Sources of volcanic detritus in the basal Chinle Formation, southwestern Laurentia, and implications for the Early Mesozoic magmatic arc

The inception of the Early Mesozoic Cordilleran magmatic arc along western Laurentia is recorded by plutons in the Mojave Desert and Sierra Nevada of California, United States (Barth and Wooden, 2006), and in Sonora, Mexico (Arvizu et al., 2009; Riggs et al., 2010), and by rare forearc and intra-arc sedimentary successions in Sonora (González-León, 1997; González-León et al., 2005) and the Mojave Desert (Carr et al., 1997; Stevens et al., 2005; Rains et al., 2012). One of the compelling enigmas of the earliest (i.e., Permian–Early Triassic) stage of arc evolution is the lack of a coeval retro-arc sedimentary record. Lower–Middle Permian sedimentary units exposed in eastern California through western and southern Arizona to central Sonora are part of the miogeoclinal margin that was truncated in late Paleozoic time (Walker, 1988; Stone and Stevens, 1988; Bateman, 1992; Miller et al., 1992; Saleeby et al., 1992; Dickinson and Lawton, 2001). Detrital zircons in the Lower–Middle Triassic Moenkopi Formation are considered to record magmatism to the southeast, consistent with paleocurrents from that direction (Dickinson and Gehrels, 2008). Analysis of the oldest unit that carries a widespread, unmistakable record of the arc, therefore, is necessary to understand its evolving tectonic setting and evolution of the Late Triassic landscape as arc magmatism became dominant along the southwest margin of Laurentia.

The Upper Triassic Chinle Formation (Fig. 1) was deposited in a broad basin behind the early Cordilleran arc beginning at ca. 230 Ma (Lawton, 1994; Dickinson and Gehrels, 2008). The basal Chinle Formation, which comprises the fluvial Shinarump and Mesa Redondo members, records the earliest distinctive influx of arc-related detritus in the form of detrital minerals and volcanic clasts. We report here on a detailed detrital- and volcanic-zircon study of these basal fluvial sandstones and the volcanic clasts within them. Although paleocurrent directions (Stewart et al., 1972a) in general indicate that the Shinarump river systems flowed from southeast to northwest, a distinct Triassic detrital zircon signature together with the presence of Triassic volcanic clasts and local northeast flow directions record a significant contribution from the Cordilleran magmatic arc. We propose that the initial ∼35–40 m.y. of arc magmatism occurred in an offshore arc, broadly analogous to the evolving Ryukyu arc-trench system, where the Philippine Sea is subducting under the Eurasian continental plate. As subduction became established and arc crust evolved and thickened by intra-arc contractional deformation and magmatism, the arc and retro-arc rose above sea level, providing a fluvial pathway for the deposition of arc-derived detritus into the Chinle basin. The dominant direction of flow and sediment flux, however, remained from the higher-elevation landmass to the east. Our purpose, therefore, is to document the diversity within the basal Chinle Formation and to propose how arc magmatism along the margin is reflected in these retro-arc sedimentary rocks.

The Cordilleran magmatic arc developed across a truncated Paleozoic miogeoclinal margin along western Laurentia (Fig. 1) recorded by carbonate and siliciclastic passive-margin platform strata deposited on Precambrian basement. The early to mid-Paleozoic margin trended northeast-southwest, and Pennsylvanian–Permian sinistral strike-slip faulting created a northwest-southeast–trending margin along which the Cordilleran arc formed (Walker, 1988; Stone and Stevens, 1988; Bateman, 1992; Miller et al., 1992; Saleeby et al., 1992; Dickinson and Lawton, 2001). The oldest plutonic rocks and volcanic detritus that record arc magmatism are 270–260 Ma plutons in Sonora, Mexico (Arvizu et al., 2009; Riggs et al., 2010), and ca. 280–265 Ma volcanic cobbles in forearc sedimentary strata (Riggs et al., 2010; Lindner et al., 2012). Likewise, Miller et al. (1995) and Barth and Wooden (2006) documented Permian–Middle Triassic plutonism (260–235 Ma) in the Mojave Desert of California. Related volcanic sections are exposed in the El Paso Mountains region of southeastern California, where andesite has been dated at ca. 260 Ma (Martin and Walker, 1995), and in the east-central Sierra Nevada, where successions are generally younger (ca. 230–215 Ma; Barth et al., 2011) than the oldest arc rocks in the southern Mojave Desert and Sonora.

Concurrently with the major change in tectonic activity along western Laurentia, the late Paleozoic Ouachita orogen along southeastern Laurentia (Fig. 1) created highlands that were the source of detritus moving from southeast to northwest across southern Laurentia. The Lower–Middle Triassic Moenkopi Formation contains this detritus (Dickinson and Gehrels, 2008). Dickinson and Gehrels (2008) suggested that the western facies of the Moenkopi Formation were deposited in a flexural foreland related to the Permo-Triassic Sonoma orogen. Facies indicate a marine (Stewart et al., 1972b) or tidal-zone (Reif and Slatt, 1979) environment of deposition in the westernmost deposits of the Moenkopi Formation.

