In a regional study of Permian age sedimentary rocks in the American Southwest, Peterson and Nairn (1971) reported data from one site in what they referred to as the Dewey Lake Formation, exposed near Carlsbad, New Mexico. The site yielded a north-northwest declination and shallow positive inclination magnetization estimated mean direction (thus normal polarity). Overall, the study reported few examples of progressive demagnetization behavior and no supporting rock magnetic data.

The initial investigation of strata of what were assumed to be entirely of the Quartermaster Formation (again referred to as the Dewey Lake Formation) in west Texas were reported by Molina-Garza et al. (1989). The goal of that study, conducted in the Palo Duro Canyon and vicinity, was to establish an approximate age of the end of the late Paleozoic long reverse polarity (super)chron (Kaiman superchron), given that recent K-Ar age determinations had been carried out on biotite in the volcanic ash deposits of west Texas (Kolker and Fracasso, 1985). These age determinations, ranging from ∼261 to ∼251 Ma, were associated with relatively low precision and inherently biased toward older values. Molina-Garza et al. (1989) showed that these strata yielded interpretable paleomagnetic data of dual polarity, with a remanence largely carried in hematite. Overall, four magnetozones, beginning with one of reverse polarity, were identified in the sequence. The volcanic ash deposit sampled by Kolker and Fracasso (1985) was located within the older normal polarity magnetozone at each of the three sections sampled. Molina-Garza et al. (1989) reported an estimated (reverse polarity) mean direction, based on data from 16 sites of Decl. = 152.4^o, Incl. = -11.1^o, a95 = 5.9^o, k = 39.5. This mean direction, uncorrected for any potential inclination shallowing related to compaction/diagenesis processes, provides an estimated paleolatitude of about 6^oN, with associated errors.

Molina-Garza et al. (2000) reported data from the Dewey Lake Formation exposed at the Maroon Cliffs near Carlsbad, southeast New Mexico. They reported a N-R-N polarity magnetozone sequence, with the reverse magnetozone being confined to a stratigraphically thin interval. By correlation with a continuous drill core, which contains a volcanic ash deposit, obtained some 15 km south of the Maroon Cliffs section, Molina-Garza et al. (2000) proposed that the top of their sampled section was some 20 m below the volcanic ash deposit.

Steiner (2011) provided a summary of work on strata of the Whitehorse Group and Quartermaster Formation in west Texas and laterally equivalent strata in southeast New Mexico. Actual data were shown from Quartermaster Formation strata (again referred to as Dewey Lake Formation) exposed as the upper part of the Dickens section. Steiner (2011) showed that the thick (>0.5 m) volcanic ash deposit of this section lies in a reverse polarity magnetozone with much of the remaining sampled section defining a normal polarity magnetozone. This contribution also refers to data from the Quartermaster Formation (referred to as Dewey Lake) exposed at Caprock Canyons, but no data are actually shown, other than an interpreted magnetic polarity stratigraphy and paleomagnetic pole positions derived from an unreported number of sampled sites. In the reported Caprock Canyons magnetic polarity stratigraphy, a lower volcanic ash deposit lies in a normal polarity magnetozone and the upper volcanic ash deposit is interpreted to lie in a reverse polarity magnetozone. On the basis of the graphically plotted paleomagnetic pole positions and the interpreted magnetic polarity stratigraphy as compared with results from several sections of middle to upper Permian to Lower (?) Triassic strata on different continents, Steiner (2011) proposed that the age of the Quartermaster Formation (referred to as Dewey Lake) in northwest Texas is early late Permian (Wujiapingian Stage) to early latest Permian (into the Changhsingian Stage). Several in preparation studies are listed in the references for Steiner (2011), but we are not aware of their publication.

Stratigraphic sections presented here were measured and described following standardized field geology methods (e.g., Compton, 1985). Beds were delineated by recognizable differences in color, composition, sedimentary grain size, sedimentary structure, sedimentary architecture, and were measured using a 150 cm high Jacob’s staff mounted with a topographic Abney level while paying particular attention to the upper contacts between beds as descriptions were made upward through the stratigraphic succession. In particular, this work sought to document approximations of the dominant sedimentary grain size, color, sedimentary structure and sedimentary architecture as well as the contact relationships of bedding contacts which were observed amongst the strata in the sections in Palo Duro State Park, Highway 207, Caprock Canyons State Park, and Texas State Highway 82 (i.e. Dickens) outcrops. Schematic stratigraphic sections for these sections are presented as Supplementary Figures S1–S5. They are not presented in the main text because the emphasis of this work is to demonstrate the existence of the Permian-Triassic boundary within the Palo Duro Basin. The stratigraphic sections in Supplementary Figures S1–S5 provide an accurate representation of the dominant sedimentary grain size, color, sedimentary structure and sedimentary architecture as well as the contact relationship of bedding contacts that are observed among the strata at these sections.

