Assessment of relative T i, T a, and N b ( TITAN ) enrichments in ocean island basalts

Abstract The sensitivity of trace element concentrations to processes governing solid‐melt interactions has made them valuable tools for tracing the effects of partial melting, fractional crystallization, metasomatism, and similar processes on the composition of a parental melt. Recent studies of ocean island basalts (OIB) have sought to correlate Ti, Ta, and Nb (TITAN) anomalies to isotopic tracers, such as 3 He/ 4 He and 187 Os/ 188 Os ratios, which may trace primordial deep mantle sources. A new compilation of global OIB trace element abundance data indicates that positive TITAN anomalies, though statistically pervasive features of OIB, may not be compositional features of their mantle sources. OIB show a range of Ti (Ti/Ti* = 0.28–2.35), Ta (Ta/Ta* = 0.11–93.4), and Nb (Nb/Nb* = 0.13–17.8) anomalies that show negligible correlations with 3 He/ 4 He ratios, indicating that TITAN anomalies are not derived from the less‐degassed mantle source traced by high‐ 3 He/ 4 He. Positive TITAN anomalies can be modeled using variable degrees (0.1–10%) of nonmodal batch partial melting of garnet‐spinel lherzolite at temperatures and pressures considered typical for OIB petrogenesis, and subjecting this partial melt to fractional crystallization and assimilation of mid‐ocean ridge basalt‐like crust (AFC). Correlations of TITAN anomalies with modal abundances of olivine and clinopyroxene in porphyritic Canary Islands lavas provide empirical support for this process and indicate that high abundances of these phases in OIB may create misleading trace element anomalies on primitive mantle‐normalized spider diagrams. Because partial melting and AFC are common to all mantle‐derived magmas, caution should be used when attributing TITAN anomalies to direct sampling of recycled or deep mantle sources by hotspots.