The leading wisps of Yellowstone: Post–ca. 5 Ma extension-related magmatism in the upper Wind River Basin, Wyoming (USA), associated with the Yellowstone hotspot tectonic parabola

The identification of volcanism associated with intracontinental hotspots, as well as the hotspot tracks themselves, can be challenging (Courtillot et al., 2003). Hotspot identification is often based on magmatic age progressions, which can be controversial due to errors in geochronology constraints and the issue of linking volcanic events to a specific driving mechanism (Fouch, 2012; Kuehn et al., 2015). Complicating matters, magmatism often continues long after (e.g., millions of years) the upper plate has been translated away from an upwelling plume (Bercovici and Mahoney, 1994; Sleep, 2003; Shervais and Hanan, 2008; Jean et al., 2014). Furthermore, out-of-sequence magmatism assigned to other driving mechanisms—such as “hotlines” due to convection cells in the upper mantle (Meyers et al., 1998), magma overpressurization–induced lithosphere deformation (Hieronymus and Bercovici, 2001), or lateral melt propagation off the hotspot track (Konstantinou et al., 2013)—might overlap with the timing of hotspot magmatism (Kuehn et al., 2015). We propose a new solution to assist in the identification of intracontinental hotspot-related volcanism. This specifically addresses the discrepancy between the ages and occurrence of older “off-track” volcanic products in northwest Wyoming (USA) that are adjacent to a proposed hotspot track yet are out of sequence with the hotspot track age progression.

Nearly all Cenozoic volcanism in Wyoming is associated with the ca. 2.1 Ma to present Yellowstone Plateau volcanic field (Fig. 1) (Christiansen, 2001). However, ~80–100 km southeast of the Yellowstone Plateau volcanic field, coincident with the southern margin of the Absaroka Range in the upper Wind River Basin, Wyoming, are thick packages of Eocene volcanic and volcaniclastic strata related to Farallon plate subduction (Absaroka volcanics; Feeley, 2003). Additionally, <3 Ma alkalic volcanic rocks are found in southwestern Wyoming in the Leucite Hills (Fig. 1; Carmichael, 1967; Lange et al., 2000; Mirnejad and Bell, 2006). Furthermore, in the greater Yellowstone region (northwestern Wyoming, northeastern Idaho, and southwestern Montana), scattered remnants of young (<5 Ma) near-vent deposits, lavas, and intrusives stratigraphically overlie the Eocene strata, generally at high elevations (Fig. 1). Some of these <5 Ma upper Wind River basin rocks were originally mapped by Love (1947), Keefer (1957), and Love et al. (1979) but have remained essentially unstudied by modern techniques until now, and are the focus of this study. Moreover, they appear to be part of a region in northwest Wyoming and northeast Idaho where primarily intermediate-composition rocks with arc-like trace element signatures were either erupted or emplaced during the last ~9 m.y. (Price, 2009; Adams, 2014). This includes the ca. 8.5–8 Ma Jackson Hole volcanics, dated by the K-Ar method (Fig. 1; Naeser et al., 1980; Adams, 1997; Lageson et al., 1999), and the ca. 7–6 Ma Carlton Creek Volcanics, dated by the ⁴⁰Ar/³⁹Ar method (Price, 2009). Price (2009) associated the Carlton Creek Volcanics with Basin and Range–related volcanism and noted a compositional similarity to the Jackson Hole volcanics. The proximity of the upper Wind River Basin to the position of the Snake River Plain–Yellowstone hotspot track warrants exploration of any potential relationship between the two.

Prior K-Ar geochronology from three upper Wind River Basin locations yielded ages of ca. 0.48 Ma (Lava Mountain; Rohrer and Obradovich, 1978), ca. 3.4 Ma (Pilot Knob; Rohrer and Obradovich, 1978) and ca. 3.6 Ma (Crescent Mountain; Blackstone 1966) (Figs. 1 and 2). Thus, there is some temporal overlap with the Yellowstone Plateau volcanic field, however the two older ages fall in a gap between Yellowstone volcanism and volcanism related to the Heise volcanic field, which ended at ca. 4 Ma (Morgan and McIntosh, 2005; Ellis et al., 2017).

This study presents field constraints, ⁴⁰Ar/³⁹Ar geochronology, and whole rock geochemistry from three largely unstudied upper Wind River Basin volcanoes (Lava Mountain, Spring Mountain, and Crescent Mountain) and two shallow intrusive bodies (Pilot Knob and Wildcat Hill) (Figs. 1, 2). Aside from the three prior K-Ar ages and two bulk rock geochemical analyses from Crescent Mountain (Blackstone, 1966), the data in this study are the first of their kind from these igneous bodies. The primary purpose of the study is to document the geology of these volcanoes and intrusives. We recognize that the upper Wind River Basin volcanoes and intrusives form a cluster of volcanoes that fits the broad description of a volcanic field as described by de Silva and Lindsay (2015), which we define here as the Upper Wind River Basin volcanic field (UWRB; Fig. 1). This volcanic field, along with the Jackson Hole volcanics and other locally sourced, small-volume eruptive loci in eastern Idaho, are all part of the same group of igneous activity that is not well understood in terms of broader regional tectonic influences such as Basin and Range extension and the Yellowstone hotspot.

The UWRB magmatism was influenced by the late Archean Wyoming craton, which stabilized at ca. 2.8 Ga and underlies much of present-day Wyoming, southeastern Montana, and parts of eastern Idaho, western South Dakota, and northern Utah (Mueller and Wooden, 1988; Foster et al., 2006). Craton rocks are now exposed at the surface due to Laramide thick-skinned reverse faulting in many of the regional mountain ranges (e.g., Wind River uplift leading to the Wind River Range and Washakie uplift and Washakie Range which includes Spring Mountain). Wyoming was a stable shelf environment from the Paleozoic to Early Cretaceous, with dominantly marine sedimentation in a shallow epeiric sea environment (Ray and Keefer, 1985). By the Late Cretaceous, marine and non-marine sediments were deposited in the Wind River Basin coinciding with the onset of the Late Cretaceous–early Eocene Laramide orogeny. During the Laramide, mountain ranges developed along high-angle reverse faults and the Wind River Basin began to subside through the early Eocene, resulting in a large sediment influx. Local Laramide structures are evident today in the form of large down-warped anticlines and dome structures in the basin (Ray and Keefer, 1985). By the end of the Eocene, most uplift and subsidence had stopped; however, accumulation of sediment in the Wind River Basin continued through the mid-Miocene (Ray and Keefer, 1985). Precambrian igneous and metamorphic rocks outline the northern, southern, and western flanks of the Wind River Basin (e.g., Wind River uplift to the south and Washakie uplift to the north). These include >10,000 km³ of late Archean granite, schist, and gneiss cut by mafic dikes (Keefer, 1970). From the Late Cretaceous through early Tertiary, sediments removed during uplift of the Wind River Range were deposited into the central parts of the Wind River Basin until uplift halted in the late Eocene, at ca. 51 Ma (Fan et al., 2011). Pre-Tertiary carbonate and siliciclastic sediments deposited in the basin total ~5 km in thickness (Seeland and Brauch, 1975). Volcaniclastic deposits, including volcanic conglomerate and tuffaceous sandstone, compose much of the Eocene sedimentary strata found in the upper Wind River Basin (Keefer, 1970). In the upper Wind River Basin, these mid-Eocene rocks are dominated by thick (>1 km) packages of lavas, pyroclastics, and volcaniclastic deposits of the Absaroka volcanic field (Keefer, 1970). Local Absaroka volcanism extends from north of Dubois, Wyoming, northwest into Yellowstone and into southwest Montana. The mantle source involved in Absaroka volcanism is thought to be ancient depleted lithospheric mantle that has since been enriched by both older and recent metasomatic events. The latter was most likely associated with the eastward subduction of the Farallon plate under present-day North America (Feeley, 2003).