Deposition of the Chinle Formation resulted from dynamic subsidence related to development of the Cordilleran magmatic arc (Lawton, 1994; Dickinson and Gehrels, 2008). The formation comprises several members that dominantly record diverse fluvial environments (Stewart et al., 1972a; Blakey and Gubitosa, 1983; Lupe and Silberling, 1985; Dubiel, 1991). The Late Triassic age of the unit is constrained by paleontologic data (Camp, 1930; Colbert and Gregory, 1957; Fisher and Dunay, 1984; Lucas and Hunt, 1993) and by U-Pb dating of detrital zircons (LA-ICPMS: Riggs et al., 2003; Dickinson and Gehrels, 2008; ID-TIMS: Irmis et al., 2011; Ramezani et al., 2011, 2014; Atchley et al., 2013). In most exposures, the Chinle Formation was deposited on a significant erosional surface on the underlying Moenkopi Formation and represents a marked overall change in depositional environment from marine-tidal-sabkha to continental fluvial.

The most precise date for the lower part of the Chinle Formation is provided by Atchley et al. (2013). Zircons from a tuffaceous sandstone near the base of the Mesa Redondo Member yielded a maximum depositional age of 227.604 ± 0.082 Ma. The Mesa Redondo and Shinarump members are likely lateral equivalents (see also supplementary data in Irmis et al., 2011), and thus it is reasonable to infer that deposition of the Shinarump Member began no earlier than ca. 230 Ma (cf. Dickinson and Gehrels, 2008), as the sample dated by Atchley et al. (2013) does not come from the base of the formation.

The Shinarump Member comprises coarse to pebbly sandstone interbedded with conglomerate horizons. Clasts range from 1–2 cm to a maximum of 6–8 cm in diameter and are dominantly chert, quartzite, and limestone. Volcanic clasts are latitic, dacitic, and rhyolitic lava or tuff, are sparse, and make up a maximum of 5% of total clasts. Most clasts contain quartz, feldspars, and biotite (Table 1). Vitroclastic textures such as pumice and glass shards are common and indicate a pyroclastic origin, but evidence of welding is minor or lacking in the majority of clasts. A few clasts have euhedral phenocrysts and flow banding defined by devitrified glass and are interpreted as being derived from lava flow(s).

In general, volcanic clasts decrease in size and abundance from south to north, which has led to the speculation that the source of these clasts lay to the south in the “Mogollon Highlands” (Harshbarger et al., 1957; Stewart et al., 1972a; Dodge, 1973). Stewart et al. (1986) later noted that Mesozoic igneous rocks to the south of the Colorado Plateau are younger than Late Triassic and proposed that sources may have been faulted away from the area of southern Arizona, or lie hidden beneath younger cover. Reynolds et al. (1989) suggested that an unconformity between Proterozoic/Paleozoic rocks and Mesozoic sedimentary successions in western Arizona and eastern California records uplift that may have sourced detritus in the Chinle Formation, although those authors did not propose that this area was a source of volcanic material. More recently, Oberling et al. (2010) and Oberling (2015) have shown that volcanic clasts in the Shinarump Member show geochemical similarities to Triassic plutons in the Mojave Desert. Overall paleoflow was dominantly from the southeast (Stewart et al., 1972a; Blakey and Gubitosa, 1983), where highlands associated with the Ouachita orogen continued to contribute sediment. As discussed herein, however, the presence of volcanic clasts, Triassic detrital zircons, and local northeast-directed paleocurrents indicates that the Cordilleran volcanic arc was an additional and important source of detritus (e.g., Stewart et al., 1972b).

The Mesa Redondo Member is less well exposed than the Shinarump Member, overlying and interfingering with the Shinarump Member except where the Shinarump is not present (Cooley, 1958). In its type section and other regional exposures, the Mesa Redondo Member is dominantly siltstone and claystone with minor interstratified medium- to coarse-grained sandstone (Cooley, 1958; J.W. Martz and W.G. Parker, unpub. data). In many exposures of the Shinarump and Mesa Redondo members in northeastern Arizona, both units are capped by a well-developed plinthic oxisol (highly weathered deep soil horizon), locally referred to as the “mottled strata” (Stewart et al., 1972a; Dubiel and Hasiotis, 2011; Irmis et al., 2011) or “purple mottled unit” (Dubiel, 1987). This strongly suggests that these units are at least partially lateral equivalents (Irmis et al., 2011). Indeed, Akers (1964) did not distinguish the Shinarump and Mesa Redondo members in his geological map of eastern Arizona, and Therrien et al. (1999) interpreted the Shinarump and Mesa Redondo members together as channel and overbank deposits of the same fluvial system.

Samples from eight locations, including those first discussed by Howell (2010), form the basis for this study. In total, samples were collected over an area of ∼80,000 km², between ∼150 km and ∼450 km inboard of the Cordilleran arc (Muddy Mountains, Nevada, to Hunt, Arizona; Fig. 1). Post-Triassic deformation along the edge of the Colorado Plateau and in western Arizona–southern Nevada probably resulted in net extension between these sites and the arc. Precise correlation of clastic sedimentary units is impossible across these distances, and as discussed herein, the detrital zircon signature suggests that both the Shinarump and Mesa Redondo members make up the basal unit of the Chinle Formation. Therefore, we use the term “Shinarump conglomerate” to refer to the basal clastic unit that overlies redbeds and shallow-marine strata of the Lower–Middle Triassic Moenkopi Formation. In all cases, samples were collected from as close to the base of the Shinarump conglomerate as possible; the Hunt sample was collected stratigraphically a few meters above the sandstone dated by Atchley et al. (2013). All sample sites were traditionally assigned to the Shinarump Member except as noted. Tables 2A (analyses by LA-ICPMS) and 2B (analyses by SHRIMP-RG) provide analytical data for Shinarump conglomerate clasts; the Supplemental Table¹ has all analytical data for sandstone samples.

