Fig. 2: ⁸⁷Sr/⁸⁶Sr values for single clinopyroxene grains measured in this study with LA-ICP-MS spot analyses from three plagioclase grains from the same sample (ALIA-D115-18)¹⁶.

Previous measurements of bulk clinopyroxene⁸ (representing 100s of pooled grains; gray squares with an “x”) from this same sample are shown together with the whole rock value⁷ (where WR stand for whole rock; red square), clinopyroxenes measured in this study (pink squares with an “x”), and plagioclase analyses from Edwards et al.¹⁶ (blue diamonds, circles, and triangles); all error bars smaller than symbols except where shown and represent 2σ standard errors. In the histogram, the number of clinopyroxene observations (n = 17; red bars and red figure axis label) span a larger range of ⁸⁷Sr/⁸⁶Sr than the plagioclase ⁸⁷Sr/⁸⁶Sr analyses¹⁶ (n = 157; blue bars and blue figure axis label; error bars are 2 standard errors); where the clinopyroxene and plagioclase histograms overlap, a purple shade is used. The wider range of clinopyroxene ⁸⁷Sr/⁸⁶Sr values compared with the plagioclase from the same sample is consistent with magma mixing modeling using MELTS^24,25,26 and the Magma Chamber Simulator²⁷. The magma mixing models show that magma mixing between a mafic endmember and a more silicic, trachytic endmember (as derived in this study) precipitates clinopyroxene about 50 °C prior to plagioclase saturation and thus the mixing magmas may be slightly more homogenized at this point, explaining the reduced range of ⁸⁷Sr/⁸⁶Sr exhibited by plagioclase (see Main text and “Methods” for more detail).