During the past ~2 m.y., >6000 km³ of primarily compositionally bimodal felsic and mafic magma has erupted within the Yellowstone Plateau volcanic field (Christiansen, 2001). The most commonly accepted explanation for the origin of the Yellowstone Plateau volcanic field is a stationary mantle plume with shallow manifestations (e.g., seismically imaged magma body) recognized underneath present-day Yellowstone National Park. The origin of the hotspot is still debated, however the volcanism associated with it first occurred along the Idaho-Oregon-Nevada border region, forming the Columbia River–Steens flood basalts between ca. 16.8 and 15.0 Ma (Carlson and Hart, 1987; Camp and Ross, 2004). As the North American plate migrated to the southwest, Yellowstone hotspot-related volcanism migrated to the northeast, eventually forming the eastern Snake River Plain and ultimately the Yellowstone Plateau (Fig. 1). Recent Yellowstone volcanism includes three major explosive silicic eruptions (cycles one to three) that together helped to form the present-day Yellowstone Plateau volcanic field. The youngest caldera of the Yellowstone Plateau volcanic field was created during the eruption of the 0.64 Ma Lava Creek Tuff (i.e., third cycle). Recent basaltic volcanism that postdates this includes basalts of the Snake River Group, the Osprey Basalt, and the Madison River Basalt (Christiansen et al., 2007). Volumetrically, most Yellowstone Plateau volcanic field extrusive rocks are rhyolite, making up 95% of the entire volcanic field (Christiansen, 1984). The few areas of exposed mafic rocks found within Yellowstone are primarily from the first and third caldera-forming eruptive cycles; basalts followed alternating periods of felsic eruptions in Yellowstone National Park during periods of extension (Christiansen, 2001). Basaltic eruptions after the third cycle of caldera-forming silicic activity are primarily concentrated around areas of previous basaltic volcanism and extensional faulting and are found outside the third-cycle caldera (Christiansen, 2001).

Basin and Range extensional tectonics may have played an important role in the development of UWRB magmatism. In northwest Wyoming, Basin and Range extension is best manifested by large-displacement normal faults and associated half-grabens (e.g., Teton normal fault and Jackson Hole, Wyoming), although numerous other smaller fault systems are present (Machette et al., 2001). Cenozoic normal faults have been mapped in the UWRB, cut Precambrian basement in the Wind River Range, and have also been mapped in the southeast portion of Yellowstone National Park (Figs. 1, 2; Love et al., 1979; Christiansen, 2001). The relationship between Basin and Range extension and the Yellowstone hotspot is debated, but they are likely linked to some degree (Christiansen et al., 2002; Colgan and Henry, 2009; Pierce and Morgan, 2009; Camp et al., 2015). However, regardless of that postulated link, in the greater Yellowstone region, collapse of the Sevier fold-and-thrust belt and Laramide uplifts via normal faulting initiated soon after the end of the Laramide orogeny, well before the ca. 17 Ma inception of the Yellowstone hotspot, and continued into the Miocene (Winterfeld and Conrad, 1983; Steidtmann and Middleton, 1991; Constenius, 1996). Ample evidence exists for regional extensional basin formation by the early to middle Miocene. For example, extension and basin sedimentation occurred just north of Yellowstone National Park in Montana at ca. 18–13 Ma (Barnosky and Labar, 1989; Burbank and Barnosky, 1990), and south in Jackson Hole, Wyoming, and Grand Valley, Idaho, at ca. 16–10 Ma (Barnosky and Labar, 1989; Anders et al., 2009); the latter is associated with the Miocene Jackson Hole volcanics. This Miocene phase of extension has also been linked to Yellowstone plume head–North American plate interaction (Anders et al., 2009), which complicates interpretations as to why it occurred. Subsequent extension in Jackson Hole that continues today, primarily along the Teton normal fault, was initiated at ca. 3 Ma after an ~7 m.y. hiatus (Anders et al., 1989). Mid-Miocene to ongoing extensional processes also affected southwestern Montana and northeastern Idaho, some of which have been linked to the Yellowstone hotspot (Fritz and Sears, 1993; Sears et al., 2009). This extensional tectonic regime continues today along the western side of Wyoming and extends southward past Salt Lake City, Utah, as part of the Intermountain seismic belt and eastern edge of the Basin and Range province (Payne et al., 2012).

The UWRB volcanics and intrusive bodies were initially mapped by Keefer (1957), and subsequent studies by Blackstone (1966), Love et al. (1979), and Protska et al. (1979) provided more detailed field and geochemical constraints on a few of the rocks, though these efforts simply documented their occurrence. One observation that unifies these units is that they all stratigraphically lie above Eocene Absaroka rocks, and in most cases, UWRB lavas lie disconformably on a low-relief erosional surface that developed on the Eocene rocks (Pierce and Morgan, 1992). Lava Mountain was considered Quaternary basalt on the geological map of Yellowstone National Park (Christiansen, 2001) and is a partially dissected shield volcano composed of a pile of at least 26 lava flows capped by a scoria cone (Love and Love, 1983; this study). This study demonstrates that no basalt is present at Lava Mountain, only basaltic andesite to dacite. The prior K-Ar age (0.48 ± 0.06 Ma; Rohrer and Obradovich, 1978) was obtained from a flow near the top of the lava stratigraphy and, if accurate, places Lava Mountain volcanism temporally coincident with the mafic eruptions that followed eruption of the Lava Creek Tuff, forming the third-cycle Yellowstone caldera. K-Ar dating and petrography from Pilot Knob were reported by Rohrer and Obradovich (1978) and Love and Love (1983), who identified the knob as a biotite-augite intrusive (vogesite). Blackstone (1966) mapped Crescent Mountain and described it as an ~150-m-high cinder cone, partially covered by a 60-m-thick package of “basalt” lavas. Two geochemical analyses from Crescent Mountain indicate that the local magmatic products are basaltic andesite to dacite (Blackstone, 1966). Protska et al. (1979) also briefly discussed Crescent Mountain and stated that the lavas are “basaltic” based on geological mapping.

