Whole-rock and Sm-Nd isotopic data of the main units of the Chilas
zoned ultramafic bodies (Kohistan paleo-island arc, NW Pakistan)
indicate that ultramafic rocks and gabbronorite sequences stem from a
common magma. However, field observations rule out formation of both
ultramafic and mafic sequences in terms of gravitational crystal
settling in a large magma chamber. Contacts between ultramafic and
gabbronorite sequences show emplacement of the dunitic bodies into a
semi-consolidated gabbronoritic crystal-mush, which in turn has
intruded and reacted with the ultramafic rocks to produce concentric
zoning. Field and petrological observations indicate a replacive origin
of the dunite. Bulk Mg#'s of dunitic rocks range from 0.87-0.81
indicating that the dunite-forming melt underwent substantial
fractionation-differentiation and that percolative fractional
crystallization probably generated the dunitic core.
The REE chemistry of clinopyroxene in primitive dunite samples and the
Nd isotopic composition of ultramafic rocks are in equilibrium with the
surrounding gabbronorite. Accordingly, liquids that formed the dunitic
rocks and later the mafic sequence derived from a similar depleted
source (epsilon(Nd)similar to 4.8). We propose a mechanism for the
comagmatic emplacement, where km-scale ultramafic bodies represent
continuous channels reaching down into the upper mantle. The
melt-filled porosity in these melt channels diminishes the
mean-depth-integrated density difference to the surrounding rocks. Due
to buoyancy forces, melt channels raise into the overlying crustal
sequence. In the light of such processes, the ultramafic bodies are
interpreted as melt channels through which the Chilas gabbronorite
sequence was fed. The estimated basaltic-andesitic, low Mg# (similar
to 0.53) bulk composition of the Chilas gabbronorite sequence closely
matches estimates of lower crustal compositions. Since the mafic
sequence originated from a primary, high Mg# (> 0.7) basaltic arc
magma, differentiation of such high Mg# magmas within km-scale
isolated melt conduits may explain the ``Mg#-gap'' between bulk
estimates of the continental crust and primary basaltic magmas, a major
paradox in the andesite model of crust formation. (c) 2005 Elsevier
B.V. All rights reserved.