The trachydacite complex of Mammoth Mountain and an array of contemporaneous mafic volcanoes in its periphery together form a discrete late Pleistocene magmatic system that is thermally and compositionally independent of the adjacent subalkaline Long Valley system (California, USA). The Mammoth system first erupted ca. 230 ka, last erupted ca. 8 ka, and remains restless and potentially active. Magmas of the Mammoth system extruded through Mesozoic plutonic rocks of the Sierra Nevada batholith and extensive remnants of its prebatholith wall rocks. All of the many mafic and silicic vents of the Mammoth system are west or southwest of the structural boundary of Long Valley caldera; none is inboard of the caldera’s buried ring-fault zone, and only one Mammoth-related vent is within the zone. Mammoth Mountain has sometimes been called part of the Inyo volcanic chain, an ascription we regard inappropriate and misleading. The scattered vent array of the Mammoth system, 10 × 20 km wide, is unrelated to the range-front fault zone, and its broad nonlinear footprint ignores both Long Valley caldera and the younger Mono-Inyo range-front vent alignment. Moreover, the Mammoth Mountain dome complex (63%–71% SiO2; 8.0%–10.5% alkalies) ended its period of eruptive activity (100–50 ka) long before Holocene inception of Inyo volcanism. Here we describe 25 silicic eruptive units that built Mammoth Mountain and 37 peripheral units, which include 13 basalts, 15 mafic andesites, 6 andesites, and 3 dacites. Chemical data are appended for nearly 900 samples, as are paleomagnetic data for ∼150 sites drilled. The 40Ar/39Ar dates (230–16 ka) are given for most units, and all exposed units are younger than ca. 190 ka. Nearly all are mildly alkaline, in contrast to the voluminous subalkaline rhyolites of the contiguous long-lived Long Valley magma system. Glaciated remnants of Neogene mafic and trachydacitic lavas (9.1–2.6 Ma) are scattered near Mammoth Mountain, but Quaternary equivalents older than ca. 230 ka are absent. The wide area of late Quaternary Mammoth magmatism remained amagmatic during the long interval (2.2–0.3 Ma) of nearby Long Valley rhyolitic eruptions.

The Mammoth Lakes area of Mono County (Figs. 1 and 2) has become one of California’s busiest recreational playgrounds and a regional center of real-estate development. The name applies to the town of Mammoth Lakes as well as to the cluster of resort-lined lakes in a large cirque southwest of town now called the Lakes Basin. The town has spread around the eastern base of Mammoth Mountain, a late Pleistocene (100–50 ka) pile of silicic lava domes, and has locally expanded up the lower slopes of the volcanic edifice. Looming nearly 1000 m above the downtown area (Fig. 3), much of the 5-km-wide edifice has been laced with chair lifts, gondolas, ski runs, and bike paths by the Mammoth Mountain Ski Area (MMSA), a corporate entity under permit from Inyo National Forest. Tourism to Mammoth Lakes is estimated as 1,300,000 visitors per winter and 1,500,000 per summer. Recent volcanic unrest prompted continuous geophysical and geochemical monitoring as well as the documentation of the frequency and characteristics of past eruptive episodes as contributions toward hazard mitigation and preparedness.

Peripheral to the Mammoth Mountain dome complex is a contemporaneous (230–8 ka) field (20 × 10 km wide) of scattered vents for basaltic, andesitic, and dacitic lava flows. The best known of the 37 peripheral eruptive units is the mafic intracanyon Devils Postpile flow along the Middle Fork San Joaquin River, at the western base of Mammoth Mountain and within Devils Postpile National Monument, a National Park Service enclave surrounded by extensive wilderness areas administered by the U.S. Forest Service. As many as 2000 visitors per day enter the monument during the summer season.

Geographic names cited in the text are located in Figures 1 and 2. Grid references to site locations mentioned in the text are given to 100 m using the Universal Transverse Mercator (UTM) grid (1927 North American datum, zone 11), which is shown on U.S. Geological Survey topographic maps of the area. The first three digits are easting, and the second three are northing. For example, the summit of Mammoth Mountain is approximated as 208/665; i.e., 20.8 km east, 66.5 km north. Occasionally, a fourth digit is added for precision to 10 m.

Recent Unrest

Geophysical unrest beneath Mammoth Mountain and in adjacent parts of the Sierra Nevada and Long Valley caldera has generated concern among residents, stakeholders, and geoscientists since at least 1980, when 4 magnitude 6 earthquakes shook the area. Extensive monitoring for three decades has documented numerous earthquake swarms, ground uplift and deformation, changes in hydrothermal systems, and emission of magmatic CO2 at several sites semiencircling Mammoth Mountain (Cook et al., 2001; Evans et al., 2002; Farrar et al., 1995; Foulger et al., 2003; Gerlach et al., 1999, 2001; Hill, 1996, 2006; Hill et al., 1990, 2003; Hill and Prejean, 2005; Julian et al., 1998; Langbein, 2003; Pitt and Hill, 1994; Pitt et al., 2002; Prejean et al., 2003; Rogie et al., 2001; Sorey et al., 1993, 1998). Such investigations led the U.S. Geological Survey, in cooperation with federal, state, county, and local authorities, to prepare a response plan for future episodes of unrest, calibrated for progressively more intense premonitory activity (Hill et al., 2002; currently being revised and updated).

Current Investigation

The varied and persistent unrest also prompted scrutiny of volcanological data already in hand (Bailey, 1989) and critical review of our understanding of the region’s volcanic history (Hildreth, 2004). The review led directly to our intensified geological and laboratory investigations, the results of which are reported here. In remapping the area at a scale of 1:24,000 (Hildreth and Fierstein, 2015), we identified 25 silicic eruptive units exposed at Mammoth Mountain and 37 basaltic, andesitic, and dacitic eruptive units in its near periphery (Table 1). All units were studied petrographically (Table 1) and analyzed chemically (Supplemental File 11). Most units were cored for laboratory determination of paleomagnetic directions (Supplemental File 22), and most were radioisotopically dated by the 40Ar/39Ar method (Table 2; Supplemental File 33).

Our primary goals were to determine the spatial and temporal limits of the alkaline Mammoth magmatic system, to record the number, frequency, composition, and style of its many past eruptions, and to clarify its apparent independence from the adjacent Mono-Inyo and Long Valley rhyolitic systems.

Previous Geologic Work

Comprehensive geologic mapping of the Mammoth Mountain area was first completed in the 1950s at a scale of 1:62,500 by Rinehart and Ross (1964) and Huber and Rinehart (1965a). At the same scale, Bailey (1989) adapted their regional mapping while elaborating in greater detail the volcanic and structural history of Long Valley caldera and its surroundings.

Topical and preliminary studies in the area have been many and varied. Important contributions to geochronology of the volcanic rocks were by Bailey et al. (1976), Mankinen et al. (1986), and Mahood et al. (2010). Distribution and composition of some of the volcanic rocks were presented by Huber and Rinehart (1965b, 1967), Vogel et al. (1994), Cousens (1996), and Bailey (2004). Benioff and Gutenberg (1939) investigated the young fissure called the Earthquake fault near Mammoth Lakes. Huber (1981) summarized evidence for the timing and magnitude of uplift of the central Sierra Nevada, for incision of the San Joaquin canyon, and for Pliocene closure of the local segment of the Sierran drainage divide.

Mammoth Mountain straddles the principal drainage divide (Fig. 1) separating the San Joaquin River system, which drains to the Pacific, from tributaries of the Owens River system, which is landlocked. The volcano grew atop one of the lowest saddles along the crest of the central Sierra Nevada, and the 2800 m passes adjacent to the 3369 m edifice are among the lowest. Here the divide also marks the structural transition between the Basin and Range extensional region and the rigid Sierra Nevada block. Moreover, adjacent to Mammoth Mountain, the divide is also the southwest topographic rim of Long Valley caldera (although the ring-fault structural margin of the caldera is buried ∼5 km northeast of Mammoth Mountain).

Basement Rocks

Mammoth Mountain was constructed at the convergence of four significant structural boundaries (Fig. 4). The volcanic edifice grew within a scalloped reentrant of the topographic wall of the 767 ka Long Valley caldera (Bailey, 1989), and it overlies the intrusive contact of a Late Cretaceous granite pluton with steeply foliated Mesozoic metavolcanic rocks (Huber and Rinehart, 1965a). The edifice also buries a regional contact (Morgan and Rankin, 1972; Brook et al., 1974) between the metavolcanic belt and Paleozoic metasedimentary rocks of the Mount Morrison roof pendant (Rinehart and Ross, 1964; Greene and Stevens, 2002). Mammoth Mountain overlies the Rosy Finch shear zone segment of the Sierra Crest shear zone system (Tikoff and Teyssier, 1992; Greene and Schweickert, 1995; Tikoff and Greene, 1997), a north-northwest–striking dextral transpressional structure of Late Cretaceous age that foliates both metavolcanic and granitic basement rocks nearly vertically.

Pre-Quaternary rocks beneath and near Mammoth Mountain can be summarized as Paleozoic metasedimentary rocks, Mesozoic metavolcanic rocks, Late Cretaceous granitoid plutons, and Neogene volcanic rocks.

Paleozoic metasedimentary rocks form most of the south wall of Long Valley caldera and strike northwest from Mammoth Rock (a towering monolith of late Paleozoic marble) beneath the eastern part of Mammoth Mountain. The steeply dipping strata consist of Cambrian through Permian marine formations that extend as a series of pendants and septa for ∼115 km from Bishop Creek to Mono Basin. The belt is ∼10 km wide, represents a stratigraphic thickness of ∼7 km, and was severely folded and faulted during the Permian and Mesozoic (Greene and Stevens, 2002). The rocks include siliceous, calc-silicate, and pelitic hornfels, argillite and slate, fine to coarse sandstone and quartzite, black chert, and limestone and/or marble, all metamorphosed in the hornblende-hornfels facies at ∼200 MPa during Mesozoic pluton emplacement. The fine-grained clastic sediments are interpreted to have been deposited in deep water on the continental slope or rise, the chert hemipelagically, the sands on submarine fans, and the limestone on a carbonate platform (Stevens and Greene, 1999).

Mesozoic metavolcanic rocks strike beneath Mammoth Mountain and form a steeply southwest-dipping sequence dominated by silicic pyroclastic rocks of Triassic and Jurassic age. South of Mammoth Mountain, they form a septum 2.5–4 km wide that extends 12 km southeast, separating the two large Cretaceous plutons described in the following (Rinehart and Ross, 1964). North of Mammoth Mountain, they form a 5-km-wide belt that extends more than 40 km northwest, separating the Paleozoic metasedimentary section on the east from the Cretaceous metavolcanic complex of the Ritter Range on the west (Huber and Rinehart, 1965a). Both metavolcanic belts are complexly stratified homoclines that young southwestward, having foliation and bedding that typically dip 60°–80°SW. Lithologically, massive to stratified intermediate to rhyolitic tuffs predominate, including a few quartz-bearing ignimbrites. Stratified tuffs rich in crystals and lithic clasts include fallout, pyroclastic flow, and reworked deposits. Also present are aphanitic to porphyritic lava flows, shallow intrusives, and subordinate interbeds and lenses of shale, sandstone, siltstone, calc-silicate hornfels, marble, and tuffaceous sedimentary rocks (Rinehart and Ross, 1964). Major ignimbrite sheets and plane-parallel beds of intercalated sandy sediments and tuffs suggest subaerial and shallow subaqueous depositional environments of low relief. The entire sequence is in the hornblende-hornfels facies, locally schistose, and was deformed and metamorphosed before and during emplacement of the Late Cretaceous batholith.

The Mono Creek Granite (Bateman, 1992) is one of the largest Late Cretaceous plutons of the Sierra Nevada, a 50-km-long, northwest-elongate, medium- to coarse-grained biotite granite, rich in alkali-feldspar megacrysts (unit Kmo; Fig. 4). Near Mammoth Mountain, the granite forms Mammoth Crest, Crystal Crag, the north wall of Fish Creek canyon below Pumice Butte, and both walls of the Middle Fork San Joaquin canyon, terminating against metavolcanic rocks 6 km north of Devils Postpile. The western half of Mammoth Mountain banks against and conceals a northeast-sloping wall of Mono Creek Granite that formed a steep scalloped alcove in the Long Valley caldera margin. The scallop had become a middle Pleistocene cirque (during Marine Isotope Stage, MIS, 6 and probably earlier) before being filled by the trachydacite edifice in the late Pleistocene. Mammoth Mountain does not drape the caldera rim; it grew within the scallop, and its lava flows bank against the granite wall. The granite crops out as far east as Lake George and Horseshoe Lake in the Lakes Basin. Along the Middle Fork west of Mammoth Mountain, all Neogene and Quaternary volcanic units overlie Mono Creek Granite.

Another northwest-elongate pluton, correlated (by Rinehart and Ross, 1964) with Round Valley Peak Granodiorite (Bateman, 1992), is truncated by the caldera wall near Mammoth Rock but also appears to project under the northeast part of Mammoth Mountain (Fig. 4). The pluton is 3–5 km wide, crops out for 15 km southeastward, encloses the canyon of Sherwin Creek, intrudes along the contact between metavolcanic and metasedimentary sequences, and consists of Late Cretaceous medium-grained, hornblende-biotite granodiorite (unit Krv; Fig. 4).

Neogene Volcanic Rocks

Remnants of numerous Pliocene mafic and trachydacitic volcanic rocks (and a few of late Miocene age) are present within 15 km north, west, and south of Mammoth Mountain (Huber and Rinehart, 1965a; Bailey, 1989, 2004). Northwest of Mammoth Mountain, 14 Pliocene vents were indicated by Bailey (1989) along a 12-km-long reach of the Sierran drainage divide, and along the northwest to north-central rim of the caldera he mapped 18 more Pliocene mafic and intermediate precaldera vents. For these and additional precaldera mafic lavas east of the caldera, Bailey (2004, fig. 3 therein) compiled numerous radioisotopic dates that range from 4.5 to 2.3 Ma, predominantly 3.7–2.7 Ma.

Judging from surviving outcrops, the Pliocene volcanic rocks were distributed across what is now the western half of the caldera prior to glaciation and caldera collapse. Only 4 km north of Mammoth Mountain, such rocks are as thick as 466 m beneath the Long Valley caldera floor, where they were penetrated by a geothermal well at depths of 1168–1634 m (Suemnicht, 1987; well 44–16; Fig. 5). Sparse fragments of concealed intracaldera Pliocene basalt have also been expelled in the rhyolitic ejecta ring of the 150 ka West Moat coulee (unit rwm; Figs. 1 and 5), 3 km from the northeast toe of Mammoth Mountain.

North, south, and west of Mammoth Mountain, we have reinvestigated Neogene mafic and trachydacitic lavas in more detail (and we have dated several by 40Ar/39Ar), because they are spatially associated with late Pleistocene eruptive units and were not clearly distinguished from them by previous investigators. Altogether, we have mapped and studied 32 Neogene eruptive units within 15 km of Mammoth Mountain (see Table 1 for map locations). Three are olivine-rich remnants of mafic lava flows (units Tbmm, Tbrm, and Tmmc) high on the cirque rim of the Lakes Basin just south of Mammoth Mountain, one of them dated as 3.15 ± 0.1 Ma. Along and near San Joaquin Ridge northwest of Mammoth Mountain, several mafic and trachydacitic units cluster in age between 3.7 and 2.6 Ma. On the south rim of Long Valley caldera, the trachydacite of Laurel Mountain (unit Tdlm) yields a date of 3.6 Ma. In the San Joaquin canyon near Devils Postpile, the thick and extensive basalt of the Buttresses (unit Tbtb), formerly thought to be Quaternary, gave a date of 3.75 Ma. A pair of lava domes (unit Tacc) ∼6 km southwest of Mammoth Mountain has been dated as 4.3 Ma. Along and near the northwest wall of the caldera, ∼11 km north of Mammoth Mountain, several distinguishable sets of lava flows have yielded ages of 3.7–3.0 Ma. Details of these and more Neogene eruptive units are given in Hildreth and Fierstein (2015).

The distribution of these pre-Quaternary volcanic units and others mapped by Bailey (1989) east and northeast of Long Valley caldera shows that an extensive mafic to trachydacitic volcanic field spread across the subsequent sites of Long Valley caldera and Mammoth Mountain during the Pliocene. Starting in the early Quaternary, however, and for the next 2 m.y., only subalkaline rhyolites (2.2–0.3 Ma) are known to have erupted anywhere in the Long Valley region. After this hiatus of 2 m.y., mafic eruptions resumed ca. 230 ka, initiating the alkaline Mammoth Mountain basalt to rhyodacite episode.

The central dome complex at Mammoth Mountain consists of at least 25 distinguishable silicic units that erupted between 100 and 50 ka, an interval ∼50 k.y. long. The silicic edifice accumulated near the center of a distributed field of small monogenetic mafic and intermediate volcanoes that we have divided into 37 separate eruptive units (Figs. 5–13), the eruptive volume of which we estimate to total between 7 and 12 km3. This peripheral monogenetic activity, which principally produced mildly alkalic basalt and its differentiates, began ca. 230 ka and has continued sporadically into the Holocene. Growth of the trachydacitic Mammoth Mountain edifice thus started up near the middle of the basalt-driven volcanic field, in time as well as space. Vents for the monogenetic volcanic rocks peripheral to Mammoth Mountain scatter within an ovate field ∼20 × 10 km across, which evidently reflects a new, well-circumscribed domain for crustal penetration by mantle melts. Before initiation of Mammoth magmatism, there had been no eruptions nearby since the Pliocene, other than the Long Valley subalkaline rhyolite. Contemporaneous with Mammoth activity (230–8 ka), there was no mafic or intermediate eruption any closer than June Lake (20 km north of Mammoth Mountain) and Big Pine (85 km southeast). The Mammoth volcanic field is thus a new magmatic system that began ca. 230 ka and contrasts with the Long Valley system in being alkaline (Hildreth, 2004).

