PUBLICATIONS Journal of Geophysical Research: Solid Earth RESEARCH ARTICLE Uplift, rupture, and rollback of the Farallon slab reflected 10.1002/2017JB014517 in volcanic perturbations along the Yellowstone Key Points: adakite hot spot track • Volcanic perturbations in the Cascadia back-arc region are derived from uplift Victor E. Camp1 , Martin E. Ross2, Robert A. Duncan3, and David L. Kimbrough1 and dismemberment of the Farallon slab from ~30 to 20 Ma 1Department of Geological Sciences, San Diego State University, San Diego, California, USA, 2Department of Earth and • Slab uplift and concurrent melting 3 above the Yellowstone plume Environmental Sciences, Northeastern University, Boston, Massachusetts, USA, College of Earth, Ocean, and Atmospheric promoted high-K calc-alkaline Sciences, Oregon State University, Corvallis, Oregon, USA volcanism and adakite generation • Creation of a seismic hole beneath eastern Oregon resulted from thermal Abstract Field, geochemical, and geochronological data show that the southern segment of the ancestral erosion and slab rupture, followed by Cascades arc advanced into the Oregon back-arc region from 30 to 20 Ma. We attribute this event to thermal a period of slab rollback uplift of the Farallon slab by the Yellowstone mantle plume, with heat diffusion, decompression, and the release of volatiles promoting high-K calc-alkaline volcanism throughout the back-arc region. The greatest Supporting Information: • Supporting Information S1 degree of heating is expressed at the surface by a broad ENE-trending zone of adakites and related rocks • Data Set S1 generated by melting of oceanic crust from the Farallon slab. A hiatus in eruptive activity began at ca. • Data Set S2 22–20 Ma but ended abruptly at 16.7 Ma with renewed volcanism from slab rupture occurring in two separate • Data Set S3 regions. The eastern rupture resulted in the extrusion of Steens Basalt during the ascent and melting of a dry Correspondence to: mantle (plume) source contaminated with depleted mantle. The contemporaneous western rupture resulted V. E. Camp, in renewed subduction, melting of a wet mantle source, and the rejuvenation of high-K calc-alkaline [email protected] volcanism near the Nevada-California border at 16.7 Ma. Here the initiation of slab rollback is evident in the westward migration of arc volcanism at 7.8 km/Ma. Today, the uplifted slab is largely missing beneath the Citation: Oregon back-arc region, replaced instead by a seismic hole that is bound on the south by the adakite hot spot Camp, V. E., M. E. Ross, R. A. Duncan, and track. We attribute slab destruction to thermal uplift and mechanical dislocation that culminated in rapid D. L. Kimbrough (2017), Uplift, rupture, – and rollback of the Farallon slab tearing of the slab from 17 15 Ma and possible foundering and sinking of slab segments from 16 to 10 Ma. reflected in volcanic perturbations along the Yellowstone adakite hot spot Plain Language Summary Yellowstone National Park is underlain by a rising plume of hot rock track, J. Geophys. Res. Solid Earth, 122, from the Earth’s deep mantle that has provided the melt source for three supereruptions over the last 2 7009–7041, doi:10.1002/2017JB014517. million years. This hot spot, or mantle plume, appears to be a long-lived feature that resided offshore beneath the oceanic Farallon plate before being overridden by the westward moving North American plate about 42 Received 5 JUN 2017 Accepted 15 AUG 2017 million years ago (Myr). Between 42 and 17 Myr the Yellowstone plume was shielded beneath the subducting Accepted article online 17 AUG 2017 Farallon slab, with little surface expression on the overriding North American plate. After 17 Myr the plume Published online 9 SEP 2017 reemerged to produce a great outpouring of basaltic lava known as the Columbia River flood basalts of eastern Oregon and southeastern Washington. New and compiled chemical and age data suggest that the Farallon slab was uplifted and dislocated by the thermally buoyant Yellowstone mantle plume between 30 and 20 Myr, with oceanic crust of the Farallon slab melting to generate unusual rocks called adakite. Eventual destruction of the Farallon slab beneath eastern Oregon was associated with a long-lived period of thermal erosion and tearing of the slab from 30 to 17 Myr, followed by the foundering and sinking of slab segments from 16 to 10 Myr. 1. Introduction Cenozoic volcanism in the Cascade magmatic arc is derived from subduction of the Farallon plate, with the southern limit of volcanism progressing northward, contemporaneous with northward migration of the Mendocino triple junction [Atwater, 1970; McBirney, 1978; Dickinson and Snyder, 1979]. The modern magmatic arc is composed of active volcanoes of the High Cascades and related volcanic rocks that erupted after ca. 4 Ma. The older portion of the arc erupted from ca. 45 to 4 Ma and includes highly dissected volcanic rocks