The Muddy Mountains, Nevada, site (Fig. 1; sample 111311-1, Supplemental Table) comprises coarse-grained sandstone interbedded with minor matrix- to clast-supported conglomerate lenses. Clasts are well rounded, poorly to moderately sorted, and granule- to cobble-sized chert and quartzite; no volcanic clasts were found at this site.

Detrital zircons from the Muddy Mountains are euhedral to rounded, with about equal parts rounded, euhedral, and subhedral grains. Euhedral grains are acicular to barrel shaped. In general, cathodoluminescence (CL) imagery shows that zoning bands within euhedral grains are parallel to grain boundaries and concentric through the entire grain, whereas those within rounded grains are rarely complete. In some euhedral grains, however, truncated zoning bands are overgrown by euhedral zoned rims.

The Paria, Utah, sample site (Fig. 1; sample 080310-1, Supplemental Table) exposes medium- to coarse-grained sandstone interbedded with minor matrix-supported conglomerate. Chert and quartzite clasts are typically well-rounded granules to pebbles; volcanic clasts are tan to white, angular, porphyritic fragments 1–3 cm in diameter that commonly occur in lenses. Volcanic clasts were not collected for analysis, however, due to their small size.

Detrital zircons from the Paria sample are dominantly subhedral to rounded, although rare grains are euhedral. Zoning bands are commonly truncated at grain boundaries. Euhedral and subhedral grains are barrel shaped to acicular.

The Cedar Ranch, Arizona, site (Fig. 1; sample 120310-2, Supplemental Table) is a small exposure on the northern margin of the San Francisco volcanic field. Medium- to coarse-grained sandstone is interbedded with conglomerate that contains well-rounded clasts of limestone, quartzite, and chert as much as 5 cm in diameter. Volcanic clasts were not found in the conglomerate at the Cedar Ranch site.

Far more of the detrital zircon grains from the Cedar Ranch sample are broken or blocky than in other samples. Most grains are barrel shaped or rounded; very few are acicular. In general, barrel-shaped grains have complete zoning bands preserved.

Howell (2010) interpreted sandstone near Cameron, Arizona, (Fig. 1; sample Cameron, Supplemental Table) as Mesa Redondo Member. The sampled bed comprises large-scale trough cross-stratified, resistant but slope-forming medium-grained arkosic arenite with localized pebbles of chert and felsic volcanic material; the latter make up 2%–3% of clasts.

Clasts (sample 110609, Tables 2A, 2B) were collected from a moderately indurated, medium- to coarse-grained sandstone with conglomerate interbeds; this site is traditionally assigned to the Shinarump Member (Stewart et al., 1972a). The clasts range in size from pebble to cobble and are well rounded with poor to moderate sorting. The clast compositions range from quartzite, chert, and limestone to silicified and/or sericitized, porphyritic, rhyolitic tuff and lava; maximum clast size is 5–6 cm. CL imagery was obtained only on zircons from one clast; zoning bands in general parallel grain margins.

The Holbrook, Arizona, site (Fig. 1; sample 052111, Supplemental Table; samples 103009, 052111, Table 2A) has the highest percentage of volcanic clasts of any of the Shinarump conglomerate exposures. The section comprises medium- to coarse-grained sandstone with conglomerate interbeds that contain clast- to matrix-supported, well-rounded, poorly sorted granules to cobbles. Petrified wood is common. Clasts are dominantly quartzite, chert, and limestone; volcanic clasts are ∼5% of all clasts. Maximum clast size overall is 5–6 cm.

The majority of grains are euhedral and have acicular habits, though some are barrel shaped. Although most zircons have well-preserved zoning bands, zoning patterns are commonly not apparent in rounded grains. Volcanic zircon grains are also euhedral, and in most cases zoning bands parallel crystal rims, although overgrowths are apparent in some cases.

The Shinarump Member at Joseph City, Arizona, (Fig. 1; sample JCNEWss, Supplemental Table; samples JC, JCNEW, Table 2A) is medium- to coarse-grained sandstone with poorly to moderately indurated, matrix-supported conglomerate interbeds. Clasts are well-rounded, moderately sorted granules to cobbles of limestone, quartzite, and chert, with rare porphyritic volcanic clasts. Maximum clast size is 4–5 cm and volcanic clasts are ∼2% of the total.

Cathodoluminescence imaging was not done on Joseph City detrital zircons, but zircon grains have a high percentage of euhedral, acicular forms. Fewer grains are rounded to barrel shaped. CL images from clast zircons are of very poor quality.