Vents for the 37 peripheral units are distributed no farther from the summit of Mammoth Mountain than 4.6 km west, 6.8 km northeast, 11 km north, and 9 km south. Some exposed vents are as close as 4 km south, 4.4 km north, 3 km northeast, and 3 km southeast of the summit, some right at the toe of the edifice. Vents for several other mafic and intermediate units are buried by the edifice, by adjacent surficial deposits, or by the cluster of 150–100 ka rhyolite and trachydacite domes at the northeast toe of the edifice.

230–150 ka

The 37 peripheral units include 13 basalts, 15 mafic andesites, 6 andesites, and 3 dacites. A formal unit name and 3 letter label, along with phenocryst mineralogy and SiO2 content, is given for each in Table 1; 9 of them erupted before ca. 150 ka. The oldest exposed unit dated is a phenocryst-poor apron of basaltic trachyandesite lava flows (unit mcl) that issued at the Canyon Lodge scoria cone (Fig. 6) and spread as far as 9 km northeast and 8 km east. Distal exposures are at the foot of Lookout Mountain and near the geothermal plant at the former site of Casa Diablo Hot Springs. Unit mcl is encountered in several wells (Hildreth and Fierstein, 2015) but is widely concealed by younger units, and yields dates as old as ca. 190 ka, but the most precise are 170–180 ka. Phenocrysts include sparse olivine and plagioclase and rare clinopyroxene (cpx).

Directly beneath unit mcl, a trachyandesite lava flow (unit asr; undated) is exposed only near Shady Rest Park and Campground (Fig. 7). The nearly aphyric flow (56.5% SiO2) carries rare phenocrysts of cpx and plagioclase. Several wells in the west moat and in Mammoth Lakes township penetrated the mcl-asr pair. Locations and lithologic logs of numerous wells in Long Valley caldera are given in Hildreth and Fierstein (2015).

In the Inyo-4 corehole near Inyo Craters (Eichelberger et al., 1985, 1988), a set of flows compositionally identical to unit mcl (upper Group II of Vogel et al., 1994) yields a 40Ar/39Ar date of 186 ± 2 ka. This set is underlain by a 170-m-thick stack of ∼18 mafic lava flows (Groups II–IV of Vogel), possibly a small shield nowhere exposed, for which 3 40Ar/39Ar dates (Table 2) are as old as 233 ± 4 ka. These unexposed mafic lavas below unit mcl are the oldest known products of the Mammoth volcanic field. Their vents are unknown but certainly buried in the west moat or beneath Mammoth Mountain.

Also in the west moat is a glacially scoured trachybasalt spatter cone (unit bcf; 164 ± 2 ka) just south of Crater Flat and 3 km north of the subsequent toe of Mammoth Mountain (Fig. 8). Its agglutinate and poorly exposed lava flows contain 1% plagioclase, traces of olivine and cpx, and sparse xenocrysts of quartz and feldspar.

On the south slopes of Knolls Vista (hill 2517; Fig. 7), adjacent to downtown Mammoth Lakes, units asr and mcl are successively overlain by a basaltic lava flow (unit bsm; 165 ± 2 ka) and a set of fountain-fed flows of basaltic trachyandesite (unit mkv; 153 ± 1 ka) that issued from an agglutinated vent complex on the hilltop, against the north edge of which another set of basaltic trachyandesite flows (unit msc; 154 ± 2 ka) banked. Vents for units bsm and msc were buried by effusion of the rhyolitic West Moat Coulee (unit rwm; ca. 150 ka). Unit bsm also crops out as a steep ridge (Fig. 7) that extends eastward from beneath the rhyolite coulee, 3–4 km north of downtown Mammoth Lakes, and as low-relief exposures near the southwest foot of Lookout Mountain (Fig. 13). It contains sparse olivine and plagioclase phenocrysts as well as quartz and feldspar xenocrysts. Units mkv and msc both form 100-m-high ridges consisting of several poorly exposed flows, and both are phenocryst poor. Unit msc carries 1%–2% each of plagioclase and olivine and rare cpx, whereas unit mkv contains <1% of all three phases combined.

Eastward along the south moat (Fig. 7), a basaltic trachyandesite lava flow (unit mmc; 162 ± 2 ka; Mahood et al., 2010) is compositionally indistinguishable from agglutinate of its presumed vent complex mkv. It directly underlies the Casa Diablo Till (MIS 6). Still farther east, the stratigraphically lowest lava flow exposed in the south moat sequence is a low-relief sheet of trachybasalt (unit bsc; 172 ± 2 ka) that crops out near Chance Ranch south of Mammoth Creek. It contains ∼1% phenocrysts of plagioclase and olivine and sparse quartz and feldspar xenocrysts. In water wells near Mammoth Lakes (Hildreth and Fierstein, 2015), this basalt overlies units mcl and asr.

The earliest of several mafic monogenetic eruptions south of Mammoth Mountain (Figs. 9 and 10) also took place during this opening interval. Approximately 7 km south of the edifice, basaltic lava flows of unit bdc (155 ± 2 ka), rich in olivine and plagioclase, form a platform that extends south from beneath andesitic Cone 2962 (unit a62; 118 ± 10 ka).

150–100 ka

The next 50-k.y.-long interval, from ca. 150 ka to ca. 100 ka, was marked by eruption of Long Valley’s west moat rhyolites (units rwm, rdm, rdc, and rmk), the first extrusion of phenocryst-rich Mammoth Mountain trachydacite (unit d81; 99 ± 8 ka), and 9 more peripheral mafic and andesitic units. Among these were two 150-m-high trachyandesite scoria cones, Pumice Butte (unit apb; 142 ± 5 ka) and Cone 2962 (unit a62; 118 ± 10 ka), which erupted 7–8 km south of Mammoth Mountain (Figs. 9 and 10). Separated by only 2 km, each cone produced lava-flow aprons; that of Cone 2962 is much more extensive. Three vent stars indicated just north of Pumice Butte on the map of Bailey (1989) are not vents but simply rugged surface features on the partly unglaciated apron of unit a62. Products of both cones contain sparse phenocrysts of olivine and plagioclase and trace amounts of cpx.

A scoria cone of basaltic trachyandesite (unit mss; 121 ± 2 ka) erupted 4 km north of Devils Postpile (Fig. 11), through the bed of the Middle Fork San Joaquin River, which subsequently incised its interior, exposing two dikes on the riverbank opposite Soda Springs Campground. Lava flows crop out along the riverside both upstream and downstream of the cone. The phenocryst-poor lavas and ejecta carry sparse olivine and plagioclase and are overlain by unit mdp (82 ± 1 ka), which is far richer in olivine and plagioclase and also has cpx.

In the south moat of Long Valley caldera (Fig. 7), a phenocryst-rich lava flow of basaltic trachyandesite (unit mlc; 130 ± 1 ka) is exposed only at its rugged distal terminus, where it is sandwiched by flows of units bsc (172 ± 2 ka) and bfh (92 ± 2 ka) just south of Mammoth Creek. Entirely beyond the limit of glaciation, exposures are pervasively vesicular and exhibit unusual local relief of ∼30 m, including ridges, hillocks, tumuli, and tilted surfaces that are fissured and polygonally jointed. The rock is rich in cpx, olivine, and plagioclase. Its source vent is buried by younger units, probably many kilometers to the west.

In the south moat, a low-relief apron of phenocryst-rich basalt (unit bcd; 125 ± 2 ka) extends 3.5 km southeastward to Mammoth Creek and the geothermal plant (Fig. 7). No exposure is thicker than 15 m and, because outside the limit of glaciation, most exposures are highly vesicular. Outcrops are gently rolling scoriaceous surfaces or block-jointed ledges, commonly crudely columnar. The unit is rich in cpx, olivine, and plagioclase. Its source vent is probably spatter cone 2580+ (unit bed; 121 ± 3 ka; Fig. 8) just west of Dome 2861 (unit d61; 87 ± 2 ka), which now separates the cone from exposures of its apron lavas; the cone agglutinate is similar to unit bcd petrographically and in major and trace element composition.

In the west moat, an undated basaltic trachyandesite scoria cone (unit mdm) crops out at the northeast toe of Deer Mountain dome (Fig. 8). Plagioclase-bearing but nearly aphyric scoria and agglutinate are partly covered by rhyolitic Deer Mountain (unit rdm; 101 ± 8 ka) and heavily mantled by 1350 CE phreatic ejecta from Inyo Craters. In the west moat, a stack of at least five phenocryst-poor trachyandesite lava flows (unit aic; 131 ± 1 ka) is exposed on the walls of South Inyo Crater and along a south-striking fault scarp for ∼1 km south from that crater (Fig. 8). Flows are thin and rubbly, but their massive interior zones support ledges 3–6 m thick. All have sparse plagioclase and sparser olivine and cpx (both ≤1 mm), and some flows carry rare feldspar and quartz xenocrysts as large as 10 mm. The source vent is buried by younger deposits and has not been located. Another west moat unit in this time interval is a glacially scoured coulee of the weakly porphyritic orthopyroxene (opx)-plagioclase trachydacite of Dry Creek (unit ddc; 103 ± 9 ka; Fig. 8), which erupted from a vent subsequently buried by construction of the Mammoth Mountain edifice. The flow is as thick as 40–50 m, and is everywhere glacially eroded into ridges, knobs, bluffs, and ledges. Exposures range from glassy to finely devitrified with pale gray mottling, and from massive and block jointed to flow foliated and thinly platy; the platy jointing is commonly ramped or nearly vertical.

100–50 ka

The next 50 k.y interval included incremental construction of the Mammoth Mountain edifice (ca. 90 ka to ca. 50 ka) and eruption of 14 peripheral units; 5 basalts, 6 mafic andesites, 2 andesites, and 1 dacite. Two of the oldest in this interval are extensive lava flows on opposite sides of Mammoth Mountain, unit bsr (ca. 104 ka) in the south moat and unit drf (98 ± 1 ka) along the Middle Fork; both may have erupted from vents now covered by Mammoth Mountain. Unit bsr is a low-relief apron of phenocryst-poor (plagioclase, olivine, cpx) trachybasalt that emerges from beneath Tioga moraines and extends 5.5 km southeast to Laurel Creek (Fig. 7). The trachydacite of Rainbow Falls (drf) is a single coulee that floors the Middle Fork for 6 km south of Reds Meadow (Fig. 11), fills paleochannels in Cretaceous granite, thickens downstream from 30 m to 80 m, and contains sparse plagioclase, opx, biotite, and hornblende. Both units were overrun by aprons of phenocryst-poor trachyandesite lavas (unit amp; 97 ± 1 ka), which erupted on the drainage divide at the later site of Mammoth Mountain and poured both eastward along Mammoth Creek and westward into the Middle Fork (Figs. 11 and 12). Exposures of unit amp are ∼30 m thick near upper Twin Lakes, 100 m near Horseshoe Lake, and >140 m along Boundary Creek where at least 5 cliffy flows are exposed, each 10–30 m thick. The vent is covered by the Mammoth Mountain dacite pile. Vents shown by Bailey (1989) at Hill 2938 (Fig. 12) and Sherwin Creek Campground (Fig. 7) do not exist. Phenocrysts include ∼1% plagioclase and traces of olivine and cpx.

Undated trachyandesite unit aml is a small glaciated remnant of phenocryst-poor hornblende-opx-plagioclase silicic andesite at McLeod Lake (Fig. 12), just south of Mammoth Mountain. The vent is not exposed or located, but a 10-m-thick lapilli-fall member of the unit penetrated by drilling at Horseshoe Lake suggests its proximity. In this well, the fall deposit directly underlies a Mammoth Mountain trachydacite lava flow (elsewhere unexposed), and in a well drilled near downtown Mammoth Lakes a lava flow compositionally identical to unit aml closely underlies unit amp (ca. 97 ka), suggesting that the undated unit may not be much older. It is certainly younger than 130 ka, because in the downtown well the aml lava overlies the Casa Diablo Till (MIS 6), which in turn overlies mafic units msc (ca. 154 ka) and mcl (ca. 175 ka). Lithologic logs for these wells are given in Hildreth and Fierstein (2015).

The famous colonnade called Devils Postpile is part of a 110-m-thick intracanyon flow of basaltic trachyandesite (unit mdp; 82 ± 1 ka) preserved discontinuously along the floor of a 7-km-long reach of the Middle Fork (Fig. 11), where it overlies units mss, drf, and amp. Outcrops are widely striated and plucked, glacially eroded into knolls, ridges, and sidewall benches, but largely massive with only sparse scattered vesicles. Despite columns of many orientations, evidence for ice-contact emplacement is not observed; the flow covered wide axial areas of the granitic canyon floor, and a valley glacier would have been unlikely during the warm interval ca. 82 ka. The rock is phenocryst rich (7%–10% plagioclase, 2%–3% olivine, and ∼1% cpx), distinguishing it from the three crystal-poor subjacent units just mentioned.

A basaltic scoria cone (unit bmc; 92.7 ± 2.4 ka) is preserved atop the Sierran drainage divide at Mammoth Crest, above the Lakes Basin and only 4 km south of the summit of Mammoth Mountain (Fig. 12). It was the source of an apron of lava flows (unit bfh; 92.5 ± 2 ka) rich in large plagioclase phenocrysts that extended 18 km down Mammoth Creek (Figs. 6 and 7), but is exposed for only its distal 6 km. The unit is also identified beneath glacial deposits in water wells near downtown Mammoth Lakes (Hildreth and Fierstein, 2015). Although eroded or covered by younger lavas and till for the first 12 km away from the cone, distal lava near the Hot Creek flow (unit rhc) is chemically, petrographically, and paleomagnetically identical to the cone; moreover, the proximal and distal sites yield indistinguishable 40Ar/39Ar ages. An in situ remnant ∼55 m above Crystal Lake shows that the Lakes Basin cirque contained little or no ice at the time of eruption. Phenocrysts include 8%–15% plagioclase (1–22 mm), 1%–2% olivine, and sparse cpx.

Another basaltic scoria cone is well preserved on the caldera wall 2 km northeast of Minaret Summit and 4.4 km north of Mammoth Mountain (Figs. 6 and 8; see also Fig. 14). It was the source of two aprons of mutually distinctive lava flows in the west moat; trachybasaltic unit bmn (87 ± 7 ka) and overlying phenocryst-richer basaltic unit bar (88 ± 5 ka). Thick scoria fall incised by a gulch at the south foot of the cone contains bombs of both lithologies, and is overlain by a lava-flow remnant of unit bar. Flows of unit bmn extend only 3 km northeast from the cone, are everywhere glacially scoured, contain 2%–5% plagioclase and 1%–3% olivine phenocrysts, and overlie unit ddc. Unit bar is far more extensive and contains 15%–25% plagioclase, 2%–5% olivine, and trace cpx. After spreading across much of the southwestern moat, flows of unit bar extended 13 km northeast to bank against the north wall of Long Valley caldera, thence 10 km eastward to a terminus along the Owens River in the caldera’s north moat (Fig. 6). Flows are glaciated south of Crater Flat but ruggedly scoriaceous beyond the glacial limit. A 400-m-wide slab of unit bar (Figs. 6 and 8) was uplifted (along with subjacent unit mcl) on the roof of Dome 2861 (unit d61; 87 ± 2 ka).

A lobate flow field of basaltic trachyandesite lavas extends 10 km northeast along the west moat to the caldera’s north wall near Big Springs (Fig. 13). The southern half is exposed as a series of windows through MIS 2 till, but beyond the glacial limit the northern half branches into three lobes, the central one widening into a single coulee 5 km long and 30–50 m thick. We separated the lobate field into two map units (mnd and mor), variants that range in phenocryst content from aphyric to ∼4% plagioclase (and trace olivine), but there are gradational areas and many rootless vents and fissures that complicate the outflow pattern. The mapped variants have overlapping ranges of composition (51.2%–54.2% SiO2) and indistinguishable paleomagnetic directions. The coulee, unit mor, gave a 40Ar/39Ar age of 66 ± 2 ka. The vent may be a poorly exposed patch (unit msd) of scoria lapilli and bombs (Fig. 8) that is compositionally and mineralogically similar to units mor and mnd but is largely covered by glacial and colluvial deposits in the saddle between trachydacite Domes 2861 and 2781.

Unit msj is moderately porphyritic, cpx-plagioclase-olivine, subalkaline basaltic andesite lava that crops out directly beneath unit mnd (66 ± 2 ka) in a small window at the north end of the graben followed by Highway 395 (Fig. 13). Its source vent is unknown, and it crops out nowhere else. No source vent has been recognized for an undated lithic-rich scoria-fall deposit of basaltic trachyandesite (unit mic), several meters thick and exposed (poorly) only on the walls of the Inyo Craters (Fig. 8), where it directly overlies unrelated unit aic (131 ± 1 ka). Scoriae are mostly lapilli (but as large as 20 cm), contain 54.3%–55.8% SiO2, and carry sparse olivine and plagioclase phenocrysts. The weathered top of the deposit is overlain by thin Mono Craters ash fall, then by magmatic and phreatic ejecta of the 1350 CE Inyo episode (Mastin, 1991). Unit mic may be a phreatomagmatic deposit or, alternatively, a phreatic deposit that mined till and scoria from strata concealed somewhere beneath unit aic. The vent was evidently nearby but much older than the 1350 CE excavation of Inyo Craters.

50–0 ka

Subsequent to the last silicic eruptions of Mammoth Mountain, four mafic eruptions took place in its periphery as well as extrusion of a chain of five trachydacite lavas in the northwest moat of Long Valley caldera.