Sampling locations were targeted based on published geological maps and satellite imagery. At Lava Mountain, stratigraphic relationships between lavas guided sampling, and the goal was to derive a detailed stratigraphy of the volcanic pile, from its base to the capping scoria cone (Downey, 2015). At Spring Mountain and Pilot Knob, geological mapping accompanied sampling; mapping did not occur at Crescent Mountain or Wildcat Hill. Keefer (1955) mapped highly weathered basaltic lava to the south of Spring Mountain extending down into the upper Wind River Basin north of Dubois; these exposures were unsampled in our study. Only the freshest rocks were targeted for collection and further processing and analysis. Weathered surfaces were removed from samples larger than fist size, which were then split into smaller pieces using a RockLabs hydraulic press. These pieces were powdered using a Spex Industries shatterbox (alumina vessel).

For ⁴⁰Ar/³⁹Ar analysis, two samples were submitted to the Geochronology Laboratory at University of Alaska Fairbanks (UAF; Alaska, USA) where they were crushed, sieved, washed, and hand-picked for phenocryst-free whole-rock groundmass chips. The monitor mineral TCR-2 (sanidine from the Taylor Creek Rhyolite; Duffield and Dalrymple, 1990) with an age of 28.619 Ma (Renne et al., 2010) was used to monitor neutron flux and calculate the irradiation parameter (J) for all samples. The samples and standards were wrapped in aluminum foil and loaded into aluminum cans of 2.5 cm diameter and 6 cm height. The samples were irradiated in position 8c of the enriched-uranium research reactor of McMaster University (Ontario, Canada) for 20 MWh. Upon their return from the reactor, the samples and monitors were loaded into 2-mm-diameter holes in a copper tray that was then loaded in an ultrahigh-vacuum extraction line. The monitors were fused, and samples heated, using a 6 W argon-ion laser following the technique described in York et al. (1981), Layer et al. (1987), and Benowitz et al. (2014). Argon purification was achieved using a liquid nitrogen cold trap and a SAES Zr-Al getter at 400 °C. The samples were analyzed in a VG-3600 mass spectrometer at the Geophysical Institute, UAF. The argon isotopes measured were corrected for system blank and mass discrimination, as well as calcium, potassium, and chlorine interference reactions following procedures outlined in McDougall and Harrison (1999). Typical full-system eight-minute laser blank values were generally 2 × 10^–16 mol ⁴⁰Ar, 3 × 10^–18 mol ³⁹Ar, 9 × 10^–18 mol ³⁸Ar, and 2 × 10^–18 mol ³⁶Ar, which are 10–50 times smaller than the similar sample and standard molar gas releases. Correction factors for nucleogenic interferences during irradiation were determined from irradiated CaF₂ and K₂SO₄ as follows: (³⁹Ar/³⁷Ar)Ca (³⁹Ar/³⁷Ar produced from Ca in the nuclear reactor) = 7.06 × 10^–4, (³⁶Ar/³⁷Ar)Ca = 2.79 × 10^–4 and (⁴⁰Ar/³⁹Ar)K = 0.0297. Mass discrimination was monitored by running calibrated air shots. The mass discrimination during these experiments was 0.8% per mass unit. While doing our experiments, calibration measurements were made on a weekly and monthly basis to check for changes in mass discrimination, with no significant variation observed during these intervals. A summary of both of the ⁴⁰Ar/³⁹Ar results is given in Table 1, with all ages quoted to the ±1σ level and calculated using the constants of Renne et al. (2010). We did not take into account the error on the decay constant nor the error on the age of the flux monitor in our calculations, which withstanding the effects on overall error would be small given their limited contribution to overall uncertainty. Detailed step heat data are provided in Supplemental Table S1¹. The integrated age is the age given by the total gas measured and is equivalent to a potassium-argon (K-Ar) age. The spectrum provides a plateau age if three or more consecutive gas fractions represent at least 50% of the total gas release and are within two standard deviations of each other (mean square weighted deviation <2.5).

Inductively coupled plasma–optical emission spectrometry (Agilent 720ES ICP-OES) analyses of seventeen samples (AD14 and ZD14 series) were completed at Miami University (Ohio, USA) following the methods outlined by Katoh et al. (2016). This included major (Si, Al, Ti, Fe, Mn, Mg, Ca, Na, K, and P) and trace (Ni, Cr, Sc, V, Ba, Rb, Sr, Zr, Nb, Cu, Co, Zn, Pb, Y) element analyses. Table S2 [footnote 1] provides the results for standard reference materials BHVO-2 and AGV-2 run with the AD14 and ZD14 series samples and for multiple analyses of sample AD14-16. Five MB13 series samples were analyzed for major and trace element chemistry via a Panalytical PW 4204 X-ray fluorescence (XRF) spectrometer at Franklin and Marshall College (Pennsylvania, USA), following the methods outlined in Mertzman (2000, 2015) and online at http://www.fandm.edu/earth-environment/laboratory-facilities/.

A full rare earth element (REE) suite—La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu—was determined for selected AD14, ZD14, and MB13 series samples at Miami University by inductively coupled plasma–mass spectrometry (ICP-MS). Initial solutions for unknowns and reference materials BHVO-2 and AGV-2 (Table S2 [footnote 1]) were generated via flux fusion involving the following steps: (1) 200 mg of dried powder was combined with 600 mg lithium metaborate (LiBO₂) in a pre-fired graphite crucible and fused at 950 °C for 20 min, and (2) after cooling, the resulting glass beads were added to 80 mL of 5% high-purity HNO₃ and agitated on a gyratory shaker until total dissolution was achieved. A zero standard (processing blank) was prepared following the same steps using only 200 mg of LiBO₂. The REEs were quantitatively isolated from these solutions through cation exchange chromatography using Dowex 50W-X12 resin, and the REE fractions were eluted with 40 mL of 5N HNO₃. After drying on a hotplate, the REE fractions were brought back into solution in 20 mL of 0.3N HNO₃. These final solutions were analyzed in a Varian 820 ICP-MS equipped with an autosampler. The blank and a set of four synthetic REE standard solutions were used for calibration and were analyzed at the beginning of the instrument session. Background adjusted counts were corrected for within-run drift through the use of mass-appropriate internal standards (¹¹⁵In and ¹⁸⁵Re).

We collected LMT and CRM series samples, and these were processed and analyzed for major and trace elements and REEs via XRF and ICP-MS at Washington State University (Washington, USA) using a ThermoARL Advant’XP+ sequential XRF spectrometer and an Agilent 7700 ICP-MS (REEs) following the methods of Johnson et al. (1999) and Gaschnig et al. (2011).

Twenty-five Lava Mountain, five Spring Mountain, six Pilot Knob, two Wildcat Hill, and six Crescent Mountain samples were analyzed for bulk rock geochemistry. Total volatile loss on ignition was determined for AD, ZD, and MB series samples and is reported with representative major and trace element results in Table 2 (and also in Table S3 [footnote 1]). Iron was split into Fe⁺² and Fe⁺³ according to the procedure described by Le Maitre (1976), and all major element data used in diagrams and the discussion are reported as anhydrous using the split iron data.

The geochemistry of UWRB magmas will be discussed in detail below; however, to establish a common nomenclature and describe their petrography, initial classification has been made via a total alkalies–versus–silica diagram (Le Bas et al., 1986) (Fig. 3). Lava Mountain rocks plot as basaltic andesites, andesites, and dacites, and Spring Mountain rocks are basalts. Crescent Mountain samples are basaltic andesites and dacites, while Pilot Knob samples plot as basaltic trachyandesites and Wildcat Hill rocks are andesite to dacite.