of the Western Cascades in Oregon, Washington, and northernmost California. These older volcanic products constitute the northern segment of the ancestral Cascades magmatic arc (Figure 1) [du Bray and John, 2011]. A lesser known but concurrent southern segment of the ancestral arc was proposed by Noble [1972] and fi ©2017. American Geophysical Union. Christiansen and Yeats [1992] but de ned more precisely in a series of recent studies as a broader belt of All Rights Reserved. calc-alkaline volcanism straddling the Californian-Nevada border region and active from ca. 30 to 4 Ma CAMP ET AL. RUPTURE OF THE FARALLON SLAB 7009 Journal of Geophysical Research: Solid Earth 10.1002/2017JB014517 Figure 1. Location map of geographic and physiographic features in the arc and back-arc regions of southern Oregon, northeastern California, northwestern Nevada, and western Idaho. Distribution of the ancestral Cascades volcanic arc from du Bray and John [2011] and du Bray et al. [2014]. Volcanic rocks from the northern segment of the ancestral Cascades arc range from >45 Ma to 4 Ma, but in Oregon from ca. 35 to 4 Ma [du Bray and John, 2011]. Rocks of the southern segment of the ancestral Cascades arc range in age from ca. 30 to 4 Ma [du Bray et al., 2014]. The proposed tear in the Farallon/Juan de Fuca slab between these two arc segments was suggested by Cousens et al. [2008] and Colgan et al. [2011] to account for the apparent left-lateral offset of their volcanic axes. (Figure 1) [Putirka and Busby, 2007 Busby et al., 2008a, 2008b; Cousens et al., 2008, 2011; Hagan et al., 2009; Busby and Putirka, 2009; Putirka et al., 2012; John et al., 2012; du Bray et al., 2014]. Volcanism of the ancestral Cascades appears to have occurred above a Farallon slab shown to be intensely dislocated in recent seismic investigations [e.g., Burdick et al., 2008; Schmandt and Humphreys, 2010; Sigloch, 2011; Tian et al., 2011; Liu and Stegman, 2011; Tian and Zhao, 2012]. Other studies attribute this dis- memberment to interaction of the Farallon slab with the Yellowstone hot spot, where rupturing of the slab resulted in plume rise and the Miocene eruption of the Columbia River flood basalts [Xue and Allen, 2007, 2010; Obrebski et al., 2010; Humphreys and Schmandt, 2011; Coble and Mahood, 2012; Darold and Humphreys, 2013; Wells et al., 2014; Camp et al., 2015]. Although disruption of the Farallon plate, and/or changes in slab dip, should also result in distinct changes in the style and distribution of arc/back-arc mag- matism, such perturbations have not been recognized or thoroughly explored. Here we use field data and a large database of geochemical and geochronological data to define time- dependent variations in the distribution, composition, and eruption style of Oligocene-to-Miocene volcanism in eastern Oregon and adjacent Nevada, California, and Idaho. Our intent is to examine whether or not these variations in time and space are consistent with a dynamic model of plume-slab interaction as conceived in tectonic models [e.g., Coble and Mahood, 2012] and seismic investigations [e.g., Xue and Allen, 2010; Obrebski et al., 2010]. CAMP ET AL. RUPTURE OF THE FARALLON SLAB 7010 Journal of Geophysical Research: Solid Earth 10.1002/2017JB014517 2. Analytical Data and Methods We present new geochronological data on 20 samples and new geochemical data on 299 samples from vol- canic rocks in the Oregon-Nevada-California border region. Major and trace element data were derived from X-ray fluorescence (XRF) at the San Diego State University Geoanalytical Laboratory, with a subordinate num- ber of XRF analyses from the P. R. Hooper Geoanalytical Laboratory at Washington State University. This data set is available at the EarthChem repository [Camp and Ross, 2017]. We use and compare these analyses with a large geochemical database of 2637 analyses compiled from specific site investigations [Norman and Leeman, 1990; Langer, 1991; Mathis, 1993; Camp et al., 2003; Carmichael et al., 2006; Colgan et al., 2006, 2011; Scarberry et al., 2010], together with regional compilations from the ancestral southern Cascades, the ancestral northern Cascades, the Columbia River Flood Basalt Province, and the northern Nevada rift system [du Bray and John, 2011; du Bray et al., 2006, 2014; Hooper, 2000; John et al., 2000; Camp et al., 2013] (North American Volcanic and Intrusive Database (NAVDAT), http://www.navdat.org). Table 1 summarizes geochronological data on 12 mafic lavas and 3 rhyolitic rocks determined by the 40Ar/39Ar incremental heating method at the Noble Gas Mass Spectrometry Laboratory at Oregon State University. The summary age data for five additional rhyolite samples determined by U-Pb zircon ablation geochronology at the University of Arizona Laserchron Center are presented in Table 2. The analytical detail for each data set is available at the EarthChem repository [Duncan et al., 2017; Kimbrough et al., 2017].