A glaciated sheet of phenocryst-rich olivine-cpx-plagioclase basaltic lava flows (unit bhl; Fig. 12) banked against the southeast toe of Mammoth Mountain near Horseshoe Lake at 31 ± 1 ka. The thickest exposures are ∼60 m on cliffs overlooking Twin Lakes, and the remnants flooring parts of the Lakes Basin show that the cirque was ice free at the time of eruption. The vent was beneath Horseshoe Lake, which was subsequently glacially excavated, and its proximal ejecta are marginally preserved along the eastern lakeshore. Distinctive erratics of 31 ka unit bhl in all of the ∼40 late Pleistocene moraines nested along Mammoth Creek prove that much of the till came from the Lakes Basin and was deposited during MIS 2, not earlier.

A large scoria cone of crystal-poor basaltic trachyandesite, 6 km north of the summit of Mammoth Mountain and just west of Crater Flat (Fig. 8), produced a lava-flow apron (unit mcv; ca. 33 ka) that extends 9 km northeast to bank against the caldera wall at Crestview (Fig. 13). All but the distal 3 km of the apron is concealed by younger lavas and surficial deposits. Its rough, little-eroded distal surface is marked by low swells and hillocks, a few crags, and scattered small tumuli. Flows are block jointed, vesicular to scoriaceous, and commonly oxidized reddish-brown; they contain only 1%–2% plagioclase and traces of olivine and cpx.

A northwest-trending set of three glacially ravaged remnants of basaltic trachyandesite scoria, agglutinate, and fountain-fed lava (unit mdn; 16 ± 2 ka) is the product of a vent alignment on the caldera’s west wall (Fig. 8), 1.5 km northeast of Deadman Pass and 5–6 km north of Mammoth Mountain. Scoria-fall lapilli of unit mdn are scattered sparsely atop Dome 2861 (unit d61; 87 ka), ∼6 km southeast of the vents (Fig. 15). Phenocrysts include 10%–15% plagioclase and 1%–2% each of olivine and cpx. Except for an outcrop on the floor of Deadman Creek (UTM 2115/7618), the lava-flow apron is largely buried by till and 1350 CE fallout. Banked against the toe of Lookout Mountain (Fig. 13), 9 km northeast of the vents and 2 km beyond the glacial limit, a fragmental flow deposit >10 m thick consists entirely of angular clasts of unit mdn, some larger than 1 m; one such block gave an age of 17 ± 1 ka, indistinguishable from that of the dense proximal lava.

Red Cones (unit brc) is a pair of 120-m-high basaltic scoria cones, 3–4 km southwest of Mammoth Mountain (Figs. 10 and 11). They jointly fed a 1.2 km2 fountain-fed apron of thin scoriaceous lava flows southwest of the cones. The most recent eruption in the Mammoth Mountain periphery, this episode released ∼0.01 km3 of magma, equally divided between fragmental deposits and lavas. Phenocrysts include 5%–15% plagioclase and 5%–10% spinel-bearing olivine; cpx is common as microphenocrysts. Bits of charcoal in sediment beneath a thin ash-fall layer 4.5 km northeast of the cones gave a limiting radiocarbon age of ca. 8.5 ka (Browne et al., 2010).

A southeast-trending nonglaciated chain of 5 comagmatic trachydacite domes and coulees (unit dnw) crosses the northwest moat of Long Valley caldera, 11–12 km north of Mammoth Mountain (Fig. 13). All five are glassy and rich in phenocrysts of plagioclase, sanidine, hornblende, and biotite, but they are variably contaminated with mafic enclaves, microdioritic blebs, disequilibrated crystals, and granite-derived xenocrysts, thus ranging in bulk composition from 67.3% to 60.4% SiO2. The five vents are approximately aligned, the lavas are generally more contaminated southeastward, and nearly identical paleomagnetic directions suggest they all erupted in a brief episode. However, 40Ar/39Ar dates range from 42 ± 1 ka to 27 ± 1 ka (Mahood et al., 2010; Table 2). The two southeastern coulees, although most contaminated, yield ages younger than the three northwest units. The southeasternmost coulee (27 ± 1 ka) directly overlies mafic apron lavas of unit mcv (ca. 33 ka), supporting its young 40Ar/39Ar date. Our repeated attempts to verify the 27–30 ka ages of the two southeastern coulees, however, have been foiled by apparent contaminants.

Mammoth Mountain is a pile of overlapping silicic domes and coulees (Figs. 3, 14, and 15) that extruded from a vent cluster only 2 km across and spread laterally to form a 5-km-wide edifice that has ∼850 m of relief. The edifice consists of 22 compositionally and morphologically distinguishable units of effusive lava, among which only 5 (units dfl, dnh, dsd, dwr, and rce) represent more than a single flow. Two additional extrusions (units d81 and d61) are off-edifice domes (Fig. 15) of similar trachydacite, centered 3.5 and 4.5 km, respectively, northeast of the summit of Mammoth Mountain. One unit is a pumice-fall deposit (unit rfp) as thick as 5 m, preserved only north and east of the edifice.

The 24 effusive units range in SiO2 content from 63% to 71%, and all are mildly alkaline, widely glassy, and generally flow foliated. Flow layers range in thickness from meters to laminae, are commonly folded or sheared, and are widely vuggy along foliation planes. Many layers or lenses and extensive exterior domains consist of gray to black vitrophyre, dense to vesicular. Many other layers are felsitic, variously cream-white or gray, often with variegated gray mottling, but widely oxidized pink, reddish-brown, or dark brown; most felsite is only partly devitrified, and only a few deeply eroded flow interiors approach full devitrification. The original vesicular carapace facies has rarely survived glacial scour, except on a few ridge crests like Dragons Back (Fig. 12), where lava blocks retain nearly primary scoriaceous rinds. Outcrops are typically ice-scoured ribs and ledges that form slabs or are block jointed.

Of the 24 effusive units, 19 are phenocryst-rich trachydacites, but 5 lavas with 70%–71% SiO2 and fewer phenocrysts (units rce, rmf, rrc, rsq, and rss) are designated alkalic rhyodacites. A pumice-fall deposit (unit rfp; Fig. 16), compositionally similar to edifice lavas of unit rce, is preserved at 5 proximal sites northeast of the glaciated edifice and at a site in Long Valley ∼19 km east of its vent. Key features of all 25 units are summarized in Table 1.

All 25 Mammoth Mountain units contain Fe-Ti oxides, biotite, sanidine, and plagioclase, the last being the dominant phenocryst phase in every sample. Hornblende is the dominant mafic phenocryst in units dbp, ddl, ddu, dom, and dtl (all east of the summit), as well as in unit drc and off-edifice unit d61. However, hornblende is sparse or absent in units dgr, dnh, dsk, and dwr, and absent in all the rhyodacites (units rce, rfp, rmf, rrc, rsq, and rss). In the remaining trachydacite units (dfl, dlp, dml, dms, dnk, dsd, dsu, and d81), biotite and hornblende are similar in abundance, and either may exceed the other in particular samples. Although subordinate to other mafic phases, pyroxene microphenocrysts (and sparse prisms 0.5–2 mm long) are common (∼0.5%–1%) in units dfl, dgr, dlp, dml, dms, dnh, drc, dsd, dsu, dtl, d61, and d81. Pyroxenes appear to be absent in units dbp, dnk, dom, rce, and rfp, and they have been observed in merely trace amounts in the remaining eight silicic units at Mammoth Mountain. Most pyroxenes occur in clusters or in clots with other phases, suggesting entrainment from crystallizing rinds at margins of conduits or magma pods. Orthopyroxene is more abundant than cpx, but both are present in some units. Fine-grained mafic blebs, typically 2–5 mm across, were observed in thin sections of 14 of the 25 units; they are especially common in units dml, dnh, and dtl. Groundmasses of virtually all ∼100 Mammoth Mountain silicic samples sectioned are at least partly glassy and charged with microlites. Petrographic databased on thin-section examination are assembled in Table 1.

All of Mammoth Mountain was ice covered and thus either surficially scoured or deeply excavated during late Pleistocene glaciation. Most of the silicic lava flows and domes that make up the edifice were substantially reduced in size, and linkage of lava remnants on the flanks to higher source domes has required chemical, petrographic, and paleomagnetic correlation. With the exception of unit rfp, any pyroclastic fall and flow deposits that accompanied extrusion of the 24 silicic lava units exposed at Mammoth Mountain have been eroded away or concealed by younger units.

Eruptive Sequence

The 25 silicic units were erupted in several episodes between 100 ka and 50 ka, an ∼50 k.y. interval.

  1. Dome 2781 (unit d81; 99 ± 8 ka; 69% SiO2) is a gently sloping edifice ∼2 km wide at the northeast toe of subsequent Mammoth Mountain. It hosts the postglacial fissure (Fig. 15) called the Earthquake fault (Benioff and Gutenberg, 1939). In a well at Horseshoe Lake (600 m from the south toe of Mammoth Mountain), an undated 15-m-thick lava flow of Mammoth Mountain lithology and composition (68% SiO2) directly underlies flows of unit amp (ca. 97 ka), but is not exposed.

  2. There is no evidence for further extrusions until growth of the second off-edifice Dome 2861 (unit d61; 87 ± 2 ka; 67% SiO2) adjacent to the first. Although each is ∼2 km across, Dome 2861 contrasts with the sprawling earlier dome flow in being steep sided and >300 m high. Its piston-like extrusion lifted a 400-m-wide slab of older basalts (units bar and mcl) still partly preserved on its summit. Like all trachydacites of Mammoth Mountain proper, the two domes are feldspar rich (10%–15%) and have conspicuous hornblende and biotite phenocrysts; biotite exceeds hornblende in Dome 2781, and hornblende exceeds biotite in Dome 2861; both carry small amounts of pyroxene.

  3. About the same time as emplacement of Dome 2861, the earliest units exposed on the Mammoth Mountain edifice erupted. These are the large South Summit Dome (unit dsd; 87 ± 6 ka), which may be the most voluminous unit on Mammoth Mountain (Fig. 17), and the locally exposed rhyodacite of Reds Creek (unit rrc; 83 ± 1 ka; Fig. 15), which underlies the southwest toe of South Summit Dome. As well as being among the earliest, unit rrc is chemically the most evolved product of the edifice (71% SiO2). Like other rhyodacites on Mammoth Mountain, it lacks hornblende and carries about half as much feldspar as the trachydacites. South Summit Dome (67.5%–68.4% SiO2) makes up most of the south slope of the edifice, where it has >450 m of relief (Fig. 17). Most of the unit is a massive dome, but eroded remnants to its north and colluvium-mantled ledges down its southeast extremity appear to be derivative flow lobes. Much of its summit ridge, northeast scarp, and Reds Creek margin are hydrothermally altered.

    During the early interval represented by the units just described, rhyodacite lava flows of unit rce (80 ± 1 ka) and rhyodacite pumice-fall unit rfp (80 ± 8 ka) also erupted; both have 70.2%–70.9% SiO2. Unit rce forms the broad northeast apron of the edifice (Fig. 15), its flows banking against both the Canyon Lodge scoria cone (mcl) and Dome 2781 (d81). Lithologically similar to the other rhyodacite units, unit rce is distinguishable chemically by its consistently lower Fe, Ti, Mg, Ca, P, and Sr contents and higher Nb, Zr, Mn, K, and Na (Supplemental File 1 [see footnote 1]). The vent for unit rce is buried by Solitude Dome (unit rss). Stratified pumice-fall deposits of unit rfp are exposed in roadcuts near the northeast toe of the Mammoth Mountain edifice. The most complete section (UTM 2405/69135), 5.6 m thick, overlies unit d81 and consists of 3 conformable fall sequences (Fig. 16) distinguished by independent grading and separated by a 20-cm-thick pyroclastic-flow deposit and a 14-cm-thick paleosol. Among lava flows exposed on Mammoth Mountain, only unit rce is compositionally similar to rfp pumice clasts.

  4. There followed extrusion of a relatively low silica (63%–64% SiO2), pyroxene-bearing trachydacite (unit dtl; 76 ± 1 ka), now exposed at the east base of the edifice, extending from beneath Dragons Back to Old Mammoth Road and forming the upper falls along Mammoth Creek. The flow carries abundant Fe-Ti oxides, pyroxenes, biotite, and hornblende, 15%–20% feldspars, and conspicuous multiphase crystal clots and mafic enclaves.

  5. An episode of higher silica trachydacite extrusions (66%–70% SiO2) ensued, producing the western flow-dome complex (unit dwr, 73 ± 2 ka), the Old Mammoth coulee at the Bluff (unit dom, 73 ± 1 ka), and Skyline Dome on the crest of the mountain (unit dsk, 71 ± 1.5 ka). The first (dwr) extruded at Scottys Dome on the crest and sent the White Bark Ridge coulee 1 km northward and a stairstep set of 4 flows as far as 2 km southward; representing the most compound unit at Mammoth Mountain, the subunits are compositionally indistinguishable but yield 2 distinct sets of paleomagnetic directions (discussed in the following). The flows banked against Cretaceous granite of Long Valley caldera topographic wall, thereby enclosing the Reds Lake depression. The Old Mammoth coulee (dom) extends east from beneath the Dragons Back flows (units ddu and ddl) of Panorama Dome (Figs. 12 and 15) to form a 1-km-long till-mantled bench (newly covered with houses) that terminates at the 90-m-high scarp called the Bluff, overlooking Mammoth Meadow. Skyline Dome (dsk), on the crest between Summit Dome and Scottys Dome (Figs. 14 and 15), exposes a 200-m-high north face; its upper half is hydrothermally altered.

  6. Another silicic trachydacite (66%–70% SiO2) episode next produced two coulees on the north flank, units dml (67 ± 1 ka) and dms (68 ± 1 ka). The McCoy Station coulee (unit dms) emerges from beneath younger lava flows of the Face Lift planèze floors the plateau at McCoy Station, and continues 500 m farther northward. The Main Lodge coulee (unit dml) emerges from beneath unit dms and extends 1 km downslope to its eroded terminus at the north toe of the edifice, just east of the main ski area facilities. Although they are lithologically similar, unit dms has greater SiO2 content than dml and contains fewer feldspar and hornblende phenocrysts. The age of the small dome, unit dgr (68.8% SiO2; Fig. 15), just west of Lincoln Peak, is poorly constrained but likely to have erupted during this episode or the next; it appears to be bracketed between units rce (80 ± 1 ka) and dlp (64 ± 7 ka). Poorly exposed unit dnh (68% SiO2; 65 ± 1 ka) consists of two lava ledges that crop out low on the southeast slope of the edifice, where they are overlain by the undated Bottomless Pit coulee (unit dbp), which is pervasively oxidized and forms much of the steep glaciated scarp above Twin Lakes (Fig. 17). The scarp rim is incised by a set of erosional clefts, one of which remains roofed, thus forming a steep tunnel (inspiring the name Bottomless Pit). Because unit dbp is bracketed by units dnh (65 ± 1 ka) and ddl (58 ± 2 ka), it probably erupted in this episode or the next.

  7. After a respite of a few thousand years, a short interval of renewed activity yielded a variety of silicic lavas, all north of the summit: the Lincoln Peak dome (unit dlp, 64 ± 7 ka); pyroxene-free North Knob dome (unit dnk; 60 ± 1 ka); low-silica (63%–65% SiO2) pyroxene-bearing flows of units drc and dfl (61 ± 3 ka); the summit dome (unit dsu; 61 ± 2 ka); and two alkalic rhyodacite lava flows, units rmf (61 ± 1 ka) and rsq (63.7 ± 4 ka). Lincoln Peak (unit dlp) is a 700 × 500 m dome (Fig. 3), glacially shaved into northeast elongation, that overlies rhyodacite unit rce (80 ± 1 ka). The steep dome (67.7%–68.5% SiO2) has 160 m of relief on its north face and 220 m on its east face. There is no evidence for a postglacial debris avalanche from Lincoln Peak, as suggested by Bailey (1989). North Knob dome (unit dnk; Fig. 14) is 700 m across, and its 200-m-high north face is the site of Chair 1 (the pioneering chair lift built by MMSA in 1955). The dome (66.5% SiO2) is severely eroded glacially, and its southeast slope is hydrothermally altered.

    The two rhyodacites (units rmf and rsq) erupted from separate vents (Fig. 15), but unit rsq issued at the same site where extensive rhyodacite unit rce had erupted ∼15 k.y. earlier and where unit rss subsequently built Solitude Dome (ca. 50 ka). All of the rhyodacites are compositionally distinguishable, thus representing separately evolved magma batches.

    Unit rmf is a single rhyodacite coulee, 800 m long and 200–350 m wide, that forms the dip slope surface of the planèze called Face Lift. The flow is strongly flow foliated and locally spherulitic or lithophysal. Midway down the flow, the long-lived Mammoth Mountain fumarole (Sorey et al., 1993, 1998) issues wispily between joint blocks and feeds fluid to a strip of active acid alteration that extends 200 m downslope. The upper end of the rmf lava flow was severely acid altered in the Pleistocene, facilitating deep glacial excavation of its vent area in what is now a 150-m-wide trough called the Chasm, which separates the planèze from the summit ridge of Mammoth Mountain (Fig. 18). Unit rmf differs from roughly contemporaneous unit rsq in having lower Fe, Mg, and Ca contents and marginally higher K. Unit rsq vented at the later site of Solitude Dome (unit rss) and flowed 2 km eastward along the south toe of Lincoln Peak.