Fresh hand samples were collected at each section by excavating ∼30 cm into the outcrop face. Samples were thin sectioned and examined petrographically using a transmitted polarizing light microscope. Powders used for X-ray diffraction analysis of carbonate mineralogy and analysis of carbonate δ¹³C and δ¹⁸O were drilled from hand samples using a rotary Dremel tool.

The carbonate mineralogy of each powdered sample was determined by X-ray diffraction (XRD) analysis at SMU using a Rigaku Ultima III X-ray diffractometer with a step size of 0.05° over a range of 27°–32° 2θ. The relative abundance of calcite and dolomite was determined using the ratio of the background-subtracted peak intensities of the 001 d(hkl) for the two minerals (3.03–3.04Å for calcite, 2.88–2.89Å for dolomite).

Methods for carbonate dissolution and extraction of CO₂ are based on empirically determined reaction times and temperatures for calcite and dolomite (Table 1; Al-Aasm et al., 1990). Samples containing only calcite were reacted with 100% orthophosphoric acid at 25°C for 2 h and cryogenically purified to produce CO₂ (McCrea, 1950). Samples containing both calcite and dolomite were reacted for 2 h, cryogenically purified for 30 min to remove the CO₂ gas evolved, then returned to the 25°C water bath and allowed to react for a further 24 h. After 24 h, the gases evolved were evacuated for 30 min, and the samples were transferred to a 50°C water bath for 4 h and then cryogenically purified to isolate CO₂. Samples containing only dolomite were reacted for 26 h at 25°C, and the evolved gasses were evacuated for 30 min. Then the samples were allowed to react for 4 h at 50°C, and the evolved gasses were cryogenically purified to isolate CO₂. The carbon and oxygen isotope composition of CO₂ evolved from both the 25°C and 50°C reactions was measured at SMU using a Finnigan MAT 252 or MAT 253 isotope ratio mass spectrometer and is reported here in per mil (‰) units relative to the PeeDee Belemnite standard (PDB; Craig, 1957).

Samples for magnetic polarity stratigraphy were typically collected by either drilling cores using a portable field drill with a non-magnetic diamond bit or as oriented block samples. We concentrated our sampling on hematitic siltstones and fine sandstones wherever possible, as previous work (e.g., Molina Garza et al., 1989; Molina Garza et al., 2000; Steiner, 2011) demonstrated that these rock types in this general area yield highly interpretable demagnetization data. In the case of field drilling, seven to 12+ independently-oriented core samples were typically obtained from each suitable bed at each locality. Core samples were processed into standard 2.2 cm high specimens for remanence and rock magnetic measurements. In the latter case, stratigraphic intervals lacking beds suitable for drilling or in sensitive areas (i.e. Palo Duro Canyon State Park), oriented blocks samples of suitable size were obtained by marking the orientation of any flat surface of a block that could be removed from the outcrop. At some sections, at least five oriented blocks were obtained from a single (<0.5 m) stratigraphic interval. At the Palo Duro Canyon section, oriented blocks were collected in succession, with sampling distance typically about 0.5 m. The oriented block samples were typically recut into multiple ∼8 cm³ cubic-like specimens using a non-magnetic diamond saw blade.

Remanence measurements were made on either a 2G Enterprise, DC SQUID, three-axis pulse cooled superconducting rock magnetometer, interfaced with an automated specimen handler and an on-line alternating field (AF) degausser system, or JR5A or JR6A AGICO spinner magnetometers. The principal magnetic phase in Whitehorse Group and Quartermaster Formation rocks is hematite and progressive thermal demagnetization was carried out and used an ASC TD48, dual zone thermal demagnetizer, involving 10–20 steps to maximum laboratory unblocking temperatures of 680°C (+/−). Selected, suitably indurated specimens were subjected to progressive chemical demagnetization (Collinson, 1975; Henry, 1979). To assure uniformity of leaching of authigenic hematite by reagent grade HCl, the surface area of the specimens was increased by making thin slices in the sides of the specimens with a fine, non-magnetic diamond saw. Specimens were incrementally leached over progressively longer amounts of time until the remanence remaining was less than 1% of the natural remanent magnetization (NRM) or until the specimen began to disintegrate to where it could no longer be measured. Acquisition of isothermal remanent magnetization (IRM) and backfield demagnetization of saturation IRM utilized an ASC multi-coil impulse magnet system. Three component thermal demagnetization of IRM acquired in different DC fields followed the method of Lowrie (1990). Anisotropy of magnetic susceptibility (AMS) determinations were made using a KLY-3S AGICO automated magnetic susceptibility device. All remanence measurements and demagnetization procedures, with the exception of chemical leaching, were performed in a magnetic shield (∼9 m x ∼7 m x ∼3 m) constructed by Lodestar Magnetics. Measurements of the bulk magnetic susceptibility as a function of continuous heating (to about 680°C)/cooling were made with an AGICO MFK1-A susceptibility instrument equipped with a CS-4 furnace apparatus. Additional details regarding methods utilized in this study may be found in Collins (2014) and Jackson (2016).

