Another hypothesis for the source of the Columbia River Flood Basalts involves a mantle plume that has entrained slivers of old oceanic crust (Takahahshi et al, 1998). As the heterogeneous plume rises towards the surface, the fertile oceanic crust slivers are the first to melt. This melt ponds in the lower crust before erupting to the surface. This combination of events can produce a melt that matches the TiO2 and K2O concentrations seen the the Columbia River Flood Basalts.
An argument against this hypothesis is presented by Caprarelli and Reidel (2004). They state that it would be difficult for melts to travel over 65 km through the subcontinental mantle lithosphere and the crust without being altered in any way. They state that the hypothesis, “while geochemically possible, appears to us to be geologically implausible” (Caprarelli and Reidel, 2004).