    Unit dfl on the north slope of Mammoth Mountain (Fig. 15) consists of glaciated remnants of several compositionally related, low-silica (63%–65% SiO2) trachydacite lava flows that variously overlie or bank against units dml and dms and underlie unit rmf. Relations among these four units are consistent with their radioisotopic ages. A 50-m-high cliff of unit dfl (61 ± 3 ka) also crops out on the headwall of the trough called the Chasm (Figs. 14 and 18), where it directly underlies summit unit dsu (68%–70% SiO2), which yields an indistinguishable age (61 ± 2 ka). The hydrothermally altered and deeply excavated trough area had been the extrusive site of several units in addition to dfl. Downslope north of the Chasm, eroded dfl lava flows remain as thick as 80 m along the east side of the Face Lift planèze and 50 m along its west side. On the opposite (south) side of Mammoth Mountain, unit drc is a single lava flow compositionally and paleomagnetically indistinguishable from unit dfl, and it likewise underlies the summit-forming unit dsu. Although unit dsu is an areally small dome flow (68.0%–69.6% SiO2), its north face has 120 m of relief (Figs. 3 and 14) and exposes brecciated internal shear zones as thick as 10 m; its upper half is acid altered and currently hosts a visitor center and café at the gondola terminus.

  8. Among the youngest extrusions on Mammoth Mountain are a superimposed pair of coulees (units ddl and ddu), both phenocryst-rich trachydacites, that form the narrow ridge called Dragons Back (a former glacial cleaver; Figs. 3, 15, and 17). The coulees are each as thick as 160 m, contain 66%–67% SiO2, have more hornblende than biotite, and overlie units dsd, dtl, and dbp. The lower coulee (ddl) displays a spectacularly draping contact atop unit dbp on the cliff above Twin Lakes (Fig. 17). Glacial excavation of the Twin Lakes basin and its outlet to Mammoth Creek separated the Dragons Back segment of units ddl and ddu from their eastward extensions, which survive as till-mantled Panorama Dome and roadcuts on Lake Mary Road and Old Mammoth Road. Unit ddl gave a 40Ar/39Ar plateau age of 58 ± 2 ka, and, because the conformably overlying 1-km-long upper coulee (unit ddu) is compositionally virtually identical, it is presumed to be of similar age. Both are inferred to have vented along the site of the glacially excavated trough (the Chasm; Fig. 18) just north of Mammoth Mountain’s summit.

  9. Solitude Dome (Figs. 3, 14, and 15) is capped by rhyodacite unit rss, which grew atop the vent for older units rce (80 ± 1 ka) and rsq (63.7 ± 4 ka). Dated as 50 ± 3 ka, the dome is probably as young as or younger than the Dragons Back flows. Each of the three rhyodacite lavas that erupted here, as well as units rmf and rrc (which vented elsewhere), is chemically distinctive; each thus represents a uniquely evolved magma batch.

The entire multiepisode growth of Mammoth Mountain lasted ∼50 k.y., and another 50 k.y. has passed since the last eruption of the edifice. A reasonable inference would be that the silicic magmatic interval is over, the small trachydacitic reservoir having by now crystallized.

Vent Distribution

Vents for many of the 25 eruptive units distinguished at Mammoth Mountain are obscure or have been deeply eroded glacially. Despite local incision, no dikes are exposed. Dome 2781 (unit d81), Dome 2861 (unit d61), Skyline Dome (unit dsk), North Knob (unit dnk), Solitude Dome (unit rss), Gold Rush Express Dome (unit dgr), and Lincoln Peak (unit dlp) are all lava domes that emerged from subjacent conduits, and Scottys Dome likewise covers the vent for the White Bark Ridge complex (unit dwr). In common with Skyline Dome, the crestal regions of South Summit Dome (unit dsd) and Summit Dome (unit dsu) are extensively acid altered, indicating fluid flux from directly below, also consistent with subjacent conduits. The apron of rhyodacite lava flows (unit rce) that covers much of the northeast slope of the mountain issued from a vent that was later covered by Solitude Dome (unit rss), which also covered the source of unit rsq, intermediate in age.

The glacially excavated trough (Fig. 18) that separates the summit ridge of Mammoth Mountain from both Solitude Dome and the north-dipping planèze called Face Lift is evidently where several of the remaining lavas originated. The upper end of the planèze and the narrow north-trending rib that closes the east end of the trough are both severely altered, as are parts of the colluvial fill that floors the trough. The steep northeast face of the summit ridge was the principal glacial headwall, away from which ice flowed east, north, and northwest during the late Pleistocene. Glacial action excavated a trough ∼200 m deep and ∼1.5 km long, removing as much as 20 × 106 m3 of lava, some large fraction of which had been altered hydrothermally or fumarolically. From the distribution of orphan lava flows that extend north and southeast from the trough, it is inferred that the excavated region was the central hub, a multivent complex for units dbp, ddl, ddu, dml, dms, dfl, dnh, drc, and rmf. In addition, the extrusive domes dsd, dsu, dsk, and rss are clustered close to the trough, and parts of each are hydrothermally altered.

Trachydacite units dtl and dom are exposed only distally (Figs. 12 and 15), extending eastward from beneath the Dragons Back coulees. Although they are less certainly linked to the excavated near-summit vent complex, they are likely to have originated there, because both are younger than the massive South Summit Dome (unit dsd), which precludes outflow along alternatively plausible azimuths. Unit rrc is one of the oldest on the edifice, and, because the flow is exposed only locally beneath the toe of South Summit Dome (Fig. 15), its vent is concealed and thus unknown.

Postglacial Phreatic Activity and Fumarolic Discharge

Four small postglacial pits interpreted as phreatic craters were mapped by Huber and Rinehart (1965a, 1967) at elevations near 2800 m just north of Lincoln Peak. Simplified to three pits, they were reproduced on the map of Bailey (1989), but were never described. Although the sites have been partly bulldozed for ski runs, our repeated scrutiny of undisturbed areas around and between them has turned up no evidence for craters, ejected diamict, juvenile clasts, or thermal or fluid alteration. Undisturbed forested ground within and adjacent to the sites formerly depicted as craters is covered by combinations of thin colluvium, patchy ground moraine, bedrock and regolith of unit rce, and the 1350 CE Inyo pumice-fall deposit (with or without a thin eolian sand layer beneath it). We conclude that explosion pits never existed there, and we note that the MMSA repeatedly excavates, modifies, and replaces numerous shallow water-holding pits for run-off control and for producing artificial snow.

A contiguous pair of authentic phreatic craters beneath Chair 11 was mapped by Huber and Rinehart (1965a) and Bailey (1989), ∼300 m southwest of the Main Lodge (at UTM 200/685). Although bulldozed for ski runs and construction of a reservoir, parts of their phreatic (nonjuvenile) apron deposits are intact; they are 1–10 m thick, fines rich, and thermally discolored. Phreatic diamict directly underlies 1350 CE Inyo pumice-fall deposits and directly overlies a thin organic layer containing charcoal dated as 930 ± 20 14C yr B.P. (before present, present being 1950 CE). This yields a calibrated age of 850 ± 60 cal yr B.P. (Fig. 19), or ca. 1100 CE. A stump buried in the phreatic deposit gave radiocarbon ages in the range 659–737 cal yr B.P. (Sorey et al., 1998). Despite uncertainties in age and calibration, stratigraphic relations are unequivocal that the phreatic activity here took place before the magmatic Inyo eruptions of 1350 CE, whereas the phreatic activity at Inyo Craters took place after them.

Present-day fumarolic discharge on and near Mammoth Mountain is weak, and sites are few but widely scattered. Diffuse emission of water vapor carrying CO2, He, and H2S is detectable along the upper gorge of Dry Creek (UTM 210/670), along a small fault at the south toe of the edifice (UTM 211/652), and more vigorously from cracks in bedrock of unit rmf at elevation 3030 m on the Face Lift planèze (UTM 209/673) ∼800 m north of the summit (Sorey et al., 1998). Several fumaroles along the summit ridge of Mammoth Mountain were reported to have been active in the 1950s (Huber and Rinehart, 1967), and patches of early snowmelt along that ridge (over acid-altered areas of units dsd, dsu, dsk, and dwr) are now observed annually and attributed by MMSA personnel to warm ground. Off the edifice, ∼1 km east of Twin Lakes, the persistent odor of H2S is detectable (without visible emission) along a zone of altered ground that extends ∼500 m northwest from Mammoth Rock, coinciding with a Mesozoic fault.

Cryptic CO2 Discharge

Approximately 10 sites around the base of the Mammoth Mountain edifice (Fig. 4) have, since 1990, developed soil-gas CO2 concentrations great enough to kill trees (Farrar et al., 1995; Sorey et al., 1998; Cook et al., 2001; Evans et al., 2002). None of these has produced craters, visible orifices, or sulfur-bearing discharge, but two of the tree-kill areas overlap the older Chair 11 phreatic craters and the fumarolic area at the south toe of the edifice.

The tree-kill sites are distributed along a half-ring around the west half of the edifice, extending from its south toe to its northwest toe (150°–350°). This coincides with the arcuate scallop in the basement-rock cirque wall against which Mammoth Mountain was built. The cross section in Figure 20 depicts how permeable rubble along the inward-sloping contact could distribute centrally ascending CO2 to numerous sites along the arcuate surface trace of that contact. It is inferred that the isotopically magmatic CO2 degasses from basaltic magma lodged in the middle crust, as suggested by persistent long-period earthquakes at depths of 10–18 km (Shelly and Hill, 2011). Sorey et al. (1998) discussed processes by which the small amount of H2S present in some of the fumarolic gases might be removed from the CO2-rich gas that reaches the tree-kill periphery.

Postglacial Diamicts

In addition to the late Holocene ejecta sheet surrounding the twin phreatic craters near Chair 11, various thin colluvial diamicts have been exposed in numerous MMSA water pits, typically 1–5 m deep and 5–50 m wide, that are widely distributed across the northern base of Mammoth Mountain between Reds Lake and Chair 4 (UTM 228/683). Diamicts in most such exposures are tan or orange-brown, 0.5–3 m thick, vaguely stratified in layers 5–50 cm thick defined by slight grain-size variations, and might at first glance be taken for thin nonwelded pyroclastic-flow deposits. They overlie Mammoth Mountain lavas or discontinuous sheets of Pleistocene ground moraine, and they are generally overlain by 1350 CE Inyo pumice-fall deposits, either directly or with an intervening sheet of eolian sand. Virtually all clasts are Mammoth Mountain dacite and rhyodacite, glassy or partly so, typically dense and nonvesicular, subrounded to subangular, and mostly 0.5–4 cm across, but rarely and locally, 10 cm or larger. Neither prismatically jointed clasts nor thermal and/or oxidative effects of clasts on the sandy-silty matrix were observed.

In some pits, the diamicts also contain granules and lapilli of biotite-feldspar pumice as large as 4 cm. Pumice clasts analyzed are strongly hydrated (loss on ignition, LOI > 10 wt%), depleted in silica and alkalies, and enriched in Al2O3 (>23 wt%), suggesting long residence in wet colluvium. The pumices were probably reworked from fall unit rfp and soaked in groundwater for 80 k.y. In 2004, two former pits (near UTM 211/688 and 226/684) on flats at the north toe of Mammoth Mountain exposed ashy diamicts, 0.5–1.5 m thick, rich in such hydrated and/or altered pumice lapilli; these may have been remnants of nonwelded pyroclastic-flow deposits related to units rfp or rce or colluvially remobilized equivalents.

We interpret the diamicts as thin postglacial colluvium that by creep, slumping, or sheetwash has filled minor depressions on low-relief surfaces. In addition to near the break in slope at the toe of the edifice, similar diamict is exposed in water pits on a 2930 m plateau, ∼300 m southeast of McCoy Station, where sparse clasts of dense variably rounded dacite and rhyodacite are as large as 9 cm but mostly 0.5–4 cm across. Here, as at several lower sites, a tan or orange-brown layer of massive or crudely lenticular crystal-vitric sand, 30–110 cm thick, is between the diamict and the capping 1350 CE Inyo pumice falls. From site to site, its median grain size ranges from 0.3 to 0.7 mm, fine to medium sand. It contains granules and sparse pebbles of dacite, granite, and biotite-feldspar pumice as large as 4 cm. We interpret the sand to have resulted from eolian reworking of the postglacial diamict.

Paleomagnetic directions were determined at ∼135 sites, complementing data for 25 sites previously reported in Mankinen et al. (1986). All lava units on Mammoth Mountain and 29 of the 37 contemporaneous peripheral units (basalt to dacite) were sampled, in addition to 3 nearby Pliocene mafic units and 4 Long Valley rhyolites (150–100 ka). The data were obtained with three goals: (1) to verify (or challenge) correlations among scattered outcrops based on petrography, composition, and age; (2) to search for dated or datable geomagnetic field anomalies; and (3) to determine if any eruptive units were extruded close enough in time to have similar paleomagnetic directions or to record systematic changes in field directions.

Paleomagnetic samples were collected, processed, and interpreted using standard methods described by McElhinny (1973). Field samples were taken with a 2.5 cm coring drill and oriented with a sun compass. At each site, typically eight 10-cm-long cores were taken. In the Menlo Park laboratory, 2.5-cm-long core specimens were measured, initially using a manual magnetometer and later a fully automated cryogenic magnetometer, subjecting cores from most sites to alternating field (AF) demagnetization to remove secondary components of magnetization. An isothermal component from nearby lightning strikes, though rare, was the main source of secondary magnetization in these young volcanic rocks. Specimens from a few sites were subjected to progressive thermal demagnetization to isolate low-temperature partial thermoremanent magnetization components acquired subsequent to (limited) deformation of their outcrops. The characteristic direction of remanent magnetization for each site was calculated using Fisher statistics on data from either a blanket level of AF treatment, line fits of several data points on vector component diagrams, planes fit on equal-area diagrams, or mixtures of lines and planes data. Mean site directions are illustrated in Figures 21 and 22, and all paleomagnetic data are presented in Supplemental File 2 (see footnote 2).

Paleomagnetism of Peripheral Units

The paleomagnetic data provide the following inferences and conclusions for the basalt to dacite eruptive units peripheral to Mammoth Mountain (see Fig. 21).

  1. The basalt of the Buttresses (Tbtb) and the two-flow remnant knob west of Crater Creek (Tmcw) have identical directions of reverse polarity (Supplemental File 2 [see footnote 2]), supporting the middle Pliocene radioisotopic age (3754 ± 7 ka) of Tbtb and verifying that these units long predate the Mammoth magmatic system. The several lava flows of the Tbtb stack (as thick as 470 m) are bimodal in SiO2 content (∼47% and ∼50%), but the two compositions have identical directions. The 2 Tmcw flows, 2 km downcanyon and mineralogically similar to Tbtb, although more evolved (52.2%–52.6% SiO2), have virtually the same direction as Tbtb and may thus have been a related intracanyon package of similar age. That the base of the Tbtb stack overlies granite at the present-day canyon floor shows that this reach of the Middle Fork canyon was already cut to its present depth by the middle Pliocene.

  2. Unit bmn and directly overlying unit bar erupted sequentially (ca. 88 ka) at the site of the Minaret Summit scoria cone, and provide sets of overlapping directions (for 3 and 6 sites, respectively), the means of which differ by only 1°–2° (Fig. 21C). Sites on the proximal apron of unit bar have directions indistinguishable from that in the caldera’s north moat, where 2 bar flows extend as far as 23 km from the vent cone. A late-erupted flow, compositional variant bar′ (averaging ∼0.70% higher SiO2), has a direction only 2°–3° steeper and slightly more east directed.

  3. Unit amp, which erupted (ca. 97 ka) on the Sierran crest where Mammoth Mountain was later constructed, shed opposing aprons of phenocryst-poor lava flows (57.2%–61.6% SiO2) down both sides of the divide. Unusually shallow paleomagnetic directions for flows in Old Mammoth and along the Middle Fork, though 7 km apart, are indistinguishable in declination and differ in inclination by only a few degrees. Unit drf, which erupted ca. 98 ka and directly underlies amp, yields a similarly shallow direction (Fig. 21C). The anomalous directions may represent a widely detected excursion thought to be a global feature of about this age (Singer et al., 2013; Laj and Channell, 2007).

  4. Units bmc and bfh have similar ordinary west-directed paleomagnetic declinations (Fig. 21C; two sites each), supporting correlation based on composition, age (both ca. 92.5 ka), and their unusually large feldspars, despite their 18 km separation. A lava remnant near Crystal Lake has the same direction as the bmc agglutinate cone that caps Mammoth Crest, showing that cone-derived lavas (later glacially excavated) had flowed through the Lakes Basin en route via the caldera’s south moat to the Fish Hatchery and Hot Creek.

  5. Units mor and mnd have indistinguishable mean directions (Fig. 21C; 3 sites each), sustaining the inference from compositional overlap and field relations that they are elements of an apron of flows emplaced within a brief time interval ca. 66 ka (Mahood et al., 2010).

  6. Paleomagnetic directions are virtually identical for unit mdn (16 ± 2 ka; 53.0%–53.7% SiO2) and the undated basaltic trachyandesite of June Lake (unit mjl; 53.5%–54.1% SiO2), the vents of which are close to range-front faults and 10 km north-south of each other. Both are fountain-fed lavas and agglutinate rich in phenocrysts of cpx, olivine, and plagioclase. Although clearly of late Pleistocene age, unit mjl has not been dated precisely, probably due in part to xenocrystic contaminants. Units mdn and mjl are both overlain by MIS 2 glacial deposits, and their compositional equivalence and unusually steep paleomagnetic directions (Fig. 21C) are permissive of nearly identical (late Last Glacial Maximum) eruptive ages.

  7. Unit bcd lava apron near the site of Casa Diablo Hot Springs and spatter cone unit bed are chemically and mineralogically identical and yield similar 40Ar/39Ar dates (125 ± 2 ka and 121 ± 3 ka, respectively). The 7 km between them is covered by younger eruptive units and till. Two sites on the apron gave northeast directions, but dense agglutinate on the cone rim gave a steeper north direction that differs from that of bcd by ∼24° in declination and ∼6° in inclination (Fig. 21B). Because the correlation is otherwise strong and no alternative vent for unit bcd is exposed, we speculate that the discrepancy may be attributable either to vent-proximal distortion of the local magnetic field during protracted spatter accumulation or to late creep of the agglutinate pile.