Based on its shape and areal extent, Lava Mountain is a shield volcano and the volume of erupted material is ~5.5 km³ (Downey, 2015). Lava Mountain shows evidence of being affected by post-volcanic glaciation, including truncated cliffs that expose the lavas (Figs. 4A, 4B). Field observations and geochemistry indicate that there are stratigraphic variations in magma composition; i.e., the basal ~100 m of the volcano was dominated by broadly mafic and felsic eruptions (basaltic andesite and dacite; Fig. 3), while the upper 400 m, including the capping cone, is andesitic. The youngest and stratigraphically highest unit on Lava Mountain is a scoria cone composed of slightly oxidized andesite scoria and bombs, located on the eastern side of the mountain (Figs. 4A, 4C). Andesite lavas that crop out stratigraphically lower than the scoria cone exhibit both platy and blocky characteristics in cross-section (Figs. 4B, 4D). Most andesite lavas are platy and exhibit sheet jointing in those units immediately above a basal breccia consisting of scoriaceous clasts. Individual platy lavas range from 10 to 50 m thick. Most are not horizontal throughout their lateral extent, but tend to ramp and curve due to flow (Fig. 4D). Elongated vesicles ~3 cm and smaller in maximum dimension are common in these lavas. Outcrops of blocky basaltic andesite lavas range up to 15 m high and are commonly autobrecciated (Fig. 4E). Dacite lavas occur near the base of Lava Mountain and are characterized by upper and lower vitrophyres.

The basaltic andesites from Lava Mountain are aphanitic with sparse phenocrysts of 0.25–2 mm orthopyroxene and are commonly vesicular. Two groups of basaltic andesites were observed. One consists of light tan to gray rocks that show light banding and contain ~5% orthopyroxene phenocrysts. Rocks of the other group are blocky and contain more phenocrysts but fewer vesicles. The orthopyroxenes are subhedral to anhedral, and many are resorbed. The major distinguishing petrographic feature separating the basaltic andesites from other Lava Mountain volcanics is the presence of olivine phenocrysts in all samples (Fig. 5A). Although not very abundant (~3%–7% modally), the olivines are euhedral to anhedral and average 0.3–0.6 mm in size; some exhibit skeletal morphologies. Groundmass ranges from 75% to 90% of the rocks and consists of plagioclase, orthopyroxene, and Fe-Ti oxides. A single 4.6 mm quartz xenocryst is also present in one sample. The andesites generally have an aphanitic groundmass (~65%–98%) with pyroxene phenocrysts (up to 3 mm). All contain vesicles (elongated up to 3–5 cm) and have ~6%–24% orthopyroxene phenocrysts modally; none contain olivine. Orthopyroxene is typically euhedral to subhedral and includes slightly resorbed and zoned crystals. Rounded and complexly zoned plagioclase (2–6 mm) is also present, as well as orthopyroxene with complex reaction rims. Major xenocrysts include an anhedral and partially resorbed alkali feldspar and plagioclase crystals with reaction rims. The most silica-rich samples collected at Lava Mountain are the dacites, which are dominated by groundmass; collected samples are vitrophyric (~96%–97% groundmass) and are black and glassy in hand sample. Trachytic texture is apparent in the groundmass, and most of these plagioclase crystals have a swallowtail texture. Some 0.4–0.6 mm plagioclase crystals are also rounded and resorbed (Fig. 5B) (<2% modally), and one 1.8 mm clinopyroxene was also observed, as well as trace apatite microlites.

Pilot Knob is ~2 km west of Lava Mountain (Figs. 2, 4F) and was emplaced through Eocene sedimentary strata (Dodd, 2015). The intrusion is rounded, but tapers to a southeast-trending ridge-like “fin” at the summit, which we interpret to possibly reflect the base of a dike (Fig. 4F). Glacier erosion–formed boulder fields made of the Pilot Knob intrusive rocks surround in situ bedrock (Figs. 4F, 4G). All Pilot Knob samples we collected are coarse to medium grained and average ~67.1% plagioclase, 14.5% clinopyroxene, 6.5% phlogopite, 6% orthopyroxene, and 6% apatite (Fig. 5C). Although plagioclase is the dominant mineral phase, it occurs primarily in the groundmass, where it is fine grained (<0.25 mm), is euhedral to subhedral, and exhibits strong Carlsbad twinning. Some plagioclase phenocrysts were also noted. Apatite crystals are commonly poikilitically enclosed within the plagioclase. Clinopyroxene crystals range from 0.5 to 1.5 mm in size; they typically exhibit sieve and resorption textures and were noted in both phenocryst and groundmass occurrences. The poikilitic orthopyroxenes present do not appear as weathered as the clinopyroxene and are intergrown with plagioclase crystals as glomerocrysts. Phlogopites are 2–25 mm in size throughout all Pilot Knob samples and commonly appear altered along the outer rims (Fig. 5C). Trace amounts of zircon were also noted in the groundmass. Pilot Knob rocks identified in this study are classified as lamprophyre (Rock, 1977; Woolley et al., 1996). Many lamprophyres plot within the field of trachybasalt on a total alkalis–versus–silica diagram (Le Bas and Streckeisen, 1991), but this geochemical classification overlooks some of the distinctive petrographic features of lamprophyres. For this reason, lamprophyres are commonly classified based on their mineralogy rather than their geochemistry. We classify Pilot Knob as a lamprophyre based on modal mineralogy because it fits the lamprophyre characteristics outlined by Le Maitre et al. (2002). This classification is consistent with its bulk chemical composition. The hydrous mineral phase in Pilot Knob is exclusively mica, and its mineralogical composition is primarily phlogopite (Dodd, 2015). Therefore, the Pilot Knob intrusion is a kersantite lamprophyre (Le Maitre et al., 2002).

Wildcat Hill is exposed ~5 km northwest of Pilot Knob and occurs along an apparent fault zone on strike with a major regional normal fault zone that is mapped ~10 km northwest of Pilot Knob (Fig. 1; Christiansen, 2001). Wildcat Hill also intrudes Eocene rocks (e.g., Absaroka volcanics and volcaniclastic strata) (Fig. 4H). Well-developed columnar jointing and two different petrographic types characterize the intrusion: a glassy outer margin of the intrusion (sample ZD14-19b) and a vesiculated (in places), blocky interior (ZD14-19a). The Wildcat Hill intrusion is hypabyssal in origin, with more fine-grained textures than Pilot Knob. Both samples from Wildcat Hill (ZD14-19a and ZD14-19b) appear homogenous in composition, with ~72.4% groundmass, 8.8% clinopyroxene, 8.5% plagioclase, 7.5% biotite, and 3.2% orthopyroxene (Fig. 5D). A plagioclase-dominated matrix within Wildcat Hill consists of primarily <0.1 mm plagioclase crystals with some plagioclase phenocrysts up to 0.3 mm in length. Pyroxene crystals are ~0.5 mm.