Results of progressive demagnetization were inspected using orthogonal demagnetization diagrams (Zijderveld, 1967) and the directions of magnetization components identified by the colinearity of several demagnetization data points determined using principal components analysis (Kirschvink, 1980), typically using the Remasoft software provided by AGICO. Magnetization directions at the site level were estimated as mean directions using as many independent observations as accepted, following the method of Fisher (1953). These estimated site-mean magnetization directions are transformed into virtual geomagnetic poles (VGPs), which then can be transformed into rotated VGP latitudes relative to a paleomagnetic pole position for North America for the time of the Permian-Triassic transition and plotted on stratigraphic columns to show the magnetic polarity stratigraphy of each section (Opdyke and Channell, 1996). Estimated paleomagnetic pole positions for the late Permian/Early Triassic for North America (e.g., Torsvik et al., 2012; their Table 2, APWP for Laurentia, corrected for inclination shallowing in sedimentary rocks: 260 Ma: 54.9 N, 125.4E, a95 = 3.5^o; 250 Ma: 53.1 N, 114.8 E, a95 = 13.6^o; 240 Ma: 53.2N, 115.3 E, a95 = 4.7^o; their Table 11, Global APWaP in North America coordinates: 260 Ma: 56.1N, 128.0 E a95 = 2.6^o; 250 Ma: 55.8N, 126.6 E, a95 = 3.6^o; 240 Ma: 55.5N, 118.1E, a95 = 3.6^o) are such that normal polarity magnetizations in Whitehorse Group and Quartermaster Formation strata in west Texas should have a north-northwest declination and shallow to moderate positive inclination.

U-Pb zircon CA-TIMS analytical techniques are described in detail in Mundil et al. (2004) and adaptations and calibrations pertaining to this study in Mitchell (2014). Zircons from sedimentary rocks and volcanic ash deposits were separated using standard mineral separation techniques. Selected zircon crystals were subsequently annealed at 850°C for 48 h and chemically abraded for 8–12 h at 220°C in HF (Mattinson, 2005) followed by HF dissolution and mixing with tracer solution, ion exchange chemistry (where indicated) and thermal ionization mass spectrometry (Mundil et al., 2004; Mitchell, 2014). The utilized tracer solutions were repeatedly calibrated against standard solutions and natural zircon standards at the Berkeley Geochronology Center (see Mundil et al., 2004; Mitchell, 2014 for methods and results). Additional calibrations against artificial age solutions made available by the EARTHTIME initiative suggest that our U-Pb zircon ages are accurate within the 1 permil level (for details see Irmis et al., 2011, Mitchell, 2014; Griffis et al., 2018). In addition, the ²⁰²Pb-²⁰⁵Pb-²³³U-²³⁵U spike mixed at the Berkeley Geochronology Center (BGC) was calibrated against replicate analyses of Temora 2 (collected by Black et al., 2004) zircon yielding a weighted mean of 417.69 ± 0.17 Ma (n = 10, MSWD 1.2, 0 rejects), which is consistent with the age of 417.43 ± 0.06 Ma (n = 11, MSWD = 0.8) reported by (Davydov et al., 2010) using the EARTHTIME 535 tracer. The 100 Ma EARTHTIME solution yielded a mean age of 100.247 ± 0.040 Ma (n = 7, MSWD 1.5, 0 rejects).

