Over 500,000 km3 of intermediate- to silicic-composition ash flow tuffs and lavas were erupted from large-volume volcanic fields in western North America during the middle to Cenozoic. Of the commonly used isotope systems, Nd isotope data provide the best constraint on the proportions of crust and mantle components in the tuffs; all tuffs that have been studied contain a major, often dominant, mantle component. The proportion of mantle to crustal components is best constrained for ash flow tuffs that were erupted on Precambrian crust. There is no simple correlation between inception of extensional faulting with development of calderas, although all calderas formed in or adjacent to regions that ultimately underwent crustal extension to some degree. Similarly, there are no first-order correlations between the proportion of mantle-derived component in the silicic magmas and the tectonic setting. The common thread to all caldera complexes is that they are generated by large fluxes of mantle-derived basaltic magmas. Detailed isotopic and petrologic studies of several caldera complexes indicate that the silicic magmas were fundamentally derived by fractional crystallization of mantle-derived magmas, accompanied by assimilation of continental crust. Cenozoic caldera-related magnatism in western North America represents a major episode of crustal growth and hybridization; crustal growth rates in the Cenozoic may rival those that occurred in the Proterozoic during initial crust formation. New results of petrologic and isotopic studies of multicyclic caldera complexes in the northern Rio Grande rift region, in addition to several other areas of the western United States, confirm previous models that predict large changes in O, Sr, and Nd isotope compositions of the crust during injection of mantle-derived basaltic magmas.

Continued injection of basaltic magmas results in an increase in the mean density of the crust, which may be reflected in upward movement of the locus of silicic magma evolution in the crust; such changes may be monitored in areas underlain by ancient crust using Pb isotope ratios if the crust is stratified in U/Pb ratios. If accompanied by assimilation, crystallization of mantle-derived magmas that stall near the crust-mantle boundary will return crustal components to the mantle in the form of mafic and ultramafic cumulates, in addition to producing a petrologically complex crust-mantle boundary. This model can explain many geophysical characteristics of the crust and upper mantle in western North America and supports the conclusions of many recent studies of lower crustal xenoliths, which propose recent magmatic underplating in tectonically active regions. ¿1991 American Geophysical Union