Helium–oxygen–osmium isotopic and elemental constraints on the mantle sources of the Deccan Traps

Abstract

The Deccan Traps, a 65 million-year-old continental flood basalt province located in western India, is the result of one of the largest short-lived magmatic events to have occurred on Earth. The nature and composition of its mantle source(s), however, have been difficult to resolve due to extensive assimilation of continental crust into the ascending Traps magmas. To circumvent this issue, using high-precision electron microprobe analysis, we have analyzed olivine grains from MgO-rich (up to 15.7 wt.%) lavas that likely erupted before substantial crustal assimilation occurred. We compare olivine, pyroxene and plagioclase mineral chemistry and He–O–Os isotope compositions with bulk rock major- and trace-element abundances and 187Os/188Os for both bulk-rocks and mineral separates. Helium isotope compositions for the olivine grains generally show strong influence from crustal assimilation (<3 RA), but one ankaramite from the Pavagadh volcanic complex has a 3He/4He ratio of 10.7 RA, which is slightly lower than the range of 3He/4He measured for present-day Réunion Island volcanism (∼12–14 RA). Olivine-dominated mineral separates span a more restricted range in 187Os/188Os (0.1267 to 0.1443) compared with their host lavas (0.1186 to 0.5010), with the separates reflecting a parental magma composition less affected by lithospheric or crustal interaction than for the bulk-rocks. Despite significant He–Os isotopic variations, Δ17O is relatively invariant (−0.008±0.014‰) and indistinguishable from the bulk mantle, consistent with high-3He/4He hotspots measured to-date.Compositions of olivine grains indicate the presence of up to 25% of a pyroxenite source for Deccan parental magmas, in good agreement with ∼20% predicted from isotopic data for the same samples. Modeled pyroxenite signatures are similar to geochemical signatures expected to arise due to other types of mantle differentiation or due to assimilation of continental crust; however, we show that crustal assimilation cannot account for all of the compositional features of the olivine. Weak correlations exist between a global compilation of Xpx (Deccan: 0.2–0.7) and 3He/4He, δ18O (Deccan olivine: 4.9–5.2‰) and 187Os/188Os. Robust relationships between these parameters may be precluded due to a lack of two-reservoir source mixing, instead involving multiple mantle domains with distinct compositions, or because Xpx may reflect both source features and crustal assimilation. Notwithstanding, geochemical similarities exist between Deccan Traps olivine (3He/4He = 10.7 RA; 187Os/188Osi = 0.1313 ± 45, 2σ) and Réunion igneous rocks (3He/4He = 12–14 RA; 187Os/188Osi = 0.1324 ± 14). These relationships imply that a characteristic geochemical ‘fingerprint’ may have persisted in the mantle plume that fed the Deccan Traps, since its inception at 65 Ma, to ongoing eruptions occurring on Réunion up to the present-day.