Mass spectrometric data were reduced using Isoplot (Ludwig, 2008) and Tripoli and U-Pb Redux (Bowring et al., 2011; McLean et al., 2016), and include corrections for procedural common Pb, spike Pb, mass fractionation, oxygen isotope interference for UO₂, and initial ²³⁰Th disequilibrium. Numerical age estimates for each sample are reported as weighted mean ²⁰⁶Pb/²³⁸U ages using the youngest subset with a permissible probability of fit (>0.05). Dispersion of ²⁰⁶Pb/²³⁸U ages suggests that residual Pb loss occurs (most notably in OCH12-7) and that slightly older crystals, predating the eruption age, are present in most samples. Analytically discordant ages (resolved from the Concordia curve including errors on λ²³⁵U and λ²³⁸U), as well as analyses with elevated common Pb (>3 pg) and elevated uncertainties (>2% on ²⁰⁶Pb/²³⁸U) are reported but excluded from the calculation of the mean. Ages are reported at the 2σ level. Results from U-Pb single zircon CA-TIMS analyses are summarized in Table 2 from the seven samples analyzed and presented in detail in Table 3.

The composite stratigraphic thicknesses of sections measured at the Palo Duro Canyon, Highway 207, Caprock Canyon, and Dickens localities (Figure 1) are 63, 28, 125, and 45 m, respectively. A comparison of rock types, bedding characteristics, sedimentary structures, and paleocurrent directions allows separation of the upper Whitehorse Group and Quartermaster Formation into salt-lake, sand-flat and fluvial facies belts (or associations). The depositional facies within each of the facies belts are summarized in Tables 4, 5, 6 and examples of all three facies belts exposed within an outcrop at the Caprock Canyons State Park section are shown in Figure 2 (see also Supplementary Figures S1–S5).

Sedimentary rocks of the salt-lake facies belt are present only in the upper Whitehorse Group strata, which are well exposed in the sections at Palo Duro Canyon, Highway 207, and Caprock Canyons (Figures 1–3). The salt-lake facies belt lies at the base of the Dickens stratigraphic section but is poorly exposed in areas of public access. The salt-lake facies belt consists of interbedded sandstones, mudstones, dolostones and evaporites deposited within a restricted basin (Figures 2, 3; Table 4). The most abundant and widespread salt-lake facies is red to orange, gypsiferous quartz sandstone beds ranging from a few centimeters to 1.1 m thick. Generally ripple cross-laminated, some beds are massive or exhibit normal graded thick laminations or symmetrical ripples (e.g., Figure 4A), and nearly all beds include dolomite or calcite cement. The red to orange sandstone represents subaqueous deposition of detritus via traction transport under both unidirectional and bi-directional currents. The red and orange coloration derives from hematite cements that preserve a primary remanence (Molina-Garza et al., 1989; 2000; also see below) and suggests that waters in the basin at the time of deposition were well oxidized.

The salt-lake facies belt also contains a mudstone facies composed of red beds ranging from 1 to 15 cm thick with occasional green to greenish blue-gray reduction mottles. Most mudstones are laterally continuous and thickly laminated to massive, although some mudrocks are drapes upon underlying ripples of the gypsiferous sandstone facies. Other mudstones preserve desiccation cracks that deform the upper surfaces of the beds. Most mudrock beds are weakly cemented by dolomite and calcite cements although a few beds are noncalcareous. The laterally continuous beds of thickly laminated to massive mudrock represent subaqueous deposition of detrital material via gravity suspension settling, whereas the mud drapes represent deposition during either subaqueous gravity-suspension settling during slack-water conditions or stranding of fine-grained sediments during waning-flow conditions. Mudrocks with desiccation marks (Figure 4B) represent initial subaqueous deposition via gravity suspension settling followed by subaerial drying and cracking. Like the gypsiferous sandstone facies, the red color of the mudrocks derives from early hematite cements that are demonstrated to retain a primary remanence and suggests well-oxidized conditions during deposition (Molina-Garza et al., 1989; 2000). The green to greenish blue-gray beds may represent deposition under anoxic conditions but the irregular distribution of this color in the strata suggests it is more likely that original hematite cements were reduced and removed during post-burial (e.g., Lewis and Byrne, 1996).