At Spring Mountain, ~38 km east of Lava Mountain, basalts erupted along high-angle normal faults that cut Paleozoic and Eocene strata (1:24,000-scale field mapping by C. Haley, W. Hart, and M. Brueseke (1999–2000); Figs. 1, 2, 4I). These faults are likely part of a more extensive southeast-trending fault system mapped by Keefer (1955) that extends southeast of Spring Mountain (Spring Mountain fault; Fig. 2). Keefer (1955, 1957) showed that erupted lavas flowed south into the upper Wind River Basin. Based on our observations, Spring Mountain volcanism is likely the result of one monogenetic event. Eruptions were focused along faults and were fissure eruptions; large amounts of pervasively oxidized scoria mark the vents (Figs. 4I, 4J, 4K). Textural variation exists among the Spring Mountain basalt and three different textural varieties are present: fine-grained, coarse-grained, and scoria (Fig. 4J). Coarse-grained samples contain ~0.25–0.5 mm olivine phenocrysts and are vesicular. The fine-grained variety (sample MB13-4B) has an aphanitic, non-vesiculated groundmass, but does have fresh iridescent olivine phenocrysts (average ~0.5 mm). In thin section, Spring Mountain basalts are primarily holocrystalline, with subhedral to euhedral olivine phenocrysts and glomeroporphyritic clusters in a fine-grained groundmass (Fig. 5E). Olivine occurs as a phenocryst and groundmass phase in all samples. The phenocrysts can be euhedral, resorbed, and/or skeletal, occasionally with iddingsite rims. Groundmass plagioclase is ubiquitous and has a trachytic texture. Ophitic and subophitic textures are also common in all samples, with plagioclase intergrown with clinopyroxene and olivine. Clinopyroxenes in these rocks are smaller than olivine crystals, averaging only ~0.1 mm in size.

The Crescent Mountain volcano was not mapped in detail as part of this study, but was sampled. Blackstone (1966) mapped a partially dissected scoria cone, as well as a local intrusive dike and package of lavas, all stratigraphically overlying and/or cutting Eocene deposits (Fig. 4L). The cone is estimated to be ~5 km² in extent, and the lava package is interpreted to have erupted after cone formation; both were intruded by the dike (Blackstone, 1966). It appears to be a monogenetic volcano. New sampling and geochemistry indicate that the two geochemical analyses published by Blackstone (1966) were switched; his “andesitic basalt” unit Tpb (samples CRM1, CRM2, CRM3, CRM5) is in fact dacite lavas, while trachyandesite unit Tpi (samples CRM4, CRM4a) is a basaltic andesite dike. Crescent Mountain dacite lavas have abundant sheet jointing and are aphanitic, with ~20%–25% 1–2 mm phenocrysts of plagioclase, clinopyroxene, and rare olivine (Blackstone, 1966; Fig. 5F). Blackstone’s unit Tpi, which crosscuts the cone and lava package, is more crystal rich, with ~30%–35% phenocrysts of 1–2 mm pyroxene and olivine, and <1 mm plagioclase, in a fine-grained trachytic to pilotaxitic matrix (Blackstone, 1966). Blackstone (1966) also reported mica and alkali feldspar in this unit, however we could not substantiate this occurrence.

As discussed above, two UWRB samples—AD14-17 (basal basaltic andesite lava at Lava Mountain) and MB13-3 (Spring Mountain basalt)—were successfully dated to augment existing K-Ar ages from UWRB rocks (Fig. 6; Table 1; Table S1 [footnote 1]). We made multiple attempts to date the capping andesite scoria (AD14-4) at Lava Mountain, but limited radiogenic ⁴⁰Ar (integrated ~0.04%) precluded an interpretable age, indicating a possible young age (<ca. 400 ka) for this eruptive event (Table S1 [footnote 1]). A homogeneous, phenocryst-free whole rock separate from sample AD14-17 (integrated ~2.1% radiogenic ⁴⁰Ar) was analyzed. The integrated age (549 ± 145 ka) and the plateau age (397 ± 48 ka) and isochron age (426 ± 88 ka) are all within error of each other. We prefer the plateau age of 397 ± 48 ka for sample AD14-17 because of the higher precision over the isochron age determination. A homogeneous, phenocryst-free whole rock separate from sample MB13-3 (integrated ~12.0% radiogenic ⁴⁰Ar) was analyzed. The isochron regression to ⁴⁰Ar/³⁶Ar_i (_i is sample initial ⁴⁰Ar/³⁶Ar) indicates no evidence of excess ⁴⁰Ar. The integrated age (4791 ± 150 ka) and the plateau age (4791 ± 150 ka) and isochron age (4841 ± 337 ka) are all within error of each other. We prefer the plateau age of 4791 ± 150 ka for sample MB13-3 because of the higher precision over the isochron age determination.

We report new major and trace element bulk chemistry for 44 samples. Major elements are reported as weight percent (wt%) oxide, and trace elements as parts per million (ppm). Table 2 illustrates representative samples, with the full data set in Table S3 [footnote 1]). As discussed earlier, UWRB samples plot either as basalts, basaltic andesites, basaltic trachyandesites, andesites, or dacites (Fig. 3). They also exhibit two distinct suites: a subalkaline suite defined by Spring Mountain basalts, Lava Mountain, Wildcat Hill, and dacite lavas of Crescent Mountain, and a transitional to alkaline suite, defined by the Crescent Mountain basaltic andesite dike and Pilot Knob (Fig. 3). On an alkali-FeO*-MgO (AFM, where FeO* is total Fe expressed as FeO) diagram, only Spring Mountain basalts are tholeiitic (olivine tholeiites based on normative mineralogy); all of the rest are clearly calc-alkaline (Fig. 7). Major and trace element variations of rocks from each volcano or intrusive are plotted in Figures 8, 9, and 10. While broadly similar, it is apparent that samples from each location are chemically distinct from each other. Lava Mountain samples range from 55.5 to 62.8 wt% SiO₂, and MgO, CaO, FeO*, CaO, and P₂O₅ all decrease with increasing SiO₂ while Na₂O and K₂O increase (Fig. 8). Spring Mountain basalts average ~50 wt% SiO₂, and major elements show no obvious relationship with SiO₂, though one sample has higher TiO₂ and Al₂O₃ and lower MgO than the rest (Fig. 8). Pilot Knob samples have lower FeO*, MgO, and CaO and higher Al₂O₃, Na₂O, K₂O, and P₂O₅ at a given weight percent SiO₂ than Lava Mountain rocks (Fig. 8). Pilot Knob K₂O values range from 3.6 to 4.0 wt%, which is much higher than in any other UWRB sample. A similar relationship is exhibited by the Crescent Mountain dike (e.g., high K₂O, P₂O₅), but at the highest SiO₂ values, Crescent Mountain lavas fall along the co-linear array of increasing or decreasing major elements versus SiO₂ defined by Lava Mountain (Fig. 8). Wildcat Hill samples also fall along this array at slightly higher SiO₂ than Lava Mountain (Fig. 8). Trace element contents of UWRB samples show variations akin to those of the major elements; however, the differences between locations are more clearly depicted. This provides further evidence that each of these eruptive loci and intrusions represents a separate magmatic system (Fig. 9). Pilot Knob is characterized by high Ba (>2000 ppm), Sr (>1700 ppm), Zr (>200 ppm), Rb (>30 ppm), and La (>160 ppm); only the Crescent Mountain dike has similar trace element characteristics (Fig. 9). These incompatible element abundances are significantly higher than in any other unit, and clear differences are also present in other trace elements (e.g., Nb, Sc) that show lower concentrations at the same SiO₂ value. Sampled Spring Mountain basalts have nearly identical trace element signatures to each other, generally forming a tight cluster on Harker diagrams, except for Ni and Cr (Cr not shown in Fig. 9), which is accounted for by differences in modal olivine (Fig. 9). Lava Mountain basaltic andesites, andesites, and dacites are also distinguishable from other units on these diagrams. Lava Mountain dacites have the lowest overall Ni (<12 ppm) and Cr (<27 ppm) concentrations, which contrasts strongly with the basaltic andesites carrying Ni concentrations >115 ppm and Cr concentrations >240 ppm. Overall, compatible elements Cr, V, Co, Sc, and Ni all decrease with silica (Co, V not shown in Fig. 9). Positive correlations are observed between Sc, Ni, and Cr, suggesting pyroxene fractionation.