The North Park, Arizona (Howell, 2010; Fig. 1; sample North Park, Supplemental Table), sample material is from matrix-supported, trough cross-stratified cobble conglomerate with sandstone lenses. The sample was collected dominantly from a coarse- to very coarse-grained subarkosic arenite, with a small portion taken from a fine- to medium-grained fraction at the same stratigraphic level. Clasts are limestone and quartzite and are generally 3–5 cm, with rare white volcanic clasts <3 cm in diameter. Zircons from the North Park sample were separated and analyzed as part of Howell’s (2010) study and were not imaged.

Sandstone at Hunt, Arizona, has traditionally been assigned to the Mesa Redondo Member (Cooley, 1958) (Fig. 1; sample 050511-ss, Supplemental Table; samples 120309 and 050511, Table 2A). The sampled material is poorly to moderately indurated, coarse grained, and thin to medium bedded. Conglomerate interbeds and lenses are trough cross-bedded and 2–3 m thick and consist of matrix-supported, poorly to moderately sorted, well-rounded granules to cobbles of chert, quartzite, limestone, and ∼1% porphyritic volcanic clasts that typically weather to a pitted texture. Maximum clast size is 4–5 cm.

Detrital zircons from the Hunt sample are approximately equal parts acicular, barrel shaped, and rounded. Many rounded grains do not have clear zoning bands. More rarely, subhedral grains have zone bands that are truncated along the grain margin. CL images of clast zircons show euhedral, concentrically zoned grains.

Zircons were extracted from sandstone by standard methods (e.g., Gehrels, 2000) with minimal magnetic separation (0.1–0.3 amperes). Zircons from the Hunt, Holbrook, Joseph City, Cedar Ranch, Paria, and Muddy Mountains sandstone samples underwent annealing and chemical abrasion (cf. Mattinson, 2005; Riggs et al., 2013). Cathodoluminescence imaging to identify cores was done at Northern Arizona University using a JSM-6480LV scanning electron microscope. Zircons from the North Park and Cameron samples are from Howell (2010) and were not annealed or chemically abraded before analysis at the University of Arizona LaserChron Center. All detrital zircon samples were analyzed for U and Pb isotopes and for trace element concentrations at the University of California–Santa Barbara (UCSB) Laser Ablation Split Stream (LASS) facility using a Nu Plasma HR MC-ICPMS (high resolution multi-collector–inductively coupled plasma mass spectrometer), a Nu AttoM single collector ICPMS (Nu Instruments Ltd., Wrexham, UK), and an Analyte 193 excimer ArF laser-ablation system equipped with a HeLex sample cell (Photon Machines, San Diego, USA) using a 24 μm beam. Analyses are normalized against the 91500 zircon standard as a reference material (1062 Ma; Wiedenbeck et al., 1995). Data were reduced using Iolite 2.31 software in Igor Pro 6.3. Error assessment follows Kylander-Clark et al. (2013).

Zircons from the Cameron site and a portion of the North Park sample analyzed at the University of Arizona LaserChron Center were ablated by a New Wave UP193HE Excimer laser (operating at a wavelength of 193 nm) using a spot diameter of 30 μm. All measurements were made in static mode, using a Nu Plasma HR MC-ICPMS and Faraday detectors with 3 × 10¹¹ Ω resistors for ²³⁸U, ²³²Th, and ²⁰⁸Pb-²⁰⁶Pb, and discrete dynode ion counters for ²⁰⁴Pb and ²⁰²Hg. Ion yields were ∼0.8 mV per ppm. Each analysis consisted of one 15 s integration on peaks with the laser off (for backgrounds), fifteen 1 s integrations with the laser firing, and a 30 s delay to purge the previous sample and prepare for the next analysis. In this analysis mode, ablation pits are typically ∼15 μm in depth. Uncertainties shown in Table 2B are at the 1σ level and include only measurement errors. Data were reduced using an in-house program.

Zircons were extracted from volcanic clasts by standard methods, but were not annealed or chemically abraded. A minimum of four zircons were analyzed from each clast reported on here.

Zircons from most volcanic cobbles were also analyzed in the UCSB LASS lab using methods described above. Isotope ratios and elemental concentrations in zircons from four clasts (Table 2B) were measured using the U.S. Geological Survey SHRIMP-RG at Stanford University (California, USA). Zircons were mounted in epoxy and imaged with a CL detector on a scanning electron microscope. These images were used to guide selection of analysis points. Zircons were ablated using a ∼30-μm-diameter, 5–6 nA O₂^– primary beam. For Th and U concentrations and Pb/U ratios, data reduction used SQUID (a Microsoft Excel add-in that yields reduced and corrected isotope ratios and ages from raw SHRIMP U-Th-Pb data [www.bgc.org/isoplot_etc/squid.html]) and Isoplot software (www.bgc.org/isoplot_etc/isoplot.html), and followed procedures described in Barth and Wooden (2006). All ion microprobe zircon ages were standardized against Braintree Complex zircon R33 (419 Ma; Black et al., 2004); ages of individual analytical spots are reported at the 1σ level.