  8. The five discrete trachydacite domes and flows (unit dnw) that form a chain across the caldera’s northwest moat share a shallow north paleomagnetic direction (Fig. 21C). Despite giving 40Ar/39Ar dates that range between 42 ka and 27 ka, and contamination-related compositional variations (60.4%–67.3% SiO2), their similar remanent directions suggest a brief eruptive sequence. Rapid sequential eruption would also be consistent with progressively greater contamination (with both mafic enclaves and granitoid xenocrysts) southeast along the chain. Such an interpretation, however, encounters the complexity that the more contaminated members of the chain yield younger 40Ar/39Ar dates. Moreover, the southeastern coulee (27 ± 1 ka) directly overlies mafic unit mcv, which yielded 40Ar/39Ar dates of 31 ± 2 ka and 33.8 ± 1 ka in independent laboratories. The paleomagnetic and radioisotopic data are in conflict, suggesting unresolved problems in dating the five members of unit dnw and perhaps unit mcv.

  9. Three relatively phenocryst-poor silicic lava flows (units aml, ddc, and drf), peripheral to and contemporaneous with Mammoth Mountain, are related neither to the edifice nor to each other. They have different paleomagnetic directions and mutually distinguishable mineralogy and composition. The dacite of Rainbow Falls (drf; 98 ± 1 ka) has a shallow inclination (31°) that could represent a field excursion (see item 3).

  10. Units mss and mdp, superimposed along the floor of the Middle Fork, have different paleomagnetic directions (Figs. 21B, 21C), consistent with an age difference of ∼40 k.y. A dike-bearing scoria cone incised by the river was the vent for mss lava flows. It had long been presumed that the Devils Postpile flow (mdp) erupted there, but no evidence has been found for its vent, and our work shows it to be unrelated to unit mss.

  11. Three phenocryst-poor lava flows (unit aic; 131 ± 1 ka) exposed on the inner wall of South Inyo Crater have virtually identical directions (Mankinen et al., 1986), sustaining the inference based on chemical similarity (57.5%–59.1% SiO2) that they represent a brief eruptive sequence. An undated spatter cone (unit mdm) ∼1 km north is shown to be unrelated by both its composition (55.6%–56.0% SiO2) and a steep north paleomagnetic direction that differs from the aic lavas by >10° in inclination (I) and >20° in declination (D) (Fig. 21B). The vent for unit aic is buried and its location is unknown.

  12. The basalt of Horseshoe Lake (unit bhl) has a 40Ar/39Ar date of 31 ± 1 ka and an unusual paleomagnetic direction, steep and west (I = 62°, D = 318°; Fig. 21C). It may have erupted during the Mono Lake geomagnetic excursion, now considered to have taken place ca. 32 ka (Channell, 2006; Cassata et al., 2008).

  13. Adjacent phenocryst-poor units msc and mkv erupted ca.153–155 ka, and are just north of downtown Mammoth Lakes; they have unusually steep inclinations but differ in declination by as much as 90° (Fig. 21B). Both units may have erupted during a geomagnetic excursion referred to as the Albuquerque by Langereis et al. (1997) and Laj and Channell (2007). The lavas are difficult to distinguish lithologically, but are clearly different chemically (52.3%–52.5% vs. 54.5%–54.9% SiO2, respectively). A lava flow (unit mmc) exposed 4 km east-southeast of the mkv vent is petrographically and chemically indistinguishable from vent agglutinate, but yields a typical normal polarity direction (Fig. 21A) and a 40Ar/39Ar date of 162 ± 2 ka (Mahood et al., 2010). Because the radioisotopic ages estimated for the three units were determined in three different laboratories, their nominal age range (153–162 ka) may not be real. If they represent the same geomagnetic excursion, then their directional differences and individual spreads might reflect rapid motion of the excursional field. For the mkv vent complex, proximal distortion of the local magnetic field during protracted spatter accumulation and late creep of the agglutinate pile may also have contributed to its directional anomalies.

  14. Unit bsr (50.6%–51.7% SiO2), one of the youngest of the south moat lava-flow sequence, has yielded 40Ar/39Ar dates of 99 ± 1 ka and 103.5 ± 1.4 ka and a shallow northwest paleomagnetic direction (Fig. 21C). Along with units amp and drf (see item 3), its unusual direction may reflect eruption during the geomagnetic excursion called post-Blake by Singer et al. (2013). However, the flows of unit bsr directly overlie those of unit bmc-bfh, which has twice been dated as ca. 92.5 ka (Table 2) and has an ordinary normal polarity direction (Fig. 21C). The age conflict remains unexplained, but xenocrystic contamination is known to have made earlier attempts to date unit bsr problematic (Curry, 1971; Bailey et al., 1976; Mahood et al., 2010), and unit bmc-bfh is unusually rich in large complex plagioclase.

  15. Unit mcl, one of the oldest exposed units of the Mammoth system (ca. 175 ka), crops out at its vent (the Canyon Lodge scoria cone) as a ledge near downtown Mammoth Lakes, as a faulted ledge near the geothermal plant, as a slab uplifted on the roof of trachydacite Dome 2861, and as six separate outliers near the north end of the Long Valley resurgent graben (Figs. 6 and 13). Six exposures drilled have only 0.5%–2% plagioclase and sparse olivine, and they cluster tightly in composition (55.1%–55.9% SiO2; 7.0%–7.3% FeO* (total iron calculated as FeO); and 1.9%–2.1% K2O). Although within-site dispersions are small, inclinations (40°–71°) and declinations (30°–88°) both scatter widely. This scatter seems explicable only by deformation, as all sites except the scoria cone are on Long Valley’s resurgent uplift and all are cut by or are near resurgence-related faults that also displace post-mcl units rwm and bcd, which yield ages of 150–125 ka. The site on the undeformed scoria cone, however, itself has an unusual remanent direction (D = 47°; Fig. 21A). The scatter of the other sites should thus be considered as deviations from the unusual cone direction, not from a typical normal polarity direction.

  16. Phenocryst-poor subalkaline basaltic unit bsm (165 ± 2 ka) crops out in 4 isolated windows separated by as much as 7 km, but 7 samples are compositionally indistinguishable and unusual for the Mammoth system (48.4%–48.9% SiO2; 1.94%–1.98% TiO2; 1.22%–1.27% K2O; Supplemental File 1 [see footnote 1]). Despite probable identity, paleomagnetic directions for the outcrops scatter substantially (Fig. 21A), probably because they are within or close to the fault system crossing the resurgent uplift. Units mcl and bsm are among the oldest exposed and thus were subjected to deformation longer than most lavas of the Mammoth system.

  17. Unit mlc (51%–52.9% SiO2; 130 ± 1 ka) is exposed only as a rugged distal flow front marked by tumuli and pressure ridges. That the two sites judiciously selected for drilling give paleomagnetic directions that differ in declination by 25° probably reflects deformation of the carapace during late flowage or inflation. Only the site we judge more stable (less likely to have been deformed) is plotted in Figure 21B.

  18. Three additional mafic peripheral units have directions that are independent of other known units and, while distinctive, are unlikely to be excursional (Figs. 21A, 21C): south moat lava-flow apron of unit bsc (172 ± 2 ka); west moat spatter cone unit bcf (164 ± 2 ka); and north moat lava flows of unit mcv (31 ± 2 ka). Directions are also plotted for the four Long Valley rhyolites in the west moat (Fig. 21B), all of which represent ordinary normal polarity.

Paleomagnetism of Mammoth Mountain Units

  1. The five rhyodacite lava units have different paleomagnetic directions (see Figs. 15, 22A, and 22B), confirming their compositional and 40Ar/39Ar age differences. Unit rrc (83 ± 1 ka) has I = 46°, D = 20°. Unit rce (80 ± 1 ka) has I = 68°, D = 327°. Unit rmf (61 ± 1.5 ka) has I = 53°, D = 323°. Unit rsq (60.3 ± 6.5 ka) has I = 54°, D = 11°. Unit rss (Solitude Dome; 50 ± 3 ka) has I = 72.5°, D = 330°. Although units rce and rss yield overlapping directions, they differ in 40Ar/39Ar age estimates by at least 25 k.y.

  2. Four sites on the sprawling dome and multiflow complex, unit dwr, which forms the entire west flank of Mammoth Mountain, yield contrasting pairs of remanent directions (Fig. 22A). The north, south, and west flow lobes are compositionally indistinguishable, and the vent dome is only marginally less evolved (68.0% SiO2 vs. 68.5%–69.1% for the flows; 2.57% FeO* vs. 2.34%–2.44% for the flows). The dome (73 ± 2 ka) and southwest coulee have ordinary normal polarity directions, but the west (79 ± 12 ka) and northwest lobes share a shallower, more east, direction of magnetization. The 40Ar/39Ar dates are permissive of eruption during the Norwegian–Greenland Sea geomagnetic excursion (Langereis et al., 1997; Laj and Channell, 2007). We have discovered little evidence that dispersion of directions for the silicic Mammoth Mountain units might be attributable to chaotic distortions on their original blocky surfaces, probably because glacial erosion has deeply scoured almost all such surfaces.

  3. At least seven eruptive units of varied age and composition have fairly shallow inclinations and north-directed declinations (I = 33°–43°; D = 6°–345°). These include the South Summit Dome (dsd), one of the largest and oldest (87 ± 6 ka) extrusions on the edifice, and North Knob (dnk), one of the youngest (60.4 ± 1.2 ka). Extending eastward off the flank of South Summit Dome, a sequence of four coulees, units dom (73 ± 1 ka), dbp, ddl (58 ± 2 ka), and ddu, yields similar directions and is tightly grouped chemically (66%–67% SiO2), slightly less evolved than the South Summit Dome (68% SiO2). Isolated unit dgr, just west of Lincoln Peak, is also directionally similar to this group but chemically more evolved (69% SiO2); although undated, it is apparently older than unit dlp but younger than units rce and dsd.

  4. Several eruptive units on Mammoth Mountain have fairly steep inclinations and northwest declinations (I = 49°–61°, D = 321°–348°). Among them, low-silica trachydacite (63%–65% SiO2) units drc and dfl (61 ± 3 ka), on opposite (south and north) sides of the edifice, have indistinguishable directions (I = 56°–57°; D = 330°–335°), which supports chemical and mineralogical evidence that they were comagmatic extrusions. At the east base of the edifice, similarly low-silica (63%–64% SiO2) unit dtl is older (76 ± 1 ka) and paleomagnetically different; its direction is less steep and not as west-directed (I = 53°, D = 348°).

  5. Four more units with steep northwest directions (I = 53° –59°, D = 323°–332°) differ from the three just mentioned in having higher silica (66%–71% SiO2): unit dnh on the east slope, the Summit Dome (dsu), and two coulees on the north slope (rmf and dml, the directions of which are similar to that of low-silica unit dfl). Directly beneath units dfl and rmf, the bench-forming coulee at McCoy Station (unit dms; 69% SiO2) likewise yields northwest declinations (314°–333°) but scattered disparate inclinations (40°–60°). From lowermost to uppermost, the stratigraphic sequence stacked on the north slope of the edifice is dml-dms-dfl-rmf-dsu; all five units have fairly similar directions. If their radioisotopic dates, which range from 68 to 61 ka, are actually less precise than their stated uncertainties (±1–3 k.y.), the stack might alternatively represent a compressed eruptive sequence, compositionally varied but closely spaced in time.

  6. Older than that sequence, Skyline Dome (dsk; 71 ± 1.5 ka), also gives steep inclinations (54°–62°) but divergent declinations (348°, 342°, and 17°). The anomalous latter site, on the crest of the mountain, is within an extensive area of hydrothermal alteration.

  7. Lincoln Peak (dlp; 64 ± 7 ka) gave erratic paleomagnetic results (Fig. 22B), probably because most outcrops have been disturbed, possibly rotated during prolonged extrusion or inflation. A massive cliff site midway up the north face provided what may be the most reliable direction: I = 59°, D = 352°, steep and north directed.

  8. The two off-edifice domes, consisting of crystal-rich trachydacite indistinguishable from that of Mammoth Mountain, yield steep north- to northeast-directed declinations (Fig. 22A): Dome 2781 (d81; 99 ± 8 ka), I = 67°, D = 22° and Dome 2861 (d61; 87 ± 2 ka), I = 63°, D = 3°.

Although the many subalkaline rhyolites of Long Valley were produced by a long-lived magma system separate from the system that generated the Mammoth Mountain trachydacites and alkalic rhyodacites, the four youngest Long Valley rhyolites (150–100 ka) were contemporaneous with the Mammoth volcanic field and extruded only a few kilometers from the foot of Mammoth Mountain. Their relationship (if any) to the eruptive history of Mammoth Mountain thus warrants discussion, as does that of the late Holocene Inyo Chain a few kilometers farther north.

Pre-Mammoth Rhyolites (2.2–0.3 Ma)

The Long Valley volcanic field is part of an active regional transtensional zone at the Sierra Nevada–Basin and Range transition (Riley et al., 2012). By ca. 4 Ma, decompressing upper mantle began leaking modest batches of mafic magma to the surface across a broad belt (Moore and Dodge, 1980) that extends from the later site of Long Valley for 40 km southwest into the Sierra Nevada and 30 km northeast into the Adobe Hills (north of Adobe Valley; Fig. 1). Only close to Long Valley, which coincides with a major left-stepping offset in the Sierran range front, were the precaldera mafic products accompanied by numerous eruptions of dacite (3.6–2.6 Ma). Bailey (2004) noted that the mafic and dacitic magmas ceased erupting by ca. 2.6 Ma, not long before the onset of rhyolitic eruptions ca. 2.2 Ma. For ∼2 m.y. thereafter, an increasingly molten deep-crustal environment evidently favored entrapment of mantle-derived basaltic magma, which in turn amplified crustal melting, initiating a prolonged interval of rhyolitic magmatism, including the great caldera-forming eruption of the Bishop Tuff at 767 ka. Not until ca. 230 ka did mafic eruptions resume, and they have been limited to an area west of the structural caldera, the new magmatic focus here called the Mammoth system. Long Valley rhyolites that preceded inception of the Mammoth system were erupted in several successive episodes, as established by Bailey (1989) and Metz and Mahood (1991; and as reviewed in Hildreth, 2004).

Glass Mountain Rhyolites

The two earliest episodes took place at Glass Mountain, a sprawling precaldera complex (2.2–0.79 Ma) of ∼60 overlapping lava flows and domes, accompanied by as many as 17 fall deposits, exclusively high-silica rhyolite (76.6%–77.7% SiO2), at the northeast periphery of Long Valley caldera where the rhyolites are exposed to a thickness >1050 m on the caldera wall (Metz and Bailey, 1993). Compositional data for Glass Mountain lavas plot into two groups: (1) an older sequence (2.2–1.3 Ma) of at least 24 eruptive units, all high-silica rhyolite but chemically varied, probably tapped sporadically from several discrete bodies at different stages of evolution, including some units more enriched than the Bishop Tuff in incompatible trace elements; and (2) a younger sequence (1.2–0.79 Ma) of at least 35 eruptive units, all of them geochemically similar to the more evolved end of the compositionally zoned Bishop Tuff array, and presumably tapped from a common, by then integrated, expanding magma chamber (Metz and Mahood, 1985, 1991). The total magma volume released by the Glass Mountain center may have been 100 ± 20 km3.

Caldera-Forming Eruption

The next episode was the caldera-forming eruption of the Bishop Tuff at 767 ka, which released ∼650 km3 of compositionally and thermally zoned gas-rich rhyolitic magma (Hildreth and Wilson, 2007) in a virtually continuous ∼6-day-long eruption (Wilson and Hildreth, 1997), thereby permitting 2–3 km subsidence of the chamber roof. Approximately one-half the Bishop Tuff volume was emplaced radially as a set of sectorially distributed ignimbrite outflow sheets along with concurrent Plinian and coignimbrite fallout. The other half ponded inside the subsiding caldera, where welded intracaldera ignimbrite is as thick as 1500 m, and was subsequently buried by 500–800 m of postcaldera rhyolite tuffs, lavas, and sedimentary fill (Bailey, 1989). Postcaldera rhyolitic episodes ensued, as described in the following.

Early Postcaldera Rhyolite

What Bailey et al. (1976) termed the Early Rhyolite consists of ∼100 km3 of intracaldera, phenocryst-poor rhyolite (74%–75% SiO2) that erupted during the 100 k.y. interval following caldera collapse. The volume is comparable to that of precaldera Glass Mountain and far greater than the total of subsequent Long Valley rhyolites. Released in numerous separate eruptions from several vents (Bailey, 1989), the Early Rhyolite includes ∼11 exposed lava flows and others intersected by drilling, but it consists predominantly of varied tuffs (mostly nonwelded pyroclastic-flow deposits) that make up about two-thirds of the assemblage by volume. Eight lava flows (but no tuffs) were K-Ar dated (Mankinen et al., 1986), yielding ages from 751 ± 16 ka to 652 ± 14 ka. Several 40Ar/39Ar dates by Simon et al. (2014) are in the same range. The Early Rhyolite extends far beyond its outcrop area, as documented in many wells (Suemnicht and Varga, 1988; Bailey, 1989); the assemblage is at least 622 m thick near its center of outcrop (McConnell et al., 1995), and thicker than 350 m where deeply buried in the caldera’s southeast moat, and 230–537 m thick beneath parts of the west moat.

The compositions of Early Rhyolite lavas are similar in most respects to the last-erupted part of the zoned Bishop Tuff, except that Zr and Ba significantly extend the range of Bishop zoning, and the dominant minerals in the Bishop Tuff, sanidine and quartz, are absent. Doubling of the Ba content (to as much as 1200 ppm) in the nearly aphyric Early Rhyolite may reflect resorption of accumulative sanidine during postcaldera physical and thermal reorganization of the rhyolitic magma reservoir (Hildreth, 2004). Structural resurgence of the west-central part of the caldera has lifted, faulted, and tilted much of the Early Rhyolite (Bailey, 1989; Hildreth, 2004), but not its Lookout Mountain subunit.