The bedded gypsum facies is only present in strata ∼20 m beneath the Whitehorse Group-Quartermaster Formation contact and consists of white to gray beds ranging from 0.5 to 2.6 m thick. Gypsum beds are internally organized into thickly laminated to thinly bedded layers with irregular cryptalgal upper bedding surfaces (Figure 4C). Most beds are laterally continuous, although a few examples from the Caprock Canyons section exhibit cryptalgal textures and domal bedding surfaces suggestive of microtopography up to 30 cm (Figure 4C). Occasionally upper bedding surfaces within the bedded gypsum facies are red to pink colored from comminuted hematite-cemented silt and very fine sand grains. All examples of the bedded gypsum facies are strongly calcareous. The bedded gypsum facies represents chemical sedimentation in restricted waters that were supersaturated with respect to CaSO₄·n(H₂O). The cryptalgal textures and domal bedding in these strata represent colonization of these substrates by bacteria or algae and development of stromatolites. Because there is no evidence of reworking and transport of the gypsum beds, they were likely deposited either in slack-water environments, although cryptalgal textures and domal structures associated with photosynthetic organisms necessitates deposition within the photic zone. The red to pink colored, hematite-cemented, silt and very fine sand grains comminuted with the gypsum represents gravity-suspension deposits possibly derived from eolian dust settling through the water column and furthermore indicates well-oxidized conditions during deposition. The calcareous cements associated with the bedded gypsum facies may have been co-deposited as an evaporite at times when solutions were supersaturated with respect to CaCO₃ or CaMg(CO₃)₂, implying decreased restriction in the basin. Alternatively, some of the calcareous cements may derive from dissolution of CaSO₄ and recrystallization as carbonate in the burial environment.

The dolostone facies occurs only in the Palo Duro and Highway 207 (Figure 3C) sections and consists of buff to gray micritic beds ranging in thickness from 9 cm to 1.2 m. Dolostone facies beds that are less than 0.2 m thick are internally massive. Dolostone beds thicker than 0.2 m typically exhibit massive structure near the base of the unit grading upward into undulating, very thin to thin, bedded domal structures with cryptalgal textures. Dolostone beds often include detrital quartz silt grains without hematite cement pigments. Internal to the thicker dolostone beds are stylolites oriented parallel to bedding with amplitudes up to 1 cm. At the Highway 207 section the uppermost dolomite bed includes an irregular upper contact that breaks internal bedding in the dolostone and defines a microtopographic surface that extends up to 90 cm beneath the uppermost surfaces of the dolostone bed (Figure 3C). In addition, the upper contact of the uppermost dolostone at the Highway 207 section exhibits mm-to cm-scale flutes and runnels. The dolostone facies represents chemical sedimentation in restricted marine waters that were supersaturated with respect to CaCO₃ or CaMg(CO₃)₂. The micritic texture of the carbonate in the dolostone facies reflects deposition in a low energy marine environment, and the cryptalgal textures indicate that this facies was deposited in shallow water, within the photic zone. Considering that detrital quartz silt in the dolostone facies does not include authigenic hematite cement and that hematite cement is ubiquitous in other facies of the salt-lake facies belt, it is possible that the dolostone facies experienced anoxic conditions during deposition or soon after burial that reduced and removed hematite from this facies in the early burial environment. The presence of stylolites is indicative of pressure dissolution in post-burial, diagenetic environments. However, those stylolites and other microtopographic surfaces, flutes and runnels at the upper contact of the uppermost dolostone at the Highway 207 section reflect exposure and dissolution by meteoric water (i.e., karst, Figures 3C,D) and could represent short-term exposure surfaces, also called intraformational disconformities.

The breccia facies is only exposed at the Highway 207 section and consists of angular, granule- to cobble-size, dolostone fragments in pockets and on the uppermost bed of the dolostone facies. Hematitic coarse silt to very fine sand fills the interstitial spaces between angular dolostone fragments (Figure 3D). The breccia facies represents the residuum from surface weathering and karstification of the exposure surfaces of the dolostone facies. Descriptively, similar facies have been identified in the middle Permian strata of Niger (Tabor et al., 2011; Smith et al., 2015).

The claystone facies is represented in the salt-lake facies belt at the Caprock Canyons and Palo Duro Canyon sections (Figure 5A). This facies is a single, internally massive bed at each locality. The claystone facies is a white, 12 cm thick, bed at the Palo Duro Canyon section and a red, 8 cm thick, bed at the Caprock Canyons section. Both of these facies are ∼18–20 m beneath the base of the Quartermaster Formation. The dominant mineral in these beds is analcime, although sanidine and zircon grains have been isolated from these beds for radioisotopic age determinations (Chang, 2008). The claystone facies is a gravity-suspension deposit of volcanic ash that settled through the water column, also referred to as a “tuff” or “tuffite.” The similar stratigraphic position of the claystone facies at Caprock Canyons and Palo Duro Canyon sections suggests that they are contemporaneous and representing the same volcanic event, although geochronologic data leave that correlation equivocal (see below). The hematitic color of the claystone facies at the Caprock Canyons section may be related to more intense chemical alteration during burial.

