Mixing of Mantle Sources Preserved in Melt Inclusions at Coaxial Segment of the Juan de Fuca Ridge

Coaxial segment of the Juan de Fuca Ridge (JdFR) has a poorly defined axial rift, a wide rift valley, and erupts primarily pillow lavas, suggesting relatively low magma supply compared to more magmatically robust ridge segments to the south. To evaluate magma supply at Coaxial, we have analyzed olivine-hosted melt inclusions (MIs; N=113) and host glasses (N=7) for major, trace and volatile element contents, which we compare with new and existing data from Vance and Cleft segments of the JdFR. Of the three segments, Coaxial has the most diverse range of MI compositions. We divide the Coaxial MIs into two categories based on major and trace element systematics: NMORB (La/Sm > 0.5 and K2O/TiO2 > 0.05; N=68) and DMORB (La/Sm ≤ 0.5 and K2O/TiO2 ≤ 0.05; N=45), with the NMORB inclusions more closely resembling those of Cleft and Vance. DMORB inclusions are also characterized by higher mean FeO/MgO (0.93) and H2O/Ce (0.076) compared to NMORB (FeO/MgO =1.1 and H2O/Ce = 0.032). As a whole, the MIs from Coaxial have relatively low equilibration depths in the crust (average of 1656 m; determined by vapor saturation pressures of CO2-H2O) compared to the depth of the seismically imaged melt lens (2329 m; Carbotte et al., 2008). However, when considered separately, NMORB have peak MI depths consistent with the melt lens, similar to the other segments investigated, while the DMORB MI have a shallower peak resulting from lower CO2 concentrations. We suggest that the lower La/Sm and CO2/Nb ratios in the DMORB indicate the presence of a depleted mantle source, consistent with reported mantle values (e.g., Shimizu et al., 2016) that result in anomalously low equilibration pressures. However, the lower average FeO/MgO of the DMORB MIs support their early, likely deep, entrapment compared to the majority of NMORB inclusions. While DMORB compositions have been explained by high extents of melting of a normal mantle component, we favor a hypothesis where MIs are derived by mixing a normal depleted MORB mantle component, similar to that sampled at Cleft and Vance segments, with a previously depleted mantle component.