North-Central Rhyolite Chain

The earliest of three clusters of postresurgence rhyolites that Bailey (1989) termed “moat rhyolite” is a northwest-trending chain of four extrusions (74%–75% SiO2) that crosses the northeast sector of the resurgent uplift and is therefore not really in the caldera moat or aligned along its ring-fault zone. In contrast to the voluminous Early Rhyolite, all components are phenocryst rich and of small eruptive volume, totaling ∼1 km3. Approximately 100 °C lower in Fe-Ti-oxide temperature than the nearly aphyric Early Rhyolite, the north-central rhyolites are rich in plagioclase, sanidine, quartz, hornblende, and biotite. The extrusions yielded sanidine K-Ar ages of 527 ± 12, 523 ± 11, 505 ± 15, and 481 ± 10 ka (Mankinen et al., 1986); Simon et al. (2014) determined 40Ar/39Ar ages of 570 ± 8 ka for 2 of them. Mafic enclaves (53.3%–54.9% SiO2) occur in lava flows and domes along the chain, but none have been found in any of the still-younger Long Valley rhyolites.

Southeastern Rhyolite Cluster

After an apparent hiatus of at least 120 k.y., another set of rhyolites erupted over an interval as long as ∼75 k.y. and formed a cluster of six units in the caldera’s low southeast moat (see map of Bailey, 1989); two of them arguably extend the trend of the north-central chain just discussed. Two are along the caldera’s southeastern ring-fault zone, but the other four are inboard of it. The extensive (12 km2) Hot Creek flow and three of the small peripheral rhyolite lavas are crystal poor (1%–2% feldspars plus trace biotite, quartz, pyroxene, and Fe-Ti oxides), whereas two lava flows of the cluster are phenocryst-rich hornblende-biotite rhyolites like those in the north-central chain. All six units have 75%–77% SiO2 where fresh, although parts of several were altered by interaction with saline-alkaline lake water and sediment.

The exposed lavas add up to only ∼1.5 km3, the Hot Creek flow being most of it. Five have been dated (Mankinen et al., 1986; Heumann et al., 2002). Sanidine yields ages of 362 ± 8 ka and 333 ± 10 ka for the two crystal-rich units and ages of 329 ± 23 ka and 329 ± 3 ka for two crystal-poor units. For the Hot Creek flow, obsidian gave a K-Ar age of 288 ± 31 ka (Mankinen et al., 1986), and sanidine gave a 40Ar/39Ar age of 333 ± 2 ka. Whatever process promoted reversion to crystal-poor rhyolite ca. 330 ka, it was unique in the post–Early Rhyolite evolution of the Long Valley magma reservoir, because all other Long Valley rhyolites (570–100 ka) are rich in phenocrysts. Thermal rejuvenation by basalt injection is an unlikely explanation for these small-volume low-temperature rhyolites, which lack mafic enclaves, because the crystal-poor units carry only euhedral phenocrysts and lack xenocrysts or partly resorbed relicts of an earlier generation. (Another process, high-silica melt extraction from crystal-rich felsic mush, was favored in Hildreth, 2004.)

Four Rhyolites Contemporaneous with Mammoth Mountain

After another hiatus of ∼150 k.y., a third cluster of moat rhyolites erupted, this time west of the resurgent uplift and no longer phenocryst poor. It consists of three modest lava domes (Deer Mountain, Mammoth Knolls, and Dry Creek Dome) and the extensive West Moat Coulee (8.5 km2; as thick as 574 m where drilling penetrated its cryptic feeder; Benoit, 1984). The coulee represents ∼2 km3 of rhyolite lava but the domes total only ∼1 km3 more. Three of the vents are along the ring-fault zone, whereas that of Deer Mountain apparently is ∼2.5 km outboard of it (Hildreth, 2004). All four have been dated (Mankinen et al., 1986; Heumann et al., 2002; Mahood et al., 2010); the oldest is the West Moat Coulee, for which four ages range from 163 to 147 ka, but the Dry Creek, Deer Mountain, and bilobate Mammoth Knolls domes yield overlapping ages in the range 115–97 ka. All are phenocryst-rich (15%–30%) low-temperature rhyolites, containing plagioclase, sanidine, quartz, biotite, hornblende, Fe-Ti oxides, and trace amounts of zircon, apatite, and allanite (± opx, cpx, and pyrrhotite), but chemically they are of two kinds. Deer Mountain and the coulee have high Ba (700–860 ppm) and only 72%–73% SiO2, whereas the Dry Creek and Mammoth Knolls domes are more evolved, with 76%–77% SiO2 and lower Ba (110–200 ppm). Like all Long Valley rhyolites, all are subalkaline, in contrast to the entire Mammoth Mountain suite.

These four rhyolites extruded at sites only 2–5 km from the base of Mammoth Mountain, and 3 of the 4 erupted in a time interval (ca. 100 ka) shared by the earliest Mammoth Mountain trachydacite. All four are along the edge of, and erupted during the eruptive lifetime (230–8 ka) of, the greater Mammoth volcanic field. Extruded along or near the caldera’s buried ring-fault zone, the four may have tapped marginal domains of the long-crystallizing Long Valley magma reservoir that were reenergized by inception of the contiguous Mammoth magmatic system ca. 230 ka. Deer Mountain rhyolite, which gave an eruptive age of 101 ± 8 ka (Heumann et al., 2002), contains zircons that yield crystallization ages that cluster at 230 ka (Reid et al., 1997). Residual or thermally rejuvenated Long Valley magma is apparently implicated.

Post-Mammoth Inyo Chain

Mammoth Mountain has sometimes been categorized loosely as part of the Inyo Craters volcanic chain, an association we regard inappropriate and misleading. Although roughly on strike with the rhyolitic Inyo alignment (which is directed by the range-front fault system), the basalt to dacite Mammoth array is far broader and much older. The Inyo chain is linear and single-file north to south, whereas the footprint of the Mammoth multivent array is 10 km wide east to west. The Inyo domes extruded along the range-front fault system, which ends beneath the west moat at a left step, where it jumps 15 km eastward and resumes as the Hilton Creek fault at the caldera’s southeastern margin (Figs. 1 and 2). Mammoth Mountain is 5 km outside the caldera’s ring fault, banked against the granitic wall of a Pleistocene cirque, and it does not overlie a range-front fault. We cannot confirm the existence of north-south faults depicted by Bailey (1989) cutting basement rocks 3–10 km south of Mammoth Mountain. Moreover, the Mammoth Mountain dome complex ended its period of eruptive activity (100–50 ka) long before inception of Inyo volcanism in the Holocene.

The 10-km-long Inyo chain (Bailey, 1989; Hildreth, 2004) consists of seven rhyolitic lava flows and domes, several phreatic craters, and a composite apron of pyroclastic fall and flow deposits (Miller, 1985; Sampson and Cameron, 1987; Nawotniak and Bursik, 2010). The oldest unit is North Deadman dome (∼0.04 km3), undated but probably mid-Holocene, followed by Wilson Butte (∼0.05 km3 of lava), which erupted ca. 1.3 ka. Both domes are aphyric glassy rhyolite. Wilson Butte (76.6% SiO2) is similar compositionally and petrographically to the Mono Craters domes (Kelleher and Cameron, 1990) and should be called a Mono dome were it not for its position on the north-south linear trend of the Inyo chain rather than the arcuate trend (Fig. 1) of the Mono chain. North Deadman dome (74.7% SiO2) is compositionally intermediate between Wilson Butte and the crystal-poor lower silica rhyolite (Sampson and Cameron, 1987) that dominated the youngest Inyo eruptive episode in 1350 CE. Two phenocryst-poor minidomes (each <0.001 km3) just north and south of the large Glass Creek flow erupted after Wilson Butte but prior to the 1350 CE Inyo eruption, the products of which they compositionally resemble (Sampson and Cameron, 1987).

Injection of the Inyo dike (Eichelberger et al., 1985) in late summer of 1350 CE (Millar et al., 2006) led to sequential eruption of the Deadman Creek, Obsidian, and Glass Creek flows, each preceded by substantial pyroclastic outbursts (Miller, 1985), the last of which was followed by phreatic eruptions at nearby Inyo Craters (Mastin, 1991). The total magma volume erupted during the 1350 CE episode was estimated by Miller (1985) to be 0.4 km3 as lava, 0.17 km3 as fallout, and >0.05 km3 as pyroclastic density currents.

Compositionally, the mid-fourteenth century Inyo eruption was extraordinarily complex (Sampson and Cameron, 1987; Vogel et al., 1989). As summarized in Hildreth (2004), at least four discrete magmas were confluent during the eruption: (1) crystal-poor rhyolite like that of Mono Craters; (2) crystal-rich Long Valley rhyolite like that of the 100 ka west moat domes; (3) trachydacite like that of the 42–27 ka chain of flows and domes (unit dnw) on the northwest margin of the caldera; and (4) andesitic magmatic inclusions (∼60% SiO2) found quenched in two of the 1350 CE flows (Varga et al., 1990). The compositional and petrographic similarity of the crystal-rich component to the nearby Long Valley rhyolite of Deer Mountain was pointed out by Sampson and Cameron (1987). Moreover, Reid et al. (1997) identified, in both 1350 CE crystal-rich Inyo lava and 101 ka Deer Mountain lava, zircon populations with crystallization ages that cluster at 230 ka. Residual or thermally rejuvenated Long Valley magmatic crystal mush appears to have been involved.

During the 1350 CE sequence, four closely spaced sub-Plinian pumice eruptions issued from three vents now capped by the lava flows (Miller, 1985), blanketing the region with substantial pumice-fall and derivative reworked deposits. Isopachs were drawn by Miller (1985), and grain size and componentry were given by Nawotniak and Bursik (2010). Along Glass Creek gorge between the Obsidian and Glass Creek flows, the cumulative sheet of coarse ejecta is >50 m thick, and it might also be as thick adjacent to the Deadman Creek flow where it is little incised. Many rhyolitic ejecta of varied textures are larger than 1 m proximally, and abundant rounded boulders of granite and basalt entrained from glacial deposits are 1–2.5 m across. Poorly sorted tack-welded agglutinate is several meters thick in proximal fall deposits along Glass Creek. The fall deposits thin to 10–15 m by 1–2 km away from the vents, but the combined thickness of the pair of fall units that dispersed southward is still ∼1 m at Mammoth Mountain and Devils Postpile, 7–10 km south of the Deadman Creek flow. No paleosols or erosion surfaces separate the four fall units, and no fallout mantles the 1350 CE lava flows.

Major and trace element data for all eruptive units of the Mammoth system are given in Supplemental File 1 (see footnote 1) and selectively plotted in Figures 23–28. Chemical data for several postcaldera Long Valley rhyolites were given by Heumann and Davies (1997), Heumann (1999), and Heumann et al. (2002), along with Nd, Sr, Pb and U-Th isotope data for many of them. For numerous mafic lavas, as mapped by Bailey (1989), chemical data and Nd, Sr, and Pb isotope data were given by Cousens (1996).

A conventional alkali-silica diagram (Fig. 23A) illustrates the contrast between the alkaline Mammoth Mountain suite (100–50 ka) and the postcaldera subalkaline rhyolites (750–100 ka) that erupted from the residual Long Valley system a few kilometers to the east. Relative to postcaldera Long Valley rhyolites, the Mammoth Mountain suite is less silicic but more alkalic; this generalization likewise applies to all precaldera and caldera-forming Long Valley rhyolites (Hildreth, 2004, fig. 4 therein). Figure 23B shows that, except for a few basalts, nearly all of the many peripheral eruptive units (230–8 ka) of the Mammoth system are likewise alkaline by the criteria of LeBas et al. (1986). A notably subalkaline exception is the Holocene basalt of Red Cones (unit brc), the youngest eruptive unit in the Mammoth system.

Figure 24 provides a compositional overview of the eruptive sequence, focal and peripheral, showing apparently unsystematic variation in SiO2 content with eruption age since ca. 230 ka. It is noteworthy, however, that the 5 silicic andesite units (a62, aic, aml, amp, apb) and 2 peripheral dacites (ddl, drf), all phenocryst poor and widely scattered, erupted during the interval 140–90 ka, which overlaps the final episode of Long Valley rhyolite extrusions and initiation of Mammoth Mountain edifice growth. The bulk of the silicic edifice was constructed in several eruptive pulses between 90 ka and 60 ka. In contrast, eruption of mafic magmas (48%–55% SiO2) was fairly persistent over the interval 230–8 ka, with the exception of an eruptive lull at 60–30 ka, which coincided with extinction of the Mammoth Mountain edifice.

Figures 25–28 illustrate, for a large number of samples from all units of Mammoth Mountain and its periphery, the extent of compositional variation within each eruptive unit and the chemical basis for correlating mutually isolated exposures of many such units.

Mammoth Mountain

The 25 eruptive units distinguished for Mammoth Mountain range continuously from 63% to 71% SiO2 and from 3.6% to 5.1% K2O (Fig. 25A). Contents of MgO (0.2%–1.7%), FeO* (1.4%–4.7%), TiO2 (0.30–1.04), and CaO (0.6%–3.7%) all decrease fairly linearly with SiO2 (Figs. 25B, 25C). Broader, more scattered arrays are exhibited by Na2O (4.5%–5.7%), owing in part to hydration of generally glass-rich samples, and by Al2O3 (15%–17%) and Ba (1250–1900 ppm), P2O5 (0.03%–0.40%), and Zr (270–480 ppm), owing presumably to heterogeneous distribution of feldspars, apatite, and zircon. Among the more phenocryst poor rhyodacites alone, Ba ranges from 1250 to 1620 ppm and Zr from 270 to 480 ppm (Fig. 25D), evidently reflecting wide differences in degrees of sanidine and zircon retention in the crystal mush from which successive rhyodacitic melt batches were extracted. A plot of Sr-Rb (Fig. 25E) shows a moderately coherent trend dominated by plagioclase fractionation until reaching the rhyodacite range, where sanidine retention in mush (or removal by crystal fractionation) suppressed Rb enrichment in the most evolved melts. No systematic compositional trends with time are recognized within the Mammoth Mountain suite, as more and less evolved units were emplaced both early and late during construction of the edifice (Figs. 24 and 25).

The longest arrays (1.5%–2.9% SiO2), though narrow, are shown by eruptive units dfl, dml, dms, and dwr. In contrast, units dlp, dsd, dtl, rce, rmf, rsq, and rss all exhibit tight arrays that extend through <1% SiO2. Samples of units dbp, dom, ddl, and ddu overlap linearly within the limited SiO2 range 65.7%–67.1%, consistent with their stratigraphic stacking on the southeast side of the edifice. Dome 2861 (unit d61) stands out among trachydacites in its enrichment in Ba and Zr (Fig. 25D) and in its high Na content, which is reflected in its relative isolation in Figure 23A.

The alkalic rhyodacites, poorer in phenocrysts than the trachydacites, tend to be compositionally bimodal (Fig. 25); units rce, rfp, and rrc are richer in K and Nb than units rss and rsq but poorer in Ti, Fe, Mg, Ca, Sr, and P. Unit rmf, slightly richer in phenocrysts and erupted at a different vent, is compositionally intermediate among the rhyodacites in some respects but has lower Fe and higher Ba contents than the other rhyodacites. Considering the Mammoth Mountain suite as a whole, there is remarkable chemical and mineralogical variety among the 25 silicic units (Fig. 25) that built a compact edifice over a lifetime of ∼50 k.y., a variety that presumably reflects both persistent mafic recharge (recorded as blebs and enclaves) and effective crystal-melt fractionation of its crystal-rich multiphase phenocryst assemblage.

Mafic Periphery

Of 37 eruptive units peripheral to Mammoth Mountain, most are alkaline (Fig. 23) by the International Union of Geological Sciences criteria of LeBas et al. (1986); 6 are transitional (units bcd–bed, bsm, mdn, mlc, and msj), and 4 are subalkaline olivine-rich basalts (units bfh–bmc, bhl, and Holocene brc). The six andesitic and three dacitic peripheral units are all unequivocally alkaline. The least silicic units (48.4%–50.4% SiO2), bsm, bfh–bmc, bhl, and brc, are among the least alkaline, and bhl and brc are among the youngest basaltic units in the Mammoth system. The most magnesian units are bhl (9.9%–10.5% MgO) and brc (8.1%–8.3% MgO), and units bcd–bed, bfh–bmc, and bsm have 6%–7% MgO. For clarity, plots of data for the many peripheral units are separated into three sets of figures, for units principally in the South Moat, West Moat, and San Joaquin drainage (Figs. 26, 27, and 28, respectively).

Tight compositional arrays are exhibited by units asr, bfh–bmc, bhl, brc, bsc, bsm, mcv, mdm, mdn, mdp, mkv–mmc, mlc, msc, and mss, each having <1% SiO2 content (Figs. 26–28). The widest arrays (each ranging through 2%–4% SiO2) are shown by units aic, amp, mcl, and mor, all of which are phenocryst-poorer, and by the contaminated hybrid dacite chain of unit dnw (60.4%–67.3% SiO2).

The several following points are illustrated in Figures 26–28.

  • (1) Lava-flow aprons bcd, bfh, and mmc correlate well compositionally with their inferred vent cones, respectively units bed, bmc, and mkv, from which they are surficially separated by younger units.

  • (2) Spatially associated units mnd and mor and scoria from their inferred (but poorly exposed) vent complex, unit msd, together yield an overlapping continuous array (Fig. 27).