Sediments of the sand-mud flat facies (referred to as sand-flat facies elsewhere in this text) belt are exposed in the lower Quartermaster Formation, in particular this facies belt occurs at Palo Duro Canyon, Highway 207, Caprock Canyons and Dickens sections (e.g., Figures 3A–C). The stratigraphic thickness of the sand-flat facies belt is 35 m at Palo Duro Canyon, 11 m at Highway 207, 31 m at Caprock Canyons, and 32 m at Dickens. The sand-flat facies belt consists of sub-equal amounts of interbedded sandstones facies and mud/claystone facies across the study area (Table 5). However, the sandstone facies dominates the stratigraphy at the Palo Duro Canyon and Highway 207 sections, and is less important in the southern Caprock Canyons and Dickens sections.

The sandstone facies is composed of green and orange-to-red, very fine to fine sandstone. Most of the beds in the sandstone facies are thin, a few are 20–30 cm thick although some examples reach thicknesses of ∼3.5 m and they typically are intercalated with mudstone facies rocks. Some beds within the sandstone facies are remarkably uniform in thickness and lateral continuity. One ∼0.5 m thick, ledge-forming marker sandstone (Figure 2) about 13 m above the base of the Quartermaster Formation, has been traced for several kilometers across Caprock Canyons State Park. Ripple cross-laminated, asymmetric and symmetric ripples (e.g., Figure 4A) as well as lenticular and flaser bedding are common and thin (0.3–0.5 m) trough cross-bedded strata occur rarely in the sand-flat facies. The few paleocurrent indicators recorded from this facies belt indicate a generally northward sediment transport direction. Cubic salt casts up to 1 cm on edge are present in flaggy red sandstone beds 20–30 cm thick at ∼10 and ∼8 m above the base of the Quartermaster Formation at the Highway 207 and Caprock Canyons sections, respectively. All of the beds in the sandstone facies are calcareous. There are two different types of calcareous cements recognized in this facies: 1) a bladed calcite spar texture that occludes silt and sand grains and macroscopically forms spheroidal clusters (Figure 5B) and 2) micrite and microspar cements that incompletely fill the pore spaces between sand grains (Figure 5C). The bladed calcite texture is present in two horizons in both the Caprock Canyons and Dickens stratigraphic sections. However, X-ray diffraction analyses of samples with micrite and microspar cements indicate that both dolomite and calcite cements are present. The sandstone facies represents subaqueous deposition of detritus via traction transport under both unidirectional and bidirectional currents, although a prevailing northward transport direction is most prevalent. The uniform thickness and lateral continuity of many of these strata indicates smooth depositional surfaces with low-gradient slopes likely dipping toward the north side of the Palo Duro Basin, based on sediment transport directions. The cubic salt casts indicate crystallization of halite in the depositional or very early burial environment. The presence of halite salt casts in some these sandstone beds suggests that waters associated with deposition of these beds were extremely restricted and probably required evaporation of well over 90 percent of H₂O from interstitial or bottom waters to initiate crystallization of halite crystals (Borchert and Muir, 1964). Restricted conditions throughout the deposition of the sandstone facies are further supported by the presence of inorganic dolomite and calcite cements that grow into primary void spaces between grains (Figure 5C), indicating waters that are supersaturated with respect to these minerals in early burial. Collectively, the sandstone facies correspond to shallow subaqueous sheet flows of the sand flat facies belt (e.g., Dalrymple et al., 1992). The bladed calcite cement texture that occludes silt and sand grains in the sandstone facies (Figure 5A) is evidently of a different generation of carbonate growth from those of the micrite and microspar cements, and may represent a later diagenetic phase of cementation. Bladed textures similar to those in the sandstone facies are interpreted as late meteoric (Minero, 1988) and late-burial (Mazullo, 1986; Sellwood et al., 1989) cements in other sedimentary successions.

The red mud/claystone facies is composed of dusky-red mudstone from 1 cm to ∼5 m thick. Mudstone facies beds are laterally continuous and most are internally massive although thin beds of finely laminated mudstone are present. Some mudstones preserve desiccation cracks that deform the upper surfaces of the beds (e.g., Figure 4B). The upper half of the sand-flat facies belt at the Caprock Canyons and Dickens sections includes red mud/claystone facies beds that preserve arcuate slickenplanes that cross-cut mudstone beds up to 30 cm thick. The mudrock facies likely represents subaqueous deposition of detritus via gravity suspension settling or stranding of fine-grained sediment during waning-flow conditions. Mudrocks with desiccation marks represent initial subaqueous deposition via gravity suspension settling followed by subaerial cracking of the mudrocks. The mudrock strata that preserve arcuate slickenplanes represent subaerial exposure and incipient soil development (vertic Protosols of Mack et al., 1993; Tabor et al., 2017). Collectively, the mudrock facies of the sand flat facies belt corresponds to the slack-water and episodically exposed parts of a sand-flat depositional system (e.g., Dalrymple et al., 1992).