Detrital zircon analytical data are provided in the Supplemental Table, and all age errors reported are 2σ, unless stated otherwise. Analyses were evaluated for discordance based on a comparison of ²³⁵U/²⁰⁷Pb and ²³⁸U/²⁰⁶Pb for Permian and Triassic grains, and of ²⁰⁷Pb/²⁰⁶Pb and ²³⁸U/²⁰⁶Pb for Proterozoic grains. Grains that were >10% normally discordant (i.e., ²³⁵U/²⁰⁷Pb age >10% older than ²³⁸U/²⁰⁶Pb age) or >5% reversely discordant (i.e., ²³⁸U/²⁰⁶Pb age >5% older than ²³⁵U/²⁰⁷Pb age) were not used in interpretations; these are indicated by “discordant” in Table 1 and the Supplemental Table.

Twenty-three volcanic clasts in the Shinarump Formation each yielded between three and 29 grains with Triassic ages; clasts range in age from ca. 232 Ma to ca. 224 Ma (Tables 2A, 2B). All sampled locations contain clasts across this age range. Most clasts have a narrow range of spot ages, with error assigned by Isoplot’s TuffZirc of 1–5 m.y. (Tables 2A, 2B); we take 2 m.y. as a best estimate of analytical error. Rare older grains are likely antecrysts, and in only one case (sample 110906-1D, Table 2B) does a grain lie below and outside the error bar of the grains that are included by Isoplot in a TuffZirc plot. We infer the date assigned by TuffZirc to be the best estimate of crystallization age for the individual clasts. No grouping of ages is apparent, although the range of 227–229 Ma is most strongly represented (n = 13). Th/U ratios, which provide information about magma chemistry and may be a powerful tool for understanding the relation between detrital materials and putative sources (e.g., Riggs et al., 2012, 2013; Barth et al., 2013), vary relatively little. Averaged Th/U values for individual clasts in general cluster between 0.9 and 1.7 (Table 1). It is also noteworthy that clasts from the Hunt site, which has been assigned to the Mesa Redondo Member, have identical clast ages and Th/U values to Shinarump Member clasts.

Overall, detrital zircon signatures of the Shinarump Member have strong similarities as well as differences across our sample area (Figs. 2 and 3). All samples have a distinctive Triassic population, although the youngest zircon ages vary between locations. Detrital zircons from sandstones at the Paria and North Park sites are noteworthy in the low percentage of Permo-Triassic grains (8% and 17%, respectively) compared to 34–53% at Cameron, Cedar Ranch, Holbrook, Hunt, Joseph City, and Muddy Mountains. All samples have ca. 1400 Ma grains, and ca. 1100 Ma and 1600 Ma populations are common in most samples (Figs. 2 and 3).

Age data are complemented by Th/U values (Fig. 4). The majority of Permo-Triassic grains have Th/U values between 0.3 and 2.2; very few grains have ratios lower than 0.1.

Detrital zircons have a much broader range of Triassic ages than volcanic zircons at all sites. At Hunt and Holbrook, zircon ages and Th/U values of clasts and Triassic grains overlap significantly, although at Hunt several grains form an older cluster and at Holbrook the range is broader. The Joseph City plot of age versus Th/U shows an isolated cluster of grains that has no apparent relation to clasts and just one grain that overlaps the dominant clast age and narrow Th/U ratio (Fig. 5). The North Park and Paria samples have very few grains in the 235–220 Ma range.

Detrital zircon grains fall into several major age groups. Rare Archaean and earliest Paleoproterozoic (ca. 3250–2500 Ma) grains in the Holbrook, Joseph City, and Paria samples (1.5% of total grains) were likely derived from eroded older sedimentary strata and/or cratonal rocks in the mid-continent. Dickinson and Gehrels (2008) noted similar-age grains only in their samples from the upper Chinle Formation, and considered it likely that the younger river system lay in closer proximity to these sources. Alternatively, our samples may have come from lower-order drainages that did not contribute significantly to the lower Chinle Formation trunk system described by those authors.

Paleo- through Neoproterozoic grains (ca. 2100–900 Ma) are more than half the grains in several samples and 52% of total grains. Within this large grouping, populations that represent southwest Laurentian tectonic elements are common (Fig. 1). Grains were derived from arc terrains related to the Yavapai (ca. 1800–1700 Ma) and Mazatzal (1700–1600 Ma) provinces and A-type “anorogenic” granites that range in age from ca. 1500 to 1300 Ma (Whitmeyer and Karlstrom, 2007). Lastly, Grenville-age (i.e., 1300–950 Ma) zircons are as much as 45% of analyzed grains from some Shinarump samples. Although plutons of these ages are exposed across southwestern Laurentia, Proterozoic zircons may also have been derived from uplifted and eroded Pennsylvanian–Permian sedimentary and metasedimentary strata that were part of the arc basement. Zircon grains with Proterozoic ages are common in these strata throughout southwestern North America (e.g., Soreghan et al., 2002; Soreghan et al., 2007).

The derivation of some Paleozoic grains is more complex. In contrast to the findings of Dickinson and Gehrels (2008) in Chinle Formation strata farther to the east, very few (<1%) Cambrian grains (540–500 Ma) are present in the samples analyzed here. Approximately 3% of grains lie in a poorly defined range from ca. 300 to ca. 450 Ma; zircons of this age were attributed by Dickinson and Gehrels (2008) to sources in the Ouachita orogen.