  • (3) Units mss and mdp, directly superimposed on the canyon floor of the Middle Fork, yield tightly adjacent (but nonoverlapping) compositional groupings, centered respectively at 53% and 54% SiO2 (Fig. 28). Despite a measured age difference of ∼40 k.y., the apparent chemical affinity of the two units disfavors location of the unknown vent for younger unit mdp far from that of unit mss, which is exposed near Soda Springs Campground. Confirming their independence, nonetheless, the two units have distinguishably different paleomagnetic directions (Fig. 21) and Sr, Nd, and Pb isotope ratios (Cousens, 1996).

  • (4) Subunit bar′ (51.1%–52.4% SiO2), which crops out in two areas within the primary apron of extensive phenocryst-rich unit bar (50.6%–51.2% SiO2), is a low-Mg, high-Al variant that shares similar concentrations of Ti, Fe, Ca, and Na and has only slightly lower K and P (Fig. 27). Although petrographically similar to most exposures of unit bar, the subunit carries slightly more plagioclase and only half as much olivine. It is inferred to be a late flow within the poorly exposed medial apron of bar flows.

  • (5) TiO2 concentration generally exceeds 1.2% in all the mafic units (although not in the dacites), but it peaks at >2% in intermediate units mic and mkv–mmc, in the intermediate suite mor-mnd-msd, and in basaltic unit bsr, all of which are phenocryst poor (Figs. 26 and 27).

  • (6) Phenocryst-rich unit bhl is far more enriched in CaO (10.1%–10.7%) than any other unit mapped here (Fig. 26B); unit bhl has 5%–7% cpx, 1%–2% olivine, and 10%–15% plagioclase, and is also the richest in MgO (9.9%–10.5%) and Sr, unusually rich in P, and relatively poor in Al. For its low SiO2 content, the unit has moderate K2O but unusually low Na2O, accounting for its contrasting positions in Figures 23B and 26A.

  • (7) Notably enriched in Al2O3 are units mlc (20.1%–20.9%), bar′ (19.1%–19.9%), and mdn (18.6%–19.5%). Each has 15%–20% plagioclase phenocrysts, somewhat more than other crystal-rich units like bcd, bfh, and bar. The latter three and most other mafic units here have peak Al2O3 values in the range 17.0%–18.5%, while the andesites and dacites generally have lower peak values. Noteworthy for its unusually low Al2O3 content is unit bhl (15.0%–15.7%), probably owing to its (possibly accumulative) abundance of cpx and olivine phenocrysts.

  • (8) Among the mafic units (48%–55% SiO2), a wide range in alkalinity, especially in K2O, tends toward bimodality (Figs. 23B, 26A, and 27A). Scrutiny of the units composing the contrasting arrays apparent in those diagrams reveals no correlation with age or geographic distribution.

  • (9) Units ddc and drf have 66%–67% SiO2 and are among the most evolved units peripheral to Mammoth Mountain (Figs. 23B, 27, and 28) and similar crystal-poor trachydacites. Although their outcrops are now separated by >6 km on opposite sides of the Sierran divide, they both appear to have erupted from vents later buried by growth of the Mammoth Mountain edifice, and both yield radioisotopic ages of ca. 100 ka.

  • (10) Although most of the peripheral units have yielded compositional ranges of ≤1% SiO2, several have ranges of ∼2% (units a62, aic, bar, bmn, and mcl), and a few have larger ranges (units mor, 3.1%, and amp, 4.4%). Several trachydacites of the Mammoth Mountain edifice yielded comparably wide compositional ranges. The hybrid dacite chain of the northwest moat (unit dnw), with a range of 7% SiO2, is the special case of a contamination series. Such ranges illustrate the importance of analyzing several samples from each unit. Single analyses could misrepresent a unit, would miss evidence for fractionation or mixing trends within units, and might tempt unwarranted modeling of affinities between unrelated units.

Radiogenic Isotopes

Sr, Nd, and Pb isotope ratios were measured for 28 samples of the monogenetic peripheral units of the Mammoth system (Cousens, 1996; Bailey, 2004), representing eruptive units we can now identify as a62, aic, amp, bar, bcd, bfh, bhl, brc, bsr, drf, mcl, mcv, mdp, mor, and mss. Most samples cluster narrowly in the ranges 0.70613–0.70638 for 87Sr/86Sr; 0.51243–0.51259 for 143Nd/144Nd; and 19.16–19.27 for 206Pb/204Pb. Two 87Sr/86Sr values determined for crystal-rich Mammoth Mountain trachydacites (units d81 and rce) plot within the tight mafic cluster, as does crystal-poor trachydacite unit drf. Plotting just outside the 87Sr/86Sr cluster (but not so for Nd or Pb) are units bfh (0.70591), mss (0.70596), and bhl (0.7067; contaminated?). The only other outlier is the Holocene basalt of Red Cones (0.70516–0.70531 and 0.51277), unit brc, which is one of the few subalkaline peripheral units as well as one of the most primitive and the youngest.

Although basement xenocrysts plagued early efforts to determine K-Ar ages of several of these units (Curry, 1971; Bailey et al., 1976; Mahood et al., 2010), Sr-Nd-Pb isotope data for basement rocks show that upper crustal contributions to mafic and dacitic products of the Mammoth system are small. For example, a metavolcanic schist of unit Mzmv near Agnew Meadows gave values of 0.72396 for 87Sr/86Sr and 18.94 for 206Pb/204Pb (Cousens, 1996). Numerous metavolcanic rocks from the Ritter Range Pendant gave 87Sr/86Sr values that range from 0.7053 to 0.7264, most of them >0.709 (Kistler and Swanson, 1981). Metasedimentary rocks of the Mount Morrison roof pendant just south of Long Valley caldera gave 87Sr/86Sr values that range from 0.7090 to 0.7250 (Goff et al., 1991). Granitoid plutons underlying the volcanic field have given 87Sr/86Sr values ranging from 0.70666 to 0.70987, and many other Sierran plutons nearby range from ∼0.707 to ∼0.715 (Goff et al., 1991; Cousens, 1996; Bailey, 2004). The narrow isotopic range for Pleistocene lavas of the Mammoth system is incompatible with significant contributions from such basement rocks.

Neogene Periphery

This paper makes no attempt to discuss in detail the many Neogene basaltic, andesitic, and dacitic units preserved around the periphery of the Mammoth system, on the rim of Long Valley caldera and within the Middle Fork San Joaquin drainage system (Bailey, 1989). Their eruptive volume was at least five times greater than that of the entire late Pleistocene Mammoth system, and most of them erupted during the Pliocene. No mafic or intermediate unit is known to have erupted between the end of the Pliocene (2.6 Ma) and ca. 230 ka. Because Pliocene units Tacc, Tbtb, and Tmcw were previously mapped as Quaternary (Bailey, 1989), however, we comment here on their ages and compositions. We also summarize results of our search within the Middle Fork drainage system for the unknown source vents of units Tbtb, Tmcw, and mdp.

The pair of glaciated domes combined as unit Tacc (Fig. 9) on the granitic rim high above Crater Creek is compositionally bimodal. Although both are phenocryst poor and similarly contaminated with granite-derived xenocrysts, the southwest dome has 58.3%–58.7% SiO2 and the northeast dome has 61.3%–62.6%. They differ likewise in Fe, Mg, and Ca contents, but what confirms their affinity is their contiguity and unusually low concentrations of P2O5 (∼0.3%) and TiO2 (∼0.6%; Fig. 28B). The southwest dome yielded a 40Ar/39Ar date of 4.29 ± 0.20 Ma.

The basalt of the Buttresses (unit Tbtb) is a severely glaciated stack of many lava flows preserved as 5 discrete remnants that span an elevation range of 470 m on the west wall of the Middle Fork canyon. As the largest remnant still extends down to river level, they appear to represent a conformable set of flows that once filled the canyon to a depth as great as 500 m. A columnar flow overlying granite gave a 40Ar/39Ar date of 3754 ± 7 ka, consistent with its reverse paleomagnetic polarity. Although all the flows examined are rich in olivine and cpx, the stack is compositionally bimodal, and the two compositions are interlayered in all sectors. Five flows sampled contain 46.5%–47.5% SiO2 and lack plagioclase phenocrysts; five other flows have 49.8%–50.6% SiO2 and carry a few percent plagioclase. Mg and Al increase systematically with SiO2 while Fe, Ti, and P decrease steeply and Ca modestly within a 10 sample array (Fig. 28). Preservation of the base of the Tbtb stack on granitic basement at river level indicates that the floor of this south-flowing reach of the Middle Fork canyon had already attained its present depth in the Pliocene.

Across the river and 2 km farther downstream along the east wall, a remnant pair of columnar lava flows (unit Tmcw) resembles some flows of unit Tbtb in containing subordinate plagioclase and abundant olivine and cpx; the unit yields a reverse paleomagnetic direction similar to that of Tbtb. Nonetheless, the flows are more silicic (52.3%–52.6% SiO2) and less magnesian than any flows of unit Tbtb (Fig. 28). They probably represent a Pliocene eruptive episode close in time to that of unit Tbtb. Vents have not been located for either unit.

In trying unsuccessfully to locate vents for units Tbtb and Tmcw (and for late Pleistocene unit mdp), we sampled nine more mafic remnants along the Middle Fork canyon (units Tasj, Tbdp, Tbld, Tbpl, Tmcl, Tmcm, Tmel, Tmsd, and Tmwc; Fig. 28); we recognize vents for six of them. Because none of the nine provided satisfactory chemical or petrographic matches, the search was extended, unsuccessfully, farther upcanyon, to Agnew Pass and Clark Lakes north of our map area (Hildreth and Fierstein, 2015).

Three more Pliocene mafic remnants, units Tbmm, Tbrm, and Tmcr, also plotted in Figure 28, overlie Mesozoic basement along the rim of the Lakes Basin. They neither correlate with each other nor with Pliocene units recognized elsewhere, and source vents have not been located for any of them. Unit Tbmm is the type locality of the Mammoth reverse polarity subchron (Doell et al., 1966).

The Neogene samples plotted in Figure 28 include three plugs high on the west wall of the Middle Fork drainage. That these units, Tmcl, Tmel, and Tmwc, have been reduced to nothing but massive vent-filling lavas draws attention to the severity of Pleistocene glacial erosion on the east flank of the Ritter Range (Figs. 1 and 2). Such upland erosion highlights even more strongly the contrasting evidence from unit Tbtb that the granitic floor of this segment of the Middle Fork had already been cut to its present depth by the middle Pliocene.

Of the eight late Pleistocene peripheral units along the Middle Fork (Fig. 28), all are alkaline except the Holocene basalt of Red Cones (unit brc). The Neogene units plotted in Figure 28 are similar to the Pleistocene alkaline units in total alkali contents, but almost all are more potassic (Fig. 28A), as well as richer in Ca, Mg, Sr, and Ba (Figs. 28B–28E). At 5.2% K2O, olivine-rich Pliocene unit Tmcm is shoshonitic (Fig. 28A), and in common with unit Tmcm, unit Tbpl is unusually rich in Sr (Fig. 28D).

Glacial ice accumulated upon and blanketed all of Mammoth Mountain after its growth in the late Pleistocene. The edifice completely postdates MIS 6, and no evidence is recognized here for glaciation during MIS 4, so most or all glacial erosion of Mammoth Mountain took place during MIS 2, the interval generally called the Tioga glaciation in the Sierra Nevada. Tioga Till dominates the glacial deposits of the Mammoth Lakes area, principally as a piedmont apron of ∼40 nested morainal ridges that extend as far as 8 km east of the summit of Mammoth Mountain. Because all 40 include clasts of basaltic unit bhl (31 ± 1 ka), the entire Tioga assemblage appears to have been deposited during MIS 2, regardless of the many ice-front fluctuations and stillstands represented.

No deposits assignable to a Recess Peak advance (14.2–13.1 ka; Clark and Gillespie, 1997) or to Little Ice Age (ca. 1450–1850 CE) cirque glaciers have been identified in the Mammoth region. Elsewhere in the eastern Sierra, moraines of those ages are at elevations higher than 3300 m, higher than any cirques in the Mammoth Lakes area. No evidence for a Younger Dryas glacial advance (12.9–11.5 ka) has been found in the Sierra Nevada (Clark and Gillespie, 1997; Gillespie and Clark, 2011).

Prior to growth of Mammoth Mountain, ice from the Lakes Basin and Sierran crest had advanced ∼1.5 km farther east than distal Tioga moraines, depositing some poorly preserved moraines called the Casa Diablo Till (Curry, 1968). Kesseli (1941) and Curry (1968, 1971) had recognized that mafic lava flows near Mammoth Creek separated tills of different ages and that the older (Casa Diablo) till lacks clasts of Mammoth Mountain dacite. Lava flows have now been precisely dated that underlie (162 ± 2 ka; unit mmc) and overlie (125 ± 2 ka; unit bcd) the Casa Diablo Till, confirming that it represents MIS 6.

Lavas of units amp (ca. 97 ka), bmc (ca. 93 ka), and bhl (ca. 31 ka) on the floor of the Lakes Basin cirque indicate the absence of glaciation when they were erupted. Likewise, the emplacement of lava flows of units mss (ca. 121 ka), drf (ca. 99 ka), amp (ca. 97 ka), and mdp (ca. 82 ka) directly on the granite floor of the Middle Fork San Joaquin River indicates ice-free conditions along that repeatedly glaciated canyon during those time intervals.

Two tongues of Tioga Till extend eastward from sources exclusively on Mammoth Mountain. One extends 3 km from the northeast base of the edifice into downtown Mammoth Lakes; the other extends 3 km along the valley between Dragons Back and Lincoln Peak as far as Juniper Ridge. A sandy-gravelly, fines-poor matrix predominates and encloses boulders and smaller stones that are nearly all (>99%) Mammoth Mountain dacites and rhyodacites, along with sparse mafic lavas and rare metavolcanic and granitoid clasts. Stones range from subrounded to subangular, and many are 10–50 cm across (rarely 1–2 m). In contrast to granitoid boulders, ice transport of flow-foliated Mammoth Mountain lava tends to refracture rather than round the clasts.

The downtown tongue of till was previously interpreted as the deposit of a debris avalanche from Lincoln Peak (Bailey, 1989), but it lacks hummocks, megablocks, and composite blocks. Moreover, hardly any of its varied Mammoth Mountain clasts (64.0%–70.5% SiO2; 12 analyzed) are derived from Lincoln Peak, which lacks evidence for an amphitheater or other avalanche scar. The volume of the 3 km2 downtown tongue (∼0.1 km3) exceeds that of Lincoln Peak.

A third lobe of glacial till, as wide as 3 km, extends along both sides of Dry Creek from the north base of Mammoth Mountain to a sharp terminus 8 km northeast. The hillocky sheet supports few well-formed moraines, probably reflecting in situ ablation of dead ice within a debris-laden piedmont lobe. Few stones are larger than 1 m. Clasts in the till near Dry Creek are predominantly basaltic, but include abundant dacite from Mammoth Mountain. A clast count (n = 374) in a scenic loop roadcut (UTM 259/743) gave 51% crystal-poor mafic lavas (units mnd, mor, and mcl), 20% Mammoth Mountain lavas, ∼8% each of units ddc, bar, and bmn, and 4% biotite rhyolite. Along a fault scarp 1.5 km west of there (UTM 245/738), another clast count (n = 175) gave 53% crystal-poor mafic lavas, 12% Mammoth Mountain lavas, 18% ddc, 13% bar, 4% bmn, and no rhyolite. Because the deposit overlies unit mor (66 ± 2 ka) and is younger than all other contiguous units (except the Inyo ejecta of 1350 CE), it represents a glacial advance during MIS 2.

The Lakes Basin cirque at the southeast foot of Mammoth Mountain was the principal source of the set of ∼40 nested morainal ridges and associated ground moraine that extends 6 km eastward from the basin outlet to near Sherwin Creek Campground, where its most distal moraine loop is breached by Mammoth Creek. The till was deposited during MIS 2 by a broad piedmont glacier that terminated ∼1.5 km short of the Casa Diablo moraines of MIS 6. Some moraines are as high as 40–50 m, and numerous wells adjacent to such ridges penetrate 10–90 m of till. Moraine surfaces are boulder rich, sharp crested, and little weathered. Boulders are dominantly granitoids, but include metavolcanics, mafic lavas, Mammoth Mountain lavas, and rare metasedimentary rocks. The source of ice-transported debris was principally the Lakes Basin; lesser but important contributions are from the south slope of Mammoth Mountain and metavolcanic and granodiorite debris from Cold Water Canyon and the range-front scarp south and east of Mammoth Rock. See Figure 4 for the distribution of basement lithologies.

North of Mammoth Creek, stones in the till are 80%–90% porphyritic granite (unit Kmo) as large as 5 m, ∼5% crystal-rich basalt of unit bhl (31 ± 1 ka) as large as 3 m, 3%–5% metavolcanic rocks as large as 3 m, and sparser clasts of large-feldspar basaltic unit bmc, crystal-poor scoria and lava clasts of units mcl and amp, fine-grained granitoids, and rare hornblende-biotite rhyolites. The northernmost lateral moraines are poor in metavolcanic clasts and carry only sparse Mammoth Mountain dacites, but both of these clast groups are common in inner moraines close to Mammoth Creek.

South of Mammoth Creek, granodiorite stones (unit Krv) derived from the south wall commonly outnumber porphyritic granites (unit Kmo) derived from the Lakes Basin. In the southernmost lateral moraines, granodiorite typically predominates, but metavolcanic rocks (from Cold Water Canyon via the Lakes Basin) are also abundant and locally dominant. Sparser constituents are boulders of unit bhl, unit amp, and various Mammoth Mountain dacites. Few clasts of unit bmc are present south of Mammoth Creek. Innermost moraines south of Mammoth Creek are poor in Mammoth Mountain dacite but relatively enriched in porphyritic granite; this reflects the fact that, during recession, shrinking axial ice from the Lakes Basin flowed principally through a Twin Lakes outlet trough (Figs. 12, 15, and 17) that had already been deeply incised through the several Mammoth Mountain dacite units here (principally dbp, ddl, and dtl).

