Mantle Source Lithologies for the Columbia River Flood Basalt Province

The Miocene Columbia River Basalt Group (CRBG) in the inland northwestern United States is recognized as the world's smallest continental flood basalt province, but it is also one of the best studied, and is set in a complex tectonic environment. It consists dominantly of tholeiites and tholeiitic basaltic andesites erupted between 16.8 and 16 Ma. The province's occurrence in a complex back arc setting straddling ancient and young continental crust, and proximity to the Yellowstone hotspot track, has led to several hypotheses for both a plume and non-plume origin. One observation that does not conform to predictions for a deep-seated mantle plume is the apparent complete absence of highly magnesian lavas consistent with a high mantle melting temperature; most other flood basalt provinces do contain at least some picrites. Here, we focus on the Imnaha, Steens and Picture Gorge formations of the CRBG, reviewing existing geochemical data (all 3 formations) and presenting new major, minor and trace element data from whole rocks, glasses, and olivine phenocrysts from the Imnaha and Picture Gorge. Rare earth element patterns and Fe/Mn in whole rocks and glasses, and the Ni and Ca contents of olivines, are all consistent with a variable component of garnet pyroxenite in the mantle source for the CRBG. The pyroxenite signal is seen most strongly in the Steens Basalt and the Rock Creek type of the Imhaha Basalt; it is weaker in the American Bar type of Imnaha, and weak to absent in the Picture Gorge Basalt. In addition, we show that the American Bar Imnaha type is the likely parent for the more evolved (mostly basaltic andesite) Grande Ronde lavas. These observations are consistent with previously published interpretations of isotopic data for a plume source consisting of both enriched (EM2) and depleted mantle components, but conflict with the observation that the Steens Basalt source shows long-term chemical depletion; hence we speculate that oceanic crust in the descending Farallon slab may have made a direct contribution to Steens lavas. We also note that the recognition of pyroxenite in the CRBG source region largely removes the "no picrites" objection to a mantle plume as the ultimate origin for the CRBG.