Tracing the sublithospheric sources of continental flood basalts: multi-elemental isotopic studies on
the recently found ferropicrites and meimechites from the Karoo large igneous province.
There is a substantial debate on the mantle sources and ultimate origins of continental flood basalts
(CFBs), e.g., whether they are related to deep-seated thermal upwellings (i.e. mantle plumes) or not.
The Karoo CFB province that was emplaced on the juxtaposed land masses of Africa and
Antarctica during the early stages of the breakup of the Gondwana supercontinent ~180 Ma ago has
played a central role in this debate. Although most of the structural analyses, geochemical affinities,
and temporal relationships of Karoo-related rocks point to a strong control of lithosphere on the
magmatism, paleostress estimates for some of the Karoo dikes and high mantle potential
temperatures estimated for the sources of recently found highly magnesian rocks from Antarctica
are compatible with the plume theory.
Volcanic rocks that do not show evidence of lithospheric influence in their geochemistry are
extremely rare in the African part of the Karoo province. Here we present high-precision isotopic
(Sr, Nd, Pb, and Os) whole-rock data on some primitive dike rocks (ferropicrites and meimechites)
associated with its Antarctic extension. The isotopic data together with trace element data show that
the parental melts of the studied rocks sampled two distinctive geochemical reservoirs in the deep
sub-Gondwanan mantle. The isotopic signatures of the relatively depleted types show evidence of
extensive melt extraction in the past and are indistinguishable from those of mid-ocean ridge basalts
(MORBs) of the SW Indian Ridge, the modern successor of the Jurassic Africa-Antarctica rift. On
the other hand, the relatively enriched type isotopically resembles modern oceanic island basalts
(OIBs) and may sample pyroxenitic sources either formed by melt infiltration in the upper mantle or
by reaction of peridotite with recycled oceanic crustal components. Recent Ar-Ar datings of the
depleted types indicate that they are related to the main phase of Karoo magmatism ~180 Ma ago.
Moreover, preliminary modeling implies that many of the Antarctic Karoo flood basalts originated
from the same mantle source with the depleted types prior to being contaminated by lithospheric
materials.
Although the anomalously hot mantle sources (mantle potential temperatures >1600 °C) estimated
for the depleted types are compatible with a plume origin, their MORB-like isotopic characteristics
are unusual for plume-derived rocks. Instead, our findings are more concordant with a recent model
that suggests the generation of the Karoo CFBs in an extensive melting episode caused largely by
internal heating of the upper mantle beneath the Gondwana supercontinent. The story is not
complete, however: the next phases of geochemical research will concentrate on He, Li, and B
isotopic signatures of these extraordinary rocks. Helium will hopefully shed more light on the
plume problem and Li and B are expected to reveal if the sub-Gondwanan mantle entrained
subduction-related geochemical signatures.
Traces of Recycled Fe-Ti Gabbros in the Sources of Ferropicrites
Ferropicrites are sub-alkaline or mildly alkaline primitive magmatic rocks (MgO = 12-18 wt. \%)
characterized by exceptionally high FeO$_{tot}$ contents ($>$ 13 wt. \%) compared to other
picrites. They are related to continental flood basalt (CFB) provinces and are commonly found as
relatively thin basal lava flows (e.g., Paraná-Etendeka) or as dikes with ambiguous age relationships
(e.g., Karoo). Ferropicrites have recently been in the focus of large igneous province (LIP) research
because they usually are nearly uncontaminated and provide important geochemical information on
the asthenospheric mantle. In addition, near-primary ferropicrites are characterized by peculiar
combination of high Fe-contents and primitive Mg-rich olivines (up to Fo$_{88}$) indicative of
high mantle potential temperatures ($\sim$1600 \degC) relative to mid-ocean ridges (MORs) thus
bearing evidence for mantle plume sources in LIPs.
High Sm/Yb ratios and low Al$_{2}$O$_{3}$ contents indicate that garnet was present in the
mantle residue of the ferropicrite sources. In addition, high Ni contents of ferropicrites and melting
experiments on pyroxenites and ferropicritic starting materials indicate that their source was
dominated by olivine-free pyroxenite. According to a popular theory, this kind of garnet-pyroxenite
source generates by partial melting of relatively fusible eclogite inclusions that produce SiO$_{2}$rich melts and consume olivine in the host garnet peridotite. Most of pyroxenite-derived picrites
from, e.g., Hawaii, have lower FeO$_{tot}$ contents compared to ferropicrites, however. This
raises the question as to whether the generation of ferropicritic liquids requires specific melting
conditions or exceptionally Fe-rich pyroxenite composition, or both.
Given that (1) oceanic crust is heterogeneous in terms of major and trace element geochemistry, (2)
eclogitization does not significantly affect the geochemistry of subducted oceanic crust, and (3)
crustal signature is expected to be at least partly inherited into partial melts derived from eclogiteperidotite mixtures, we have modeled the behaviour of REE and V during melting of theoretical
mantle pyroxenite with inherited eclogite signatures from distinct parts of subducted oceanic crust
in order to explain the geochemical differences between ferropicrites and common picrites. Firstly,
we modeled the partial melting of eclogite assuming 50\% modal batch melting with
clinopyroxene/garnet ratio of 9:1. Partial melting of secondary pyroxenite (eclogite/peridotite = 1:2)
was modeled assuming modal batch melting with source modes corresponding to 2.5 GPa and 5.0
GPa conditions. We emphasize that the model parameters have been adopted directly from
experimental studies and the compositional data for the different components represent averages or
typical values.
Our results strongly suggest that many ferropicrites with high V/Lu ratios contain traces of
subducted Fe-Ti gabbros whereas the geochemistry of many common picrites can be explained by
partial melting of mantle pyroxenite with traces of average MORB-signatured eclogite. Some
ferropicrites with relatively lower V/Lu may have been generated from ferrobasalt-signatured
mantle sources or may not represent primitive ferropicritic melts at all. Interestingly, some mildly
alkaline picrites from the rejuvenated parts of the Hawaii volcanic chain have high FeOtot and V/Lu
and may represent oceanic correlatives of ferropicrites.