Permo-Triassic zircons make up 40% of all detrital grains. Understanding their source(s) requires review of current knowledge about the distribution Permo-Triassic magmatic rocks in southwestern, southern, and southeastern Laurentia. Dickinson and Gehrels (2008) noted that few arc-related grains were present in their samples, and inferred that grains <232 Ma were derived from the Cordilleran arc, whereas those >245 Ma were derived from the “East Mexico” arc. The latter interpretation is supported by paleocurrents that dominantly indicate flow from the southeast. The few reliable dates that can be related to this arc, however, are from fine-grained lacustrine tuffs in the Quartermaster Formation and Dewey Lake Redbeds in west Texas that yield zircon ages of ca. 251 Ma (Fracasso and Kolker, 1985; Geissman et al., 2011). The source of tuffs of the Quartermaster and Dewey Lake Formations is unknown, but the fine grain size supports a distant source, most likely to the west or to the southeast. The majority of magmatic rocks of the East Mexico arc are dated by K-Ar (Torres et al., 1999). One important exception is the Tuzancoa Formation in Hidalgo, eastern Mexico, in which fossiliferous volcanic and volcaniclastic rocks are dated by fossil evidence as Late Permian in age (Rosales-Lagarde et al., 2005). These rocks, however, are marine and exposed ∼2000 km south of the Chinle depocenter, and thus are not a likely source. Likewise, the Totoltepec laccolith is dated by U-Pb as 289 ± 1 Ma (Keppie et al., 2004), but is much farther south than would be a likely source, as well as being an age that is very poorly represented in the Shinarump detrital population.

Cordilleran arc magmatism began by 270 Ma (Arvizu et al., 2009; Riggs et al., 2010), but detritus from that arc is present only locally in retro-arc sedimentary strata older than Late Triassic (i.e., Lower Triassic Buckskin Formation; Sanchez et al., 2014). Preliminary results from Permian forearc rocks in Sonora, Mexico, indicate a strong influx of arc detritus at ca. 273 Ma (Riggs et al., 2014). The proximity of the Cordilleran arc, together with the growing body of information concerning its range of ages, makes it the most likely source region, thus requiring explanation of how the ages represented were mixed into the Shinarump depocenter.

The range of Permo-Triassic ages also bears examination, as dispersal paths must accommodate disparate areas that would have been sources. The oldest common Permian grains in our samples are ca. 280–285 Ma. These, together with ages to ca. 260 Ma, make up <5% of the total grains and are rare in all samples except North Park, and are inferred to derive from the Sonoran segment of the arc (Fig. 1).

The San Bernardino intrusive suite and plutons in the northern Mojave Desert (Fig. 1; Miller et al., 1995; Barth and Wooden, 2006) likely sourced ca. 260 and ca. 240 Ma grains in many of the samples. The correlation with the San Bernardino suite is suggested both by age and by similar Th/U ratios (Fig. 6). Data presented by Miller et al. (1995) do not include Th/U values; our preliminary results from these rocks show a reasonable overlap in Th/U values between 260–250 Ma grains in these plutons and in the detrital zircons in the Shinarump Member. Age and Th/U also support the derivation of ca. 235–230 Ma grains from the Granite Mountain suite (Fig. 1).

Many grains younger than ca. 230 Ma have less-certain sources. The intrusive suites documented by Barth and Wooden (2006) have only rare 230–220 Ma ages. Additionally, many detrital grains in this range have Th/U ratios higher than both older components of the sandstones and rocks of the intrusive suites (Fig. 6). Riggs et al. (2012) speculated that an as-yet-undocumented source for zircons in middle Chinle Formation sandstones is buried or has been eroded, but lay approximately in the Colorado River basin.

The breakdown of ages and potential source areas allows distinctions to be drawn between the sample sites, leading, in turn, to enhanced understanding of dispersal pathways; sites are discussed here from west to east. The Muddy Mountains (Fig. 1) site contains a group of zircons between ca. 240 and 243 Ma which match the San Bernardino suite in age and Th/U, grains between 230 and 236 Ma likely derived from the Granite Mountains suite, and grains between ca. 225 and 230 Ma that do not have an obvious source. Thus we infer that this area, which was close to the arc, either was fed by a stream system that tapped a relatively confined area within the arc or received detritus from Plinian ash columns. Proterozoic grains are well represented and were derived either from uplifted arc basement or from easterly sources. The lack of igneous clasts, however, is surprising considering the likely proximity to the arc. It should be reemphasized, however, that overall the percentage of volcanic clasts in any Shinarump conglomerate bed is generally very low (i.e., ≤5%), such that the clasts are more anomalous by their presence than absence.

The Cedar Ranch sample site contains two distinct groups of Permo-Triassic zircons. A diffuse cluster between ca. 237 and 240 Ma is consistent in age and Th/U ratios with derivation from the San Bernardino and/or Granite Mountain suites in the Mojave Desert. Zircon grains between ca. 230 and 225 Ma dominate the Permo-Triassic signature of this site. These grains overlap Mojave Desert plutons in age and Th/U to a lesser extent than the older grains, and are in part distinctive from similar-age grains in the Muddy Mountains sample. Thus we infer that some recycling of older grains may have occurred, but that the 230–225 Ma group reflects derivation from a specific part of the arc and that these grains were genetically related to volcanic clasts found in other samples (cf. Riggs et al., 2013).

