Hilary Downes


Spinel and garnet pyroxenites form an integral part of the shallow sub-continental lithospheric mantle. They occur as part of spinel peridotite xenolith suites (sometimes as composite xenoliths with peridotites) or as layers or dykes within tectonically-emplaced ultramafic massifs. Judging from their abundances in mantle xenolith suites and ultramafic massifs, pyroxenites form approximately 1-5% of the directly sampled upper mantle. Lithologies range from orthopyroxenite through websterites to clinopyroxenite, with or without olivine. Pyroxenites are defined as ultramafic rocks containing >60% modal pyroxene. Spinel, garnet and olivine are the most common varietal minerals. Relative proportions of the major minerals in pyroxenites are extremely variable, ranging from orthopyroxenites, through websterites, to clinopyroxenites. Minerals such as amphibole and biotite may be present as the result of metasomatism. Less commonly, plagioclase, corundum, sapphirine, ilmenite, apatite, zircon and graphite can be present. From field evidence, it is likely that pyroxenites were formed prior to emplacement of the ultramafic massif into the continental crust.
The origin of mantle pyroxenites is controversial. Early work by Dick and Sinton (1979) suggested that they were metamorphic segregations of the host peridotite, formed by dissolution and precipitation of pyroxene during plastic flow. In contrast, many workers considered that pyroxenites are formed by crystal precipitation from asthenosphere-derived silicate magmas passing through the lithosphere. The mechanism of liquid-crystal separation was considered to be dynamic flow crystallisation (Irving,1980) or filter-pressing. Conversely, Allègre and Turcotte (1986) suggested that pyroxenites are remnants of subducted oceanic crust, streaked out into thin layers by mantle convection. An alternative hypothesis put forward by Pearson et al. (1993) suggested that some pyroxenites represent high-pressure crystal segregates from magmas derived from melting of subducted ocean crust. Garrido and Bodinier (1999) highlighted the importance of melt-rock replacement reactions between older pyroxenites, peridotites and percolating melts. Finally, an arc setting has been suggested for some pyroxenites, e.g. in Cabo Ortegal. This paper presents a review of petrological and whole-rock geochemical data, together with isotope and mineral trace element results for pyroxenites from shallow mantle xenolith suites and ultramafic massifs, to investigate and test these various hypotheses.
Pyroxenite display a wide range of major and trace element compositions and tend to show variation between at least three end-members: one is similar to clinopyroxene in composition, the second has high MgO and is similar to orthopyroxene, while the third component has concentrations of MgO, CaO, Al2O3, TiO2 and Na2O that are close to basaltic levels. Only a few mantle pyroxenites resemble bulk oceanic crust. Most pyroxenites have LREE-depleted whole-rock REE patterns, although with lower absolute concentrations than MORB or bulk oceanic crust, and most lack Eu anomalies. Trace element patterns in clinopyroxenes from pyroxenites are variable. Those from massifs are usually LREE-depleted, whereas those in xenoliths are frequently LREE-enriched. Most show depletions in Zr, Hf and Sr, although those from the Cabo Ortegal complex (NW Spain) display significant Sr peaks, attributed to the influence of subduction-related fluids or melts.
d18O values for clinopyroxenes from mantle pyroxenites tend to be in the range 5-5.8 per mil, i.e. typical mantle values. Rare garnet pyroxenite layers from a few ultramafic massifs show higher values that may reflect recycling of a crustal component. Similarly, most Sr, Nd, Pb and Hf isotopic ratios from pyroxenites fall in the depleted mantle field, although with a slightly wider range of values than is typical for shallow sub-continental mantle peridotites. Some pyroxenites show higher 87Sr/86Sr values and lower eHf than peridotites, indicating the presence of an enriched component. The majority of mantle pyroxenites are most easily explained as products of crystal accumulation of tholeiitic and (more rarely) alkaline mantle-derived magmas, together with variable amounts of trapped interstitial magma. Others may be the product of interaction between magma and peridotite wall-rock. Only samples from a few massifs (e.g. garnet pyroxenites from Beni Bousera and Ronda) display strong evidence of being subducted recycled oceanic crust. Even fewer mantle pyroxenites show good evidence for derivation by either incipient melting of the host peridotite or metamorphic segregation.

Full Text:


DOI: https://doi.org/10.4454/ofioliti.v30i2.260