The biotite-rich claystone facies lies in the sand-flat facies belt at the Highway 207, Caprock Canyons, and Dickens stratigraphic sections. The biotite-rich claystone facies is at 3.6 and 23.6 m above the Whitehorse-Quartermaster contact at Dickens, 5.8 and 25.1 m above the Whitehorse-Quartermaster contact at Caprock Canyons, and 2.7 m above the Whitehorse-Quartermaster contact at Highway 207. The lower beds of this facies are ∼10 cm thick at both Dickens and Caprock Canyons sections, whereas the upper beds are ∼10 cm thick at Dickens and ∼30 cm thick at Caprock Canyons. The single occurrence of this facies at Highway 207 is also a ∼10 cm thick bed. The biotite-rich claystone facies is typically organized into thin beds that are internally massive, although relict ripple cross laminations are occasionally preserved in upper beds of this facies. Biotite grains up to fine sand size are concentrated at the base of these beds and along the bases of ripple-cross laminated zones. As in the claystone facies of the salt-lake facies belt, the dominant mineral in these beds is analcime. Sanidine and zircon grains have been separated from these beds (Chang, 2008; Mitchell, 2014). The biotite-rich claystone facies represent a gravity-suspension deposit of volcanic ash that settled in the shallow subaqueous channels and runnels of the sand flat facies belt. Relict ripple cross laminations in the upper parts of these beds indicate local reworking by subaqueous currents. The similar stratigraphic positions of this facies in the Dickens and Caprock Canyons stratigraphic sections suggests they are contemporaneous deposits from the same volcanic eruption, although geochemical data for the biotite and zircon grains from this facies are unable to demonstrate whether they truly are contemporaneous deposits between locations (Chang, 2008; Mitchell, 2014; see also discussion below). It is unclear if the red mud-claystone facies that is exposed in the Highway 207 section correlates to exposures of this facies at the Dickens and Caprock Canyons sections. However, based on magnetic polarity stratigraphy data from the study sections (see below), the biotite-rich claystone at Highway 207 likely correlates with the lower occurrences of this facies in the sand-flat facies belt in Caprock Canyons and Dickens sections.

A remarkable exception to the character of the sand-flat facies belt in the Quartermaster Formation is evidence for macroscopic life. To date, neither macrofossils nor fossil traces have been observed in this facies belt. However, microscopic phytoliths including ones with organic cores have been identified in samples from the sand-flat facies belt in the Caprock Canyons stratigraphic section (Figure 5D). This suggests that plants existed in the region, but they did not have a substantial impact on the lithostratigraphy of the sand flat facies belt as in other places and times that include well-developed rooted horizons reflective of saltmarshes and other vegetated sand- and mud-flat landscapes (e.g., Dalrymple et al., 1992).

Sediments of the fluvial facies belt are exposed in the Caprock Canyons and Highway 207 sections (Figures 1, 2, 3B and Table 6) and conformably overlie sand flat facies belt rocks at those locations. The fluvial facies belt is ∼45 m thick in the Caprock Canyons section and is disconformably overlain by mudrock and sandstone strata of the Upper Triassic Dockum Group (Figure 2). Basal Dockum Group rocks lie 78 m above the base of the Quartermaster Formation at Caprock Canyons and are defined by a polymictic mudstone that includes extraformational detrital chert granules and pebbles. The fluvial facies belt is at least 15 m thick in the Highway 207 section but its complete thickness could not be measured because the upper contact with the Dockum Group is located on private land, and in this case was not able to be accessed. The fluvial facies belt is absent at Dickens and Palo Duro Canyons sections, perhaps due to erosion prior to deposition of the Upper Triassic Dockum Group (Figure 1), or that the fluvial facies, perhaps being an axial sedimentary system across the Quartermaster landscape, was never deposited in those regions.

