Ca-Ba isotopes reveal distinct mantle metasomatism in the mantle source of ultrapotassic magmas

As powerful geochemical tracers, Ca and Ba isotopic systems provide complementary constraints on the genesis of mantle- and crust-derived igneous rocks. However, the extent to which Ca isotopes can uniquely track recycled sedimentary carbonates remains debated, and Ba isotopes have rarely been applied to the deep carbon cycle despite their potential sensitivity to recycled components. Italian mantle-derived ultrapotassic volcanic rocks, spanning silica-oversaturated to silica-undersaturated compositions, offer an unique opportunity to evaluate the diagnostic value of combined Ca–Ba isotopes and to constrain metasomatized mantle sources. Here we report new Ca and Ba isotope data for a suite of well-characterized Italian ultrapotassic rocks. They exhibit MORB-like δ138/134Ba values varying from -0.11‰ to 0.07‰, and δ44/40Ca values of 0.75 ± 0.08‰, lower than MORB on average; one sample (SVK 08) is an outlier with unusually low δ44/40Ca values (0.43‰). Petrographic and geochemical constraints indicate that low-temperature alteration and crustal contamination are unlikely to dominate the observed Ca–Ba isotope systematics, and the combined evidence suggests limited control by diffusion and shallow magmatic differentiation. Instead, the Ca–Ba isotope variations are best interpreted to reflect heterogeneity in metasomatized mantle sources. For the silica-undersaturated rocks, light δ44/40Ca values are consistent with contributions from Ca-rich recycled components (including carbonate-bearing sediments), although low δ44/40Ca is not uniquely diagnostic of carbonate recycling; localized garnet-related effects may contribute to extreme δ44/40Ca values in the kamafugitic outlier (i.e., SVK 08). For the silica-oversaturated lamproites, comparably light δ44/40Ca can be plausibly explained by melting of a K-richterite–bearing (MARID-type) metasomatized source. Two-endmember mixing calculations provide non-unique but feasible bounds indicating that minor sediment additions can reproduce the observed δ138/134Ba range for most samples, whereas Ba isotopes alone do not uniquely resolve carbonate-specific contributions. Finally, the partial decoupling between δ44/40Ca and δ138/134Ba within the Roman Province silica-undersaturated rocks implies variable recycled-agent proportions and/or compositions among mantle domains at an intra-province (i.e., sub-regional) scale. Together, these results highlight both the strengths and limitations of Ca–Ba isotopes: Ca isotopes provide quantitative leverage on Ca-rich recycled inputs when integrated with independent tracers, whereas Ba isotopes more robustly record bulk sediment recycling but are less diagnostic of carbonate-specific contributions in ultrapotassic systems.