A suite of representative samples from each location illustrates the same enrichments and depletions on a primitive mantle–normalized multi-element diagram (Fig. 10). All have pronounced large ion lithophile element (LILE) enrichment and high field strength element (HFSE) depletion. This is characteristic of subduction zones (Fig. 10). The Spring Mountain basalts are generally enriched in all elements except for Ti and REEs relative to normal mid-ocean-ridge basalt (N-MORB) and are clearly distinct from ocean-island basalt (OIB) (Fig. 10). These enrichments and observed relative depletions are reminiscent of basalts derived from lithospheric mantle enriched by prior subduction processes (Kempton et al., 1991). All of the UWRB rocks have steep light REE (LREE) patterns. The Pilot Knob and Crescent Mountain samples exhibit the highest degree of LREE enrichment and flat to slightly curved downward heavy REE (HREE) patterns relative to chondrite (Fig. 11). All Lava Mountain REEs show a small increase in concentration against SiO₂, which is consistent with fractional crystallization of phases that exclude REEs, and small negative Eu anomalies due to plagioclase crystallization. Such LREE enrichment and flat to curved HREE patterns in relatively primitive rocks (e.g., basalt, lamprophyre) is typically attributed to partial melting in the presence of garnet. The degree of LREE enrichment is also consistent with melting of LREE-enriched lithospheric mantle.

Our petrographic evidence indicates that UWRB rocks were affected by small degrees of crustal contamination. The Lava Mountain samples have major element trends displaying primarily positive or negative linear trends with SiO₂ (Fig. 8) and have enriched incompatible trace element characteristics relative to UWRB basalts (Fig. 10) consistent with an origin affected by fractional crystallization processes. Nonetheless, anhedral, partially resorbed xenocrystic microcline is observed in a Lava Mountain andesite (sample AD14-7) and suggests interaction with crust, as do quartz and plagioclase (with reaction rims) xenocrysts in other andesites, basaltic andesites, and dacites. One potential assimilant is the Archean Wyoming craton crust, which consists of a combination of gneiss, granodiorite, and metasedimentary rocks (Keefer, 1957; Foster et al., 2006). Geochemical evidence also suggests some amount of crustal contamination. For example, average Ce/Pb is 26.1 in N-MORB and 25 in OIB, while values <4 characterize the continental crust (Niu and O’Hara, 2003; Rudnick and Gao, 2003). In the Spring Mountain basalts, Ce/Pb is ~3–8 and in the other rocks reaches slightly higher values (~5–12). Similar relationships are present with Nb/U, another trace element ratio that in basalts (Nb/U > 40; Niu and O’Hara, 2003) can indicate crustal interaction (crustal Nb/U < 4; Rudnick and Gao, 2003). However, it is possible that these ratios in the UWRB rocks also reflect enrichments and depletions of these elements due to earlier subduction-related metasomatism of the underlying mantle. For example, Cenozoic basalts and andesites erupted adjacent to the Colorado Plateau in the southwestern USA have Ce/Pb ranging from 4.6 to 37 (Kempton et al., 1991). Kempton et al. (1991) demonstrated that the geochemical and Sr-Nd-Pb isotopic characteristics of these rocks reflect a derivation via melting of Archean lithosphere affected by a prior history of subduction-related fluid and melt enrichment and not contamination of asthenospheric melts with local and/or regional crust. Thus, while minor crustal assimilation affected some of the UWRB rocks, it may not have significantly changed the bulk chemistry of the least evolved rocks (e.g., Spring Mountain basalts, Pilot Knob).

Downey (2015) developed a preliminary two-step model to explain the generation of the evolved Lava Mountain magmas: (1) closed-system fractional crystallization of basalt to basaltic andesite and dacite, followed by (2) magma mixing of basaltic andesite and dacite (plus minor assimilation) to generate the andesite. The textures observed in the andesites and their essentially aphyric nature (e.g., ~85%–90% groundmass) are consistent with magma mixing coupled with relatively quick ascent and eruption (Johnson and Grunder, 2000; Brueseke and Hart, 2009). This model is also consistent with stratigraphic evidence that the initial and basal eruptions at Lava Mountain alternated between basaltic andesitic and dacite, followed by andesite eruptions that form the greatest volume of the volcano and the final pyroclastic phase (Downey, 2015). Additional petrologic details for the UWRB rocks are beyond the focus of this contribution.

A continental magmatic arc was present in northwest Wyoming in the Eocene, and it has been suggested that subduction-related magmatism also affected the region during the Proterozoic (Dudás et al., 1987; Mueller and Wooden, 1988; Carlson and Irving, 1994; Feeley, 2003). Modern subduction processes produce magmas that are enriched in LREEs and LILEs and depleted in HREEs and HFSEs (Smithies et al., 2004); all UWRB samples have these characteristic enrichments and depletions. Figure 12 illustrates these characteristics and compares the UWRB rocks to Yellowstone Plateau volcanic field basalts (Fig. 12A) and representative samples of the Jackson Hole volcanics and Leucite Hills (Fig. 12B). Clearly the UWRB rocks, including the Spring Mountain basalts high in Mg, Ni, and Cr, are geochemically distinct compared to basalts from the Yellowstone Plateau volcanic field (Fig. 12A), which generally lack the LILE enrichments and HFSE depletions that characterize the UWRB rocks. Similar trace element patterns are present in the ca. 8.5–8 Ma Jackson Hole volcanics (Adams, 1997) (Fig. 12B), which would be expected if they have a similar origin as UWRB rocks. Furthermore, Leucite Hills volcanic rocks also exhibit similar patterns (Fig. 12B), which Mirnejad and Bell (2006) interpreted, based on other lines of evidence, as reflecting their derivation from metasomatized Wyoming craton lithospheric mantle. Adams (2014) also suggested that the Jackson Hole volcanics were derived from a lithospheric mantle source, based on their bulk chemistry and ⁸⁷Sr/⁸⁶Sr (0.7061–0.7104) and ε_Nd (–18.5 to –22.5) values. The subduction signature in UWRB rocks (and these similar regional volcanics) must result from the prior history of subduction in this part of North America because of the age of the UWRB rocks and their location in Wyoming, i.e., thousands of kilometers inboard from active Miocene to modern subduction. Thus, it is possible that the UWRB rocks represent either direct melting of Wyoming craton lithospheric mantle (e.g., Spring Mountain basalts and Pilot Knob) or evolution from a mafic melt originally derived from lithospheric mantle (e.g., Lava Mountain and Crescent Mountain). More detailed study involving radiogenic isotopes is needed to test these hypotheses and resolve the characteristics and relative roles of mantle and crustal reservoirs in UWRB magmatism.

