Layered Gabbro
Complex, east Greenland. Journal of Petrology, 34, 681−710.
Tepper, J. (1996) Petrology of mafic plutons associated with calc −
alkaline granitoids, Chilliwack batholith, North Cascades,
Washington. Journal of Petrology, 37, 1409−1436.
Von Gruenwaldt, G. (1994) Ilmenite−apatite enrichment in the
upper zone of the Bushveld complex: a major titanium−rock
phosphate resource. International Geology Review, 35, 987−
1000.
Wager, L.R. and Brown, G.M. (1968) Layered igneous rocks. pp.
588, Oliver and Boyd, Edinburgh.
Watson, E.B. (1979) Apatite saturation in basic to intermediate
magmas. Geophysical Research Letters, 6, 937−940.
Wilson, M. (1989) Igneous petrogenesis: A global tectonic ap-
proach. pp. 466, Chapman Hall, London.
Wood, J.B. and Blundy, J.D. (1997) A predictive model for rare
earth element partitioning between clinopyroxene and anhy-
drous silicate melt. Contributions to Mineralogy and Petro-
logy, 129, 166 −181
Worku, H. and Schandelmeier, H.J. (1996) Tectonic evolution of
the Neoproterozoic Adola Belt of southern Ethiopia: evi-
dence for a Wilson Cycle process and implications for ob-
lique plate collision. Precambrian Research, 77, 179−210.
Zhou, M.−F., Robinson, P.T., Lesher, C.M., Keays, R.R., Zhang, C.−
J. and Malpas, J. (2005) Geochemistry, petrogenesis and
metallogenesis of the Panzhihua gabbroic layered intrusion
and associated Fe−Ti−V oxide deposits, Sichuan Province.
SW China Journal of Petrology, 46, 2253 − 2280.
Manuscript received April 1, 2007
Manuscript accepted October 4, 2007
Manuscript handled by Hirokazu Fujimaki