The clinoform sandstone facies is composed of red-to buff-colored, fine to medium sandstone (Figures 2, 3B). Clinoform sandstone units are thick-bedded and are internally organized into trough-cross bed (Figure 4D) or plane-bed structures (Figure 3B). Plan-view exposures of the trough-cross bedded sandstones indicate northward transport. Some sandstone beds exhibit horizontal laminae with parting lineations that also indicate a dominantly northward transport direction. Thick clinoform sandstone beds are separated by drapes of thin bedded to laminated red mudstone layers that drape upon sandstone beds that imply a channelized structure. Channelized sandstones form both single and multistory sets that correspond to channel topographies of up to 7 m (Figure 3B). The bases and sides of these channel structures occasionally are lined by red mudstone clasts ranging from pebble to boulder size. Most sandstone beds are calcareous although a few noncalcareous strata lie in the upper ∼15 m of sandstone fluvial facies in the Caprock Canyons section. The upper 6 m of the Quartermaster Formation at Caprock Canyons section include permineralized axes, branches, and fallen trunks of fossilized trees (Figure 4D). Similar to sandstones in the salt-lake and sand flat facies belts, petrographic analysis shows that carbonate cements incompletely fill the pore spaces between sand grains (e.g., Figure 5C). X-ray diffraction analyses indicate that calcite is the dominant cement and dolomite cement is not present in this facies. The sandstone facies represents subaqueous deposition of detritus transported via unidirectional traction transport directed northward in meandering fluvial channel systems. Inclined mudstones separating clinoform sandstones (e.g., Figure 3B) represent deposition during slack-water conditions (standing water) or waning flow. Desiccation cracks are not present on the upper surfaces of the inclined mudstones suggesting that slack-water conditions or subaqueous waning flow may be more appropriate for the conditions of mud-drape deposition. The red mudstone clasts lining the bases of some channel forms are eroded blocks from either mud drape deposits within the channels or from overbank mudstones. The clinoform sandstone facies are hematite-rich, suggesting well-oxidized conditions during deposition. The dominance of calcite cement in the clinoform sandstone facies may indicate less restricted/evaporitic pore waters in the early-burial environment than those associated with the salt-lake and sand flat facies belts. The preservation of fossil trees in the uppermost clinoform sandstones (Figure 4D) is important, because it is the first direct evidence of macroscopic life in the Whitehorse Group-Quartermaster succession. However, microscopic organic matter associated with biogenic silica (i.e., phytoliths) have been recovered from samples throughout the fluvial facies (Figure 5D).

The mudstone facies is composed of dark red mudstone beds from a few cm to ∼30 cm thick. Mudstone facies beds are laterally continuous and organized internally as laminated, thin-bedded, or massive to angular blocky. The angular blocky mudstones are organized into wedge-shaped aggregates that define arcuate slickenplanes that cut across most of the thickness of the unit. The mudstone facies makes up only a small part of the fluvial facies belt; it is rare in the lower half, but more common in the upper half of the fluvial facies belt succession in the Caprock Canyons stratigraphic section. Most mudstones are slightly calcareous although noncalcareous mudstone beds lie near the top of the fluvial facies belt stratigraphic succession at Caprock Canyons. X-ray diffraction analysis of mudstone facies samples indicates that carbonate cements are dominated by calcite; dolomite is not present in the mudstone facies. The mudstone facies represents interchannel, perhaps floodplain or medial bar, depositional environments in which fine-grained detrital clastic material was stranded subsequent to flood-stage conditions. The blocky mudstones with slickenplanes are Vertisols (i.e., paleosols; Mack et al., 1993; Tabor and Myers, 2015) and reflect seasonal wetting and drying of the interchannel floodplains. The dominance of calcite cements in the mudstone facies may represent interaction with rainwater and groundwater of lower salinity, perhaps meteoric, than strata in the underlying sand-flat and salt-lake facies belts where dolomite was the dominant carbonate cement. The mudstone facies contains abundant authigenic hematite cement and indicates oxidizing conditions during deposition, pedogenesis, and early burial.

The calcrete facies is present in the fluvial facies belt of the Caprock Canyons stratigraphic section and occurs at only one stratigraphic level, ∼70 m above the base of the Quartermaster Formation. This horizon is ∼2 m beneath the part of the stratigraphy that preserves permineralized fossil plants and ∼8 m beneath the Quartermaster-Chinle unconformity. The calcrete is composed of light-yellow to buff colored, massive, fine to medium sandstone that is irregularly cemented by calcite in a laterally continuous, ∼30 cm thick horizon. Calcite cements are organized into 1) vermicular patterns with internal micritic calcite texture and 2) tubular micritic calcite up to 2 cm in diameter and 10 cm long oriented in vertical to subvertical patterns. The calcrete facies represents pedogenesis of an abandoned channel deposit. The vermicular cements with internal micritic calcite texture may represent either groundwater-derived crystallization (e.g., Semeniuk, 1985) or the depth of wetting and leaching of calcite in a well-drained soil. The vertical to subvertical calcite-cemented tubules are cementation after