The North Park sample is unique in its comparatively high percentage (13%) of Permian grains. Six of these grain lie in a discrete cluster from 288 to 282 Ma, and their source is uncertain. Zircons of this age are inferred by Dickinson and Gehrels (2008) to derive from the East Mexico arc; we question this interpretation for reasons provided above. On the western margin of Laurentia, however, transcurrent faulting was likely active at ca. 280 Ma, and subduction was established by ca. 270 Ma. Thus these grains are not considered likely to be derived from the Cordilleran margin. The majority of grain ages are in a comparatively narrow range, from ca. 270 to 247 Ma (n = 13); ages and Th/U are compatible with derivation from the Mojave Desert plutons and the Sonora portion of the arc (Fig. 6).

The youngest grains from the Holbrook, Hunt, and Joseph City samples are very similar in their age and chemistry; Hunt and Joseph City samples additionally have an older group of grains in the ca. 240–260 range Ma that were likely derived from the San Bernardino suite and/or older plutons in the northern Mojave Desert (Miller et al., 1995); Th/U ratios support correlation with both suites of plutons (Fig. 6). Grains as young as 218 Ma may have been subject to lead loss; on probability density plots the maxima are between 226 and 223 Ma (Fig. 4), and these older ages may be more suitable estimates of the maximum depositional age. Th/U values are from within the range of the Granite Mountain suite to well above those values (Fig. 6). Clasts from all three sites, as well as those from Cameron, overlap the older part of the younger age range and have Th/U values that match higher detrital-grain values. We infer that these grains had a similar source to those from the Cedar Ranch site.

Grains that traveled from the Sonoran segment of the arc could reflect original deposition from ash clouds and strong dilution, or possibly longshore currents that brought detritus northward from the arc to be uplifted when detritus from the arc began to reach the continental landmass. Zircons that match plutons of the Mojave Desert in age and Th/U may have been carried by river systems or from ash clouds.

Our interpretation of the evolution of topography between the arc and retro-arc in Late Permian through Late Triassic time is by necessity constrained by a lack of detailed study of Late Permian rocks in southwest Laurentia. The southwest margin of Laurentia likely underwent induced subduction (cf. Gurnis et al., 2004; Stern, 2004; Saleeby, 2011) in Permian time as strike-slip faulting transitioned southward to subduction in northern Sonora, Mexico (Fig. 7; Arvizu et al., 2009; Riggs et al., 2010, 2014). Over time, Permian–Early Triassic magmatism became more widespread in California, forming an arc that lay offshore of the continental landmass. Very little detrital record remains of this time in equivalent-age strata, however, because the arc was marine and likely did not have substantial subaerial topography. The sedimentary record of the back-arc is confined to arc-proximal units such as the Buckskin Formation in eastern California and western Arizona, which in its lower members is equivalent to the Moenkopi Formation (Reynolds and Spencer, 1989) and which reflects magmatism at ca. 253 Ma (Sanchez et al., 2014), and in the southern Inyo Mountains in east-central California, where our data from the Lone Pine and Conglomerate Mesa Formations include prominent detrital zircon peaks at 263 and 257 Ma (Stevens et al., 2015; Riggs et al., 2015) (Fig. 7).

The date of 227.604 ± 0.082 Ma for the Mesa Redondo Member of the Chinle Formation, from a sample only a few meters above the contact with the Moenkopi Formation, supports initial deposition of the Chinle Formation at ca. 230 Ma, as suggested by Dickinson and Gehrels (2008), or 1–2 m.y. later. Between Early and Late Triassic time, arc magmatism in California transitioned to include volcanism. By this time, two major changes had occurred. First, a subaerial connection between arc and retro-arc was established that allowed fluvial transport of arc material to the Chinle basin (Fig. 7), perhaps by ca. 235 Ma. Second, because the record of volcanism is prominent as 230–225 Ma clasts, especially in younger conglomerates in the Chinle Formation, we infer that volcanic activity became common at this time, with eruption of intermediate to felsic lava and ignimbrite. Related Plinian ash clouds could have been a major source of fine-grained volcanic detritus. The observation that volcanic clasts are <5% of types in Shinarump conglomerate exposures suggests that the terrestrial connection between the marine arc and the continental landmass was limited in the earliest stages of Chinle deposition. Volcanic clasts are more common in the younger Sonsela Member, indicating that over time a subaerial fluvial connection became far more stable.

The Tr-3 unconformity separates Upper Triassic from Lower–Middle Triassic and older strata in western Laurentia (Stewart et al., 1972a; Pipiringos and O’Sullivan, 1978). Howell and Blakey (2013) postulated that this surface represents response to dynamic subsidence of the Cordilleran arc, citing the similarity in trend between the erosional surface and the arc, and that development of the unconformity was driven by migration of the arc onto the continental margin. The youngest three-grain cluster in the Holbrook Sandstone Member of the Moenkopi Formation (Dickinson and Gehrels, 2008) yields an Isoplot average age of 241 ± 2 Ma, which may provide an older age constraint on this event, and initiation of Shinarump Member sedimentation at ca. 230 Ma marks the latest time at which the arc would have become wholly subaerial. Howell and Blakey (2013) suggested that the on-shore migration of the arc occurred at ca. 235 Ma based on oldest Triassic detrital zircons in the Sonsela Member of the Chinle Formation; although our results show a much broader range of Permian–Triassic detrital grains in the older Shinarump Member, the ca. 235 Ma age is reasonable. In this case, erosion of underlying strata and development of the Tr-3 surface may have been a dramatic but relatively short-lived (i.e., 5–7 m.y.) event.

