Calcium isotopic variability in hotspot lavas controlled by partial melting and source lithological heterogeneity

Crustal recycling can induce significant chemical and lithological heterogeneity of the mantle. Ocean island basalts (OIBs) are the surface expressions of rising mantle plumes from the Earth's deep interior and thus can be a probe of the compositional heterogeneity in the deep mantle. Endmember OIBs sampled the most isotopically extreme mantle components including EM1 (Enriched mantle 1), EM2 (Enriched mantle 2), HIMU (high mu; mu = U-238/Pb-204), and FOZO (Focus zone), but the lithological properties of their source regions are still not well understood. Here we explore the potential of stable calcium (Ca) isotopes in identifying mantle source lithology by investigating OIBs from the type localities for the several mantle components. We find that the delta Ca-44/40 values of classic HIMU OIBs (0.77 +/- 0.09 parts per thousand; 2SD, N = 15) and Louisville FOZO OIBs (0.78 +/- 0.06 parts per thousand; 2SD, N = 4) are indistinguishable from the average MORB (mid-ocean ridge basalt) value (0.84 +/- 0.09 parts per thousand; 2SD, N = 31), while the delta Ca-44/40 values of the Pitcairn EM1 (0.65-0.72 parts per thousand) and the Samoan (Tutuila) EM2 OIBs (0.61-0.71 parts per thousand) are distinctly lower than those of MORBs. Such Ca isotopic distinction cannot be attributed to magmatic differentiation because the EM1 and EM2 OIBs have obviously lower delta Ca-44/40 values than the HIMU and FOZO OIBs even at a same MgO content. Equilibrium fractionation during partial melting of garnet peridotite with bulk silicate Earth (BSE) delta Ca-44/40 (0.94 +/- 0.05 parts per thousand) can explain the Ca isotopic compositions of the HIMU and FOZO OIBs but is unable to reproduce the Ca isotopic variation of all OIBs in this study. Quantitative modeling suggests that the coupling of low delta Ca-44/40 (< 0.7 parts per thousand) and high Gd/Yb (> 3.5) ratios in the EM1 and EM2 OIBs most likely points to an eclogitic source component with MORB-like or even lower delta Ca-44/40 (e.g., < 0.8 parts per thousand), possibly reflecting contributions from recycled oceanic crust and sediments. A negative correlation between average delta Ca-44/40 and delta Fe-57 values (a useful proxy for mantle source lithology) of all endmember OIBs further confirms this proposal and highlights the significant contribution of distinct source lithologies to the Ca isotopic variations in these basalts. Therefore, our study indicates that Ca isotopes of basalts can be another promising tracer of lithological heterogeneity in the mantle.