The comment by Kelemen and Yogodzinski (2007) criticizes extrapolation of our results for Mt. Shasta high-Mg andesite (S-HMA) to interpretations of other occurrences of HMA. Our intent was to encourage the petrologic community to re-examine all HMAs to better understand their petrogenesis on an individual basis. Where mixing is evident, it is critical to identify the end members involved and their origins. Although we did not specifically link our study to Aleutian HMAs (A-HMA), Kelemen and Yogodzinski suggest the latter could be a common high-Mg# and high-Sr/Y end member with global significance to arc petrology, including Mt. Shasta.
S-HMA can be explained as a mixture of high-Sr/Y dacite and low-Sr/Y basalt, with ultramafic contaminants (Streck et al., 2007). S-HMA differs from A-HMA in having significantly higher MgO, Ni, Cr, etc. (Fig. 1). However, the presence of reversely zoned clinopyroxene with Fe-rich cores (Mg# ≤75) in some of the highest-Sr/Y A-HMAs (Yogodzinski and Kelemen, 1998; Kelemen and Yogodzinski, 2007), indicate that these lavas may also be mixtures of low- and high-Mg# magmas. We do accept that they are unique, but we question their role as being globally significant and that they contribute to formation of magmas at Mt. Shasta (Figs. 1 and 2).
Although Mt. Shasta dacites with narrow ranges in SiO2 (62–64 wt%) and MgO (3–4 wt%) appear to project toward A-HMA (Kelemen and Yogodzinski, 2007, their Fig. 2) (our Fig. 1A), in diagrams such as Sr versus Ba and Sr/Y versus La (Fig. 2), the field for A-HMA lies well off the Shasta dacite trend (as well as opposite the dacite trend from S-HMA) and appears to be a poor end member. It is conceivable that the Mt. Shasta dacite trend (with a large range in Sr/Y) may be an artifact of generating dacites via fractional crystallization and/or of partial melting of variable protoliths at variable pressure. It is also possible that source Sr/Y could vary in response to the addition of variable amounts of high-Sr/Y slab-derived fluid. However, if the dacites are crustal melts, then slab fluids might play little part, but rather the amount of garnet (and/or amphibole) versus plagioclase in the crust would control melt compositions. In any case, A-HMA is unusual in having low MgO but high Mg#; this implies low Fe2+ content—could this be a consequence of oxidation of iron? Shasta dacites similarly have high Mg#—coincidence, or is there a connection?
Finally, our comment regarding Pb isotope data was paraphrased from Kelemen et al. (2003, p. 616) who state: “In all of these localities [referring to other HMA localities], other than the Aleutians, most high Mg# andesites have elevated 208Pb/206Pb, compared to MORB. Thus lead isotope data suggest the presence of a component derived either from recycling of lead from subducted sediment, or from crustal interactions of primitive basalts with older, continental crust and continentally derived sediment.”