Processes that may have created a land connection between the arc and retro-arc region include crustal inflation due to emplacement of plutons, intra-arc tectonism and resultant uplift, and sea-level drop. Eustatic sea-level change in Triassic time, however, is documented as predominantly transgressive (e.g., Hardenbol et al., 1998; Kelley et al., 2014), suggesting that isostatic forces (i.e., crustal inflation or uplift) dominated. Permo-Triassic intra-arc deformation is documented in the southwestern Mojave Desert and adjacent San Bernardino Mountains (Miller, 1981; Cameron, 1981; Miller and Cameron, 1982; Matti et al., 1993), as well as in the Red Cloud thrust system (Powell, 1981; Postlethwaite, 1988) and the Sierra Nevada–Death Valley thrust system (Snow, 1992; Stevens and Stone, 2005). Deformation is poorly dated in all of these systems, but evidence from cross-cutting plutons indicates at least some Permian–Triassic movement (Tosdal, 1988; Barth and Wooden, 2006; Stevens et al., 2015). These data indicate that regional intra-arc and proximal retro-arc uplift and exhumation were likely occurring by earliest Triassic time and that arc basement exhumation probably continued until earliest Jurassic time (Fig. 7). Crustal inflation may occur in and beneath volcanic arcs due to several mechanisms (see Chin et al. [2012] for a review); Chin et al. (2012) used xenolith compositional evolution to interpret crustal thickening below the Cretaceous Cordilleran arc, but similar data are not available for the earlier manifestations of the arc, and it is unlikely that the volume of magma input was sufficient to drive substantial thickening. Thus, although we would not discount crustal thickening as a process to bring the arc and nearby land above sea level, intra-arc deformation may have held the more important role.

New zircon data from sedimentary rocks and enclosed volcanic clasts from the Shinarump conglomerate in Arizona, southern Utah, and southeast Nevada provide insight into paleotopography of the growing Cordilleran arc and the nature of arc volcanism. The Shinarump conglomerate is the oldest retro-arc unit to contain volcanic clasts as well as fine-grained detritus, and thus yields the most complete data set regarding retro-arc sedimentation and the erosion of the arc. Volcanic clasts are ignimbrite and silicic lava, indicating that the arc erupted material from differentiated magmas and likely included stratovolcanoes and calderas. The 232–223 Ma ages of theses clasts suggest derivation from a relatively confined area of the arc, in contrast to detrital grains, which have ages that span the known timing of arc plutonism (e.g., Miller et al., 1995; Barth and Wooden, 2006).

Analysis of Th/U values highlights the similarities and differences between inter- and intra-sample detrital grains and between detrital grains and clasts. Many of the detrital grains have Th/U values similar to those for Triassic Cordilleran arc plutons, and the combination of age and Th/U similarities strongly suggests derivation from those areas. Other grains, together with all of the clasts, have ages between ca. 230 and 225 Ma and Th/U values >1 that do not correspond to known Triassic Cordilleran plutonic rocks. These ages and Th/U values are similar to those in clasts and grain zircons from the Sonsela Member, higher in the Chinle Formation described by Riggs et al. (2012), and supports the speculation of those authors that material was derived from an undocumented source in the Colorado River trough where continental crust was thicker and provided a higher Th signature.

Arc activity along western Laurentia began at ca. 275 Ma (Arvizu et al., 2009; Riggs et al., 2010; Lindner et al., 2012), but the oldest detrital zircon record of this arc is in sedimentary strata some 40 m.y. younger, suggesting that the arc developed offshore. Early Permian plutonism and volcanism in Sonora, Mexico, was dominantly or entirely subaqueous. Magmatism became more widespread during Late Permian and Early Triassic time, and eruptions may have breached the air-water interface. Ultimately, volcanic edifices became common in Late Triassic time, potentially synchronous with intra-arc deformation, which had the effect of bringing the arc above water and establishing a land bridge between arc and continent at ca. 230 Ma. At this point, erosion of the growing arc began to be seen in retro-arc fluvial strata of the Chinle Formation, ushering in continental sedimentation that has dominantly characterized the region since that time.

Funding for this study was provided by the National Science Foundation through grant EAR-0711541 to Riggs. We are indebted to many colleagues whose discussions and studies have strongly informed our interpretations: R. Blakey, W. Dickinson, T. Lawton, S. Nesbitt, S. Reynolds, and S. Richard. Great thanks to C. Brailo for mineral-separation work, A. Kylander-Clark at UCSB for help in the LASS lab, the Tiffney-Gowan family for their hospitality, and E. Cullen for advice in drafting. Riggs also thanks Q. Crowley and B. Kamber at Trinity College Dublin (Ireland) for support during the sabbatical year. Reviews by Lynn Soreghan and Paul Stone were very helpful in identifying and clarifying murky ideas and writing, and are much appreciated.