The presence of clasts of unit bhl (31 ± 1 ka) in all moraines on both sides of Mammoth Creek as far east as its confluence with Sherwin Creek proves that all were deposited during MIS 2, not in earlier Wisconsin time. The deposits overlie unit amp (97 ± 1 ka), unit ddl (58 ± 2 ka), and several older mafic and trachydacite lavas. If glaciers did form on or near Mammoth Mountain during MIS 4, their deposits were overrun and remobilized during the more robust MIS 2.

Were it not for its contiguity with Long Valley caldera, the Mammoth Mountain volcanic field might long ago have been recognized as an independent magmatic system. Like the larger Lassen volcanic center in the Cascade arc (Clynne and Muffler, 2010), it consists of a central cluster of phenocryst-rich dacite lava domes surrounded by a contemporaneous array of mafic volcanoes in an extensional tectonic setting. The Long Valley system had erupted ∼850 km3 of rhyolite magma, all of it subalkaline and most of it phenocryst poor. By contrast, eruptive products of the Mammoth system are two orders of magnitude less voluminous and almost entirely alkaline. Nearly all of its silicic products are phenocryst rich, and none is rhyolitic.

Mammoth System versus Long Valley System

The Long Valley vent array was 30 km across, and for nearly 2 m.y. (2.2–0.3 Ma) it erupted nothing but rhyolite, centrally or peripherally. Its large silicic magma reservoir evidently prevented eruption of the mafic magma inferred to have sustained it thermally; mafic enclaves have been found in only 3 of ∼100 rhyolitic eruptive units. By contrast, the vent array of the Mammoth system is predominantly mafic and 10 × 20 km wide. Its central trachydacite edifice is only 5 km wide, the vent array of the edifice is only 2 km across, and many of its products carry sparse mafic enclaves and xenocrysts. The footprint of the mafic vent array is compactly circumscribed, extending west from the caldera ring-fault zone to the San Joaquin River and north-south from Deadman Creek to Pumice Butte. Beyond the limits of the well-defined Mammoth system, the nearest Quaternary mafic vents are in Mono Basin (at June Lake and Black Point) and 85 km southeast, near Big Pine.

Inception of Mammoth mafic volcanism ca. 230 ka culminated a 2-m.y.-long west-southwestward migration of the mantle-derived focus of deep-crustal basaltic injection that had successively energized the rhyolitic systems manifested as Glass Mountain, the Bishop Tuff, and the postcaldera Early Rhyolite. Growth of the Mammoth system sustained that trend, yielding a secular drift of the melting anomaly toward 240°, at a long-term average rate of ∼15 m/k.y from Glass Mountain to Mammoth Mountain.

The Mammoth Mountain edifice and almost all of the 37 monogenetic peripheral units vented outside (west or southwest of) the caldera ring-fault zone. However, proximity to the ring-fault zone of a few of the mafic vents (units bed, mdm, mkv; perhaps bsm and msc) inspires the hypothesis that their intrusive counterparts (165–120 ka) reenergized the margin of the long-crystallizing Long Valley reservoir to yield the 4 subalkaline rhyolite lavas that erupted in the west moat at 150–100 ka. Elsewhere, there had been no Long Valley eruptions since ca. 300 ka, and those between ca. 500 ka and 300 ka had been rhyolites of only modest volume.

A few of the Mammoth peripheral monogenetic vents are within the west moat, their locations thus conceivably influenced by proximity to a buried segment of the range-front (basement hosted) Hartley Springs fault zone that foundered with the 767 ka caldera. Most of the 37 peripheral units, however, vented in the Sierra Nevada, or high on the caldera wall, or beneath the Mammoth Mountain edifice. The footprint they define is broadly scattered, ovoid in plan, nonlinear, and (in contrast to the Holocene Inyo alignment) clearly not controlled by the range-front fault zone. In any case, the range-front fault zone terminated beneath what later became the west moat of the caldera, never having extended as far south as the site of Mammoth Mountain, instead having stepped left ∼15 km across the later site of the caldera, resuming as the range-bounding Hilton Creek fault south of the caldera margin (Fig. 1).

In contrast to the wide mafic array, the Mammoth Mountain trachydacite edifice is focused over a structural singularity (Fig. 4). It banks against the Long Valley caldera glacially modified topographic wall; it overlies the nearly vertical regional contact that separates thick sections of Paleozoic metasedimentary and Mesozoic metavolcanic rocks; it overlies the margins of one or more Cretaceous granitoid plutons that intrude the pre-Cenozoic sections; and it overlies segments of the nearly vertical Sierra Crest shear zone system (Fig. 4). Reflecting the field of mantle-derived magma ascent, the peripheral array of monogenetic vents is far broader than this structural nexus, but the intracrustal site of the Mammoth Mountain silicic magma body that developed at its core was probably localized by this convergence of older structures.

Rhyolitic versus Trachydacitic Culminations

Why does the edifice that extruded piecemeal from a silicic magma body at the center of the Mammoth monogenetic volcanic field consist predominantly of crystal-rich trachydacite? Why does it lack the rhyolite characteristic of Long Valley, Mono Craters, Coso, and many more Neogene and Quaternary volcanic fields in the extensional Basin and Range province? In common with many such fields, the Mammoth volcanic field tends toward compositional bimodality (Fig. 24), but the modes here are 48%–55% and 64%–70% SiO2, not basalt and rhyolite. The few silicic andesites (units a62, aic, aml, amp, apb) that occupy the gap between modes (Fig. 24) are all phenocryst poor and are thus presumed to have been melt batches that separated from more crystal-rich basaltic representatives of the mafic mode. Of the 25 silicic units exposed centrally, 19 are crystal-rich trachydacites (63%–69% SiO2) and 6 are alkalic rhyodacites (70%–71% SiO2) that carry only half as many phenocrysts. Between ca. 90 ka and 50 ka, melt-enriched fractions separated from trachydacitic mush 6 times as small rhyodacite batches buoyant and voluminous enough to erupt. Most of the trachydacite extrusions carry relatively mafic blebs and enclaves and scattered cpx crystals likely to represent persistent mafic recharge of a waxing and waning, recurrently stirred crystal-rich reservoir. As crystal-depleted extracts from the top of the reservoir, the rhyodacites lack evidence of such mafic cargo.

Products of the Pliocene volcanic field (ca. 4.0–2.6 Ma) that preceded the Long Valley rhyolitic episode are well exposed around the caldera rim (Bailey, 1989, 2004). Data define an alkalic compositional array that is fairly continuous from 49% to 69% SiO2, but, like the late Quaternary Mammoth suite, rhyolites are absent, and most of the silicic units are crystal-rich hornblende-biotite-feldspar trachydacites. The extended Long Valley rhyolitic interval (2.2–0.3 Ma), which was not accompanied by mafic eruptions, was thus bracketed by the Pliocene and Mammoth eruptive intervals, when alkalic basalts and their intermediate alkalic derivatives were accompanied principally by phenocryst-rich trachydacites but no rhyolites.

What accounts for the contrast between the exclusively rhyolitic episode and the temporally bracketing basalt to dacite episodes? It cannot be presumed that primitive magma input from the mantle remained mildly alkalic basalt throughout the 4 m.y. eruptive history. It seems likely that the contrasting patterns of intracrustal magma evolution reflect far more intensive mantle melting during the voluminous rhyolite interval. This would have entailed higher melt fraction subalkaline basalts, elevated rates of basaltic intrusion into the crust, and concomitantly greater degrees of deep-crustal partial melting, feeding back as an expanding barrier of mushy ductile crust that prevented further ascent of the basalts and thus intensified intracrustal melting. No mafic batches were erupted during the long rhyolitic interval, but rare mafic enclaves found in a few postcaldera rhyolitic lava flows are subalkaline basalt in unit rer (ca. 680 ka) and transitional basaltic andesite in unit rnc (ca. 500 ka).

Simon et al. (2007) inferred from U-Pb zircon ages and Pb isotope ratios of phenocrysts and host glasses that, during the Glass Mountain through Bishop Tuff rhyolitic interval (2.2–0.76 Ma), the Long Valley system had an increasing rate of rhyolite production, decreasing magma residence times, declining crustal contributions, and an increasing rate of mantle input. Such trends are unlikely to have reversed during the postcaldera eruption of ∼100 km3 of phenocryst-poor Early Rhyolite (ca. 750–650 ka), but the rhyolite eruption rate dropped drastically thereafter.

By contrast, during the low-flux dacite-culminating modes, lower melt fraction alkalic basalts would ascend from the mantle in smaller batches, lodge at deeper crustal depths, induce less crustal melting, and produce derivative products dominated by fractionation and remelting of mafic intrusions. Smaller batches crystallize faster, and reduced contributions from dehydration melting of crustal wall rocks would limit the water contents of fractionating magmatic hybrids, favoring crystal-rich viscous magmas of intermediate composition rather than accumulation of water-enriched, low-temperature, low-crystallinity rhyolite. During the Mammoth episode (and during the much longer Pliocene episode) of basalt to dacite eruptions, the volumetric eruption rate was an order of magnitude lower than during the extended Long Valley rhyolite interval. Assuming comparable rate ratios for fluxes of mantle-derived basalt, deep-crustal reservoirs probably remained modest during the low-flux episodes, sporadically releasing small andesite-dacite batches that could ascend to upper crustal subvolcanic chambers, where they would partially degas and crystallize extensively (Annen et al., 2006). By contrast, dacite and silicic andesite batches contemporaneous but peripheral to the Mammoth Mountain chamber ascended rapidly to the surface, erupting as crystal-poor units drf, ddc, aml, and amp (Fig. 24).

During the intervening long high-flux interval, deep-crustal melting was evidently far more extensive, areally and volumetrically, enabling the melting region to expand into the middle crust, inducing dehydration melting of hydrous protoliths, and releasing large batches of water-rich intermediate to silicic melt to shallow chambers where they could crystallize extensively and fractionate thick roof-zone layers of low-temperature, crystal-poor, high-silica rhyolite (Hildreth, 2004).

In summary, we envisage the 4-m.y.-long magmatic history to include a 2-m.y-long episode of vigorous mantle upwelling and subalkaline basaltic magma production bracketed in time by low-flux episodes of lower melt fraction alkalic basalt production. Derivatives of the main phase culminated in voluminous crystal-poor subalkaline rhyolite, whereas those of the less vigorous bracketing episodes culminated in crystal-rich trachydacite. The succession is not unlike that of oceanic hotspot volcanoes that move progressively across fixed melting columns, producing early and late alkalic episodes that bracket far more voluminous tholeiitic shield stages. Global positioning system velocities in the Long Valley region are 8–12 mm/yr northwestward (Oldow, 2003), in contrast to the southwest drift of the rhyolitic magmatic focus, and thus in complete conflict with a model of plate migration over a fixed melting column. Here, it is the mantle melting anomaly that waxed and then waned.

During the Pliocene, the basaltic flux was widely distributed, but modest, culminating in a field of trachydacite domes scattered from Crater Creek to San Joaquin Mountain, Bald Mountain, and Laurel Mountain, and probably spanning the later site of Long Valley caldera. The mantle-derived flux intensified and focused more locally by 2.2 Ma, first beneath Glass Mountain, then migrating west-southwest at ∼15 mm/yr beneath the sites of Long Valley and the Early Rhyolite, and finally diminishing after ca. 650 ka. How regional redistribution of intraplate extension may have led to a refocusing of intense mantle upwelling ∼20 km north, beneath Mono Craters, ca. 60 ka is a speculation we need not address here. Adjacent to the southwest margin of the Early Rhyolite, however, and on the trend of the west-southwest migration of the magmatic focus, a relocated but diminished and more distributed flux of mantle magma ascent initiated the Mammoth basalt to dacite system by 230 ka.

Mammoth System Magma Reservoir

Although the vent field for Mammoth Mountain and its 37 peripheral units covers nearly 90 km2, the total eruptive volume is small. The poorly constrained total eruptive volume for the peripheral array is between 7 and 12 km3, the most voluminous units being mcl, amp, bar, bfh, and mor-mnd. The scattered peripheral vent array can be compared to the present-day mid-crustal distribution of scattered long-period seismicity, which extends ∼10 km east-west from beneath Mammoth Mountain to the Middle Fork at depths of 10–18 km (Pitt et al., 2002; Hill and Prejean, 2005). The numerous long-period earthquakes are likely to represent injection of basaltic dikes and ascent of CO2-rich fluids derived from them. Beneath the long-period array, at lower crustal depths of 19–31 km, a few ascending swarms of brittle failure earthquakes in 2006–2009 were likewise interpreted (Shelly and Hill, 2011) as slip induced in otherwise ductile crust by pressurized fluids. Only a few dozen times in the 230-k.y.-long Mammoth episode did analogous dikes attain the surface and feed monogenetic eruptions; however, ascent of magma-derived CO2-rich hydrous fluids along fractures may be much more common. An 11-month-long earthquake swarm beneath Mammoth Mountain in 1989 (Hill et al., 1990) produced a vertical planar distribution of hypocenters at 6–9 km depth that was interpreted as fracture-bound fluid injection, possibly related to deeper dike intrusion beneath the brittle crust (Hill and Prejean, 2005). The several tree-kill sites peripheral to the edifice (Figs. 4 and 20), caused by elevated CO2 emission, were first observed shortly thereafter, in 1990.

Mammoth Mountain, despite uncertainties about glacial losses and depth of its concealed base, erupted no more than 4 ± 1 km3 of silicic lavas and pyroclastics (including the two off-edifice domes d61 and d81). The consistent similarity of eruptive products at Mammoth Mountain, i.e., their high crystal content, multiphase mineralogy, and limited range of bulk composition, throughout an ∼50 k.y active lifetime, suggests a stably located, relatively simple magma reservoir. Vents on the edifice are confined within a footprint only ∼2 km2 in area, and most vents are along a northwest trend (Figs. 5, 14, and 15) that crudely approximates that of the nearly vertical bedding and foliation of the basement (Fig. 4). The narrowly elongate vent corridor on the edifice may reflect northwest elongation of the silicic magma reservoir, as influenced by the basement structure, which might likewise favor a vertically prolate chamber and not favor sill emplacement. The widespread occurrence in the trachydacites of well-distributed relatively mafic cargo, i.e., enclaves, blebs, streaks, and plagioclase-cpx clots, attests to persistent mafic recharge and convective stirring within a compact unitary chamber. Nonetheless, over the course of a multiepisode 50 k.y. eruptive history, it should be expected that small intrusive masses, rinds, and dikes would crystallize, probably coexisting transiently with active pods of crystal mush that wax and wane in crystallinity, sporadically erupt, and less often fractionate melt-enriched rhyodacitic lenses.

The 87Sr/86Sr values of Mammoth Mountain trachydacites (units d81 and rce) and that of peripheral crystal-poor trachydacite unit drf are virtually the same (0.7063 ± 0.0001) as in most units of the contemporaneous mafic periphery, suggesting comagmatic descent (whether by direct fractionation or by also entailing partial remelting of forerunning crustal mafic intrusions). Sr-Nd-Pb isotope data for nearby basement rocks confirm that upper crustal contributions to mafic and dacitic products of the Mammoth system are small.

Mid-crustal seismicity and CO2 discharge under and around the Mammoth Mountain edifice persist in 2014, providing convincing evidence for mafic magma storage deep beneath the current footprint. In addition to abundant shallow brittle-failure seismicity (Prejean et al., 2003) under and within the edifice, numerous long-period earthquakes take place at depths of 10–18 km beneath the mafic vent array, from Mammoth Mountain to Red Cones and Devils Postpile (Pitt and Hill, 1994; Hill, 1996; Pitt et al., 2002; Foulger et al., 2003; Hill and Prejean, 2005). The area underlain by the long-period array is marked by an extraordinary conductive heat-flow anomaly (3.75 heat-flow units in a 250 m hole drilled in granitic unit Kmo near Devils Postpile; Lachenbruch et al., 1976), and it coincides with an extracaldera salient in the gravity low (Carle, 1988) that extends from Mammoth Mountain to Pumice Butte and Devils Postpile, an area uniformly underlain by the same Cretaceous granite. Although the silicic magma reservoir beneath Mammoth Mountain may have crystallized by ca. 50 ka, abundant geophysical evidence shows that mantle-derived mafic magma still intrudes the crust beneath the Mammoth volcanic field today, as it has since ca. 230 ka. The radioisotopic evidence summarized in Figure 24 shows that time intervals between eruptions here have rarely been as long as 10 k.y.; the most recent was ∼8 k.y. ago.

We thank Deanna Dulen and Wymond Eckhardt for engaging us in the geology, ecology, and history of Devils Postpile National Monument, which they have so thoughtfully superintended; Clifford Mann for a wealth of information on the history and logistics of the Mammoth Mountain Ski Area; James Saburomaru, Dean Miller, and Brent Turrin for contributions in the field and the lab; John Gottwald and Scott Lee for access to the Arcularius and Alpers Ranch properties; and Sally Drake and Bob Drake of Old Mammoth for hospitality and companionship. Rick Conrey of the Washington State University GeoAnalytical Lab assured production of remarkably consistent and reliable X-ray fluorescence data over a decade-long interval. Helpful manuscript reviews by Don Swanson, Bill Evans, Fraser Goff, John Geissman, Charlie Bacon, Maggie Mangan, Dave Hill, and others improved our presentation.

1Supplemental File 1. Chemical data for Mammoth Mountain and its mafic periphery, California. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental File 1.
2Supplemental File 2. Paleomagnetic data. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental File 2.
3Supplemental File 3. 40Ar/39Ar geochronology. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental File 3.