(a) Plot of the host lavas relative to various mantle compositional components (green star) and fields for basalts from different tectonic settings as defined by Weaver (1991) and Condie (2005); (b) Plot of Al versus Mn for the olivine xenocrysts; (c) Comparison of Bini Warack basanites and basalts with experimental melts of representative mantle lithologies using their TiO₂ and FeO_t contents; (d) TiO₂ versus V plot of Bini basanites and basalts. In panel (a), DEP, deep depleted mantle; EM1, enriched mantle sources; HIMU, high U/Pb mantle source; OIB, oceanic island basalt; PM, primitive mantle; REC, recycled component. In panel (b), the fields of spinel–garnet lherzolite (dashed green line), garnet–lherzolite (dashed red line), and spinel–lherzolite (yellow line) are after Yuan and Yan (2022). The green ellipse in panel (c) denotes the low-degree (<10%) experimental melts of fertile lherzolite. The black circles illustrate partial melts of fertile lherzolite, while the red ones highlight experimental melts formed at 1.5–2.5 GPa, just as the black dashes represent the composition of harzburgite partial melts and the red ones that of their experimental melt at 1.5–2.5 GPa. The thick red arrow schematically shows non-peridotite mantle rocks (e.g., amphibolite, silica-deficient, and silica-excess pyroxenites) typically with high TiO₂ contents (≥2 wt%; Kushiro, 1996). Data for experimental melts of lherzolite and harzburgite are consistent with the compilation of Dai et al. (2023). In panel (d), the compositional zones for volcanic rocks from various tectonic settings and the magmatic differentiation trend caused by Fe–Ti oxide fractionation are from Reagan et al. (2010). FAB, fore-arc basalts; EPR, East Pacific Rise.