As discussed earlier, northwestern Wyoming, eastern Idaho, and southwestern Montana were affected by two phases of extension during the last ~18 m.y.: (1) mid-Miocene extension that is likely due to collapse of Laramide and Sevier uplifts (Constenius, 1996), and (2) a later <ca. 5 Ma phase that could result from the North American plate moving over the Yellowstone hotspot tail (Anders et al., 2009). The primary effect of the Yellowstone hotspot tail on the North American lithosphere is significant uplift in a generally parabolic shape that is typical for plumes (Anders et al., 1989; Anders and Sleep, 1992; Pierce and Morgan, 1992; Wegmann et al., 2007; Smith et al., 2009; Obrebski et al., 2011), coupled with increased heat flow and melt production. Evidence for ~1 km of uplift in the upper Wind River Basin is best exhibited by the narrow, deep canyons that have been cut into the highly erodible Eocene package, and the regional erosion surface that formed on the thick package of Eocene Absaroka strata on which the UWRB volcanics erupted (Fig. 4) (Pierce and Morgan, 1992; 2009). Furthermore, the exposure of the Pilot Knob and Wildcat Hill intrusions also demonstrate uplift and exhumation during the last few million years. Anders et al. (1989), Pierce and Morgan (1992), and Anders et al. (2009) suggested that there was a wave of extensional activity before voluminous caldera-forming silicic volcanism began at any given location along the Snake River Plain, attributed to the migration of the North American plate over the Yellowstone plume tail. Rodgers et al. (1990, 2002) documented this extension north and south of the eastern Snake River Plain and attributed it to the aforementioned migration. Vogl et al. (2014) also connected ca. 11–8 Ma uplift followed by normal faulting north of the Snake River Plain in Idaho to volcanism and hotspot passage related to the ca. 10.4–6.6 Ma Picabo volcanic field (Drew et al., 2013). Similar links are also documented in southwestern Montana (Fritz and Sears, 1993; Sears et al., 2009). In contrast, Brueseke et al. (2014) linked eastward-migrating ca. 16–12 Ma Jarbidge Rhyolite volcanism in northeastern Nevada to coeval migrating Basin and Range extension that predated major hotspot-related volcanism at a given longitude.

Pierce and Morgan (2009, p. 13) defined the active region of extension around Yellowstone as a zone of “incipient activity (or rejuvenation)” that refers to neotectonic activity along active or previously active faults and seismicity that surrounds and extends from Yellowstone (e.g., the tectonic parabola or Yellowstone crescent of high terrain; Fig. 1). The leading front of the present-day active belt of extension is east of the Absaroka Range and bisects the Wind River Basin and Wind River Range (Fig. 1), encompassing the upper Wind River Basin. Figure 1 also depicts inferred topographic swells associated with the Heise and Picabo volcanic fields (Smith et al., 2009), assuming a similar ~100–200-km-wide footprint between the leading front and back of the current tectonic parabola (Yellowstone crescent of high terrain) east and northeast of the respective volcanic field (Anders et al., 1989). Regardless of whether the tectonic parabola is caused by Yellowstone hotspot–North American plate interaction or instead due to Basin and Range extension, ample evidence demonstrates that it exists and has migrated northeast across the North American plate. Geophysical studies document high heat flow and increased mantle and crustal temperatures associated with the hotspot, now adjacent to and affecting the upper Wind River Basin (Kelbert et al., 2012; Huang et al., 2015). Thus, the combination of hotter mantle and crust, coupled with extension and uplift, provided ample opportunities for UWRB melt production and volcanism during the last ~5 m.y.

We suggest that UWRB magmatism was primarily triggered by temperature increase and lithospheric extension associated with the Yellowstone hotspot tail and its tectonic parabola as it moved (relative to North American plate motion) from Idaho to its current location in northwest Wyoming. This aligns with the model of Adams (2014) based on seismic attenuation tomography, where melt generation would have occurred due to thermal input from the Yellowstone plume, coupled with adiabatic decompression of hydrous lithospheric mantle. The general spatio-temporal relationship between the Jackson Hole volcanics and UWRB rocks depicted in Figure 1 is consistent with this hypothesis, where volcanism generally gets younger to the east. If this relationship were solely related to the movement of the plate over the hotspot tail and the resulting tectonic parabola, we would expect a simple eastward age progression. However, the age relations of dated UWRB rocks do not conform to a simple progression, even though all are younger than the Jackson Hole volcanics. For example, Pilot Knob, Crescent Mountain, and Lava Mountain are all found just west of Spring Mountain, and Lava Mountain itself is much younger than Pilot Knob even though both exist in essentially the same location. Undoubtedly some other control on magma emplacement is also playing a role on where UWRB magmatism occurred. We propose that this is due to crustal heterogeneities imparted in the region during prior Laramide and Sevier compression. Reactivation of Mesozoic faults because of Basin and Range extension and extension due to the Yellowstone hotspot tectonic parabola would provide ample locations to focus magma during UWRB magmatism. Love (1939), Keefer (1955), and Winterfeld and Conrad (1983) described southeast-trending normal faults, such as the Bear Creek fault, just east and north of Spring Mountain which lie parallel to the strike of major Mesozoic thrust faults that front the Washakie uplift (Fig. 2). Moreover, Love et al. (1979) mapped southeast-trending normal faults that extend to the southeast of the UWRB in the Wind River Range (Figs. 1, 2). These are all likely the result of post-Laramide extensional collapse of the Wind River uplift, given their southeast trend that is parallel to major compressional structures in Paleozoic to Mesozoic rocks in the Wind River and Washakie uplifts (Ketner et al., 1966; Winterfeld and Conrad, 1983). Lava Mountain (as well as Pilot Knob and Wildcat Hill) lies between the faults in the Wind River Range and along the strike of the Upper Yellowstone fault zone, one of the southeast-trending normal fault–bounded corridors that extends south from Yellowstone Lake (Figs. 1, 2; Pierce and Morgan, 1992; Christiansen, 2001). Earthquake epicenter locations from 1950 to 28 February 2017 downloaded from the U.S. Geological Survey earthquake database (https://earthquake.usgs.gov/earthquakes/map/) also show that recent earthquakes have occurred just northwest and southeast of Lava Mountain near mapped normal faults (Love and Love, 1983; Machette et al., 2001), further defining this corridor as a zone of active faulting. Moreover, Protska et al. (1979) and Rohrer (1966) documented Pliocene(?) mafic dikes that trend 10° and cut Eocene strata near Crescent Mountain. Protska et al. (1979) hypothesized that these dikes, Crescent Mountain, and Lava Mountain all were emplaced or erupted along a lithosphere-scale fracture system. This feature and the corridor of active faulting that extends northwest and southeast of Lava Mountain intersect at Lava Mountain–Pilot Knob, indicating that this junction has been a locus of magmatism during the last 3.6 m.y. Finally, normal faulting at Spring Mountain provides the clearest evidence of this connection between UWRB magmatism and extension. The vent locations at Spring Mountain are fault controlled, and the basalt erupted via these fissures. The styles of magmatism we document in the UWRB are consistent with prior observations that monogenetic volcanism is primarily associated with extension and increased rates of regional differential stresses (Takada, 1994; Cashman and Sparks, 2013).

In summary, field relations and modern seismicity provide unambiguous evidence that UWRB magmatism is spatially related to local and regional normal faults, some of which must reflect preexisting crustal heterogeneity that was imparted before active upwelling and faulting associated with Basin and Range extension and the Yellowstone hotspot affected the upper Wind River Basin. This relationship also explains the location and timing of the Jackson Hole volcanics just west of the UWRB, and the ca. 7–6 Ma dacitic Carlton Creek Volcanics and Palisades Dam–Calamity Point Andesite in eastern Idaho (Armstrong et al., 1980; Anders et al., 2009, 2014; Price, 2009; Fig. 1). The Carlton Creek Volcanics ⁴⁰Ar/³⁹Ar age determinations (Price, 2009) were calculated using the Fish Canyon Tuff flux monitor with an age of 27.84 Ma, and the Palisades Dam–Calamity Point Andesite age was determined by the K-Ar method (Armstrong et al., 1980), hence approximate ages and ranges are presented and used for comparison.

The Jackson Hole volcanic sequence crops out at the southern end of Jackson Hole, near where major Laramide and Sevier faults (e.g., Cache Creek and Jackson thrusts, respectively) intersect younger Basin and Range structures like the Teton normal fault, in a zone of differential extension (Adams, 1997). The Carlton Creek Volcanics are along the northern edge of the Big Hole Mountain Mountains, and the Palisades Dam–Calamity Point Andesite is along the eastern Caribou Range, adjacent to Basin and Range normal faults (Anders et al., 2009; Price, 2009). These faults are parallel to major north-northwest–trending thrust faults that were active during the Sevier orogeny (Yonkee and Weil, 2015), suggesting that preexisting structures (e.g., Laramide-Sevier thrust faults) controlled the location of Cenozoic normal faulting and magmatism, similar to what we propose for the upper Wind River Basin. Price (2009) confirmed that Carlton Creek volcanism occurred at the same time or just before major caldera-forming silicic volcanism initiated on the Snake River Plain in the Heise volcanic field. This is the same general age relationship we document in the upper Wind River Basin, where most UWRB magmatism occurs prior to major volcanism at Yellowstone, and is essentially the same relationship noted between widespread Miocene rhyolite volcanism across northeastern Nevada and central Snake River Plain caldera-forming volcanism to the north in Idaho (Brueseke et al., 2014). While ample evidence exists that Miocene extension occurred in the greater Yellowstone region and along the trend of the Snake River Plain in Idaho, a younger episode of eastward-migrating extension also affected the greater Yellowstone region and is associated with a similar eastward migration of magmatism, prior to Yellowstone volcanism. Lava Mountain is the youngest manifestation of these processes. As a result, it appears that in multiple locations across most of the length of the Yellowstone hotspot track through North America, major caldera-forming silicic activity was preceded, at a given longitude, by typically smaller-volume volcanism, likely associated with uplift and extension. We suggest that similar relations should characterize other examples of continental hotspots as exhibited in the age-progression pattern for the proposed Anahim hotspot track in British Columbia (Canada) (Kuehn et al., 2015), and can be used to identify them in other locations in both the modern and past geological records.

1. New ⁴⁰Ar/³⁹Ar ages coupled with existing K-Ar results indicate that <ca. 5 Ma igneous activity occurred in the upper Wind River Basin, Wyoming, and define the Upper Wind River Basin volcanic field (UWRB). This includes 4.8 Ma volcanism at Spring Mountain and ca. 3.5 Ma magmatism at Pilot Knob and Crescent Mountain, both of which predate the formation of the Yellowstone Plateau volcanic field. Lava Mountain, the youngest UWRB volcano that we have studied, formed at 397 ± 48 ka, which postdates third-cycle caldera-forming activity in the Yellowstone Plateau volcanic field.

2. Prior geological mapping suggested that Lava Mountain is a basaltic volcano, however new major element chemistry demonstrates that the volcano is not basaltic, but instead formed via eruptions of calc-alkaline basaltic andesite, andesite, and dacite. Other intermediate-composition magmatism occurred at Pilot Knob (alkaline lamprophyre), Wildcat Hill (calc-alkaline andesite to dacite), and Crescent Mountain (calc-alkaline basaltic andesite and dacite), while eruptions of tholeiitic basalt characterize Spring Mountain. UWRB igneous rocks have trace element characteristics typically associated with subduction (e.g., LILE enrichment, HFSE depletions), though they erupted in a location thousands of kilometers inland from the closest active volcanic arc. They are also geochemically dissimilar from young basaltic volcanism at Yellowstone. Other <9 Ma regionally exposed igneous rocks (e.g., Jackson Hole volcanics) have identical geochemical characteristics, as do magmas from the Leucite Hills, Wyoming, all of which have previously been attributed to melting of Archean lithospheric mantle.

3. UWRB magmas were emplaced and erupted along normal fault zones associated with regional extension, some of which likely reflect preexisting structural heterogeneity that originally formed via Mesozoic compressional faulting. We suggest that this process is related to interaction between the North American plate and the progression of the tectonic parabola associated with the Yellowstone hotspot, generally ahead of, and out of sequence with, voluminous silicic magmatism associated with the hotspot on the Snake River Plain and at the Yellowstone Plateau volcanic field. This style of out-of-sequence, hotspot-related volcanism can be used to identify hotspots in other locations.

This work was motivated by many discussions about upper Wind River Basin geology among M. Brueseke, W. Hart, C. Haley, and B. Currie at Miami University’s geology field camp (Dubois, Wyoming). Brueseke, Downey, and Dodd thank the Wells family for allowing us to stay at the Timberline Ranch (Dubois) during fieldwork, and John Morton at Miami University for assistance with analytical geochemistry. We thank Associate Guest Editor Ben Ellis, Dave Hacker, an anonymous reviewer, and Editor Shan de Silva for helpful comments and/or handling of this manuscript. This research was funded by: a Tobacco Root Geological Society Field Scholarship (Downey), a Wyoming Geological Association J.D. Love scholarship (Downey), a Kansas Geological Foundation research scholarship (Downey), a Society of Economic Geologists student research grant (Dodd), a Kansas State University Small Research Grant (Brueseke), and the Janet and Elliot Baines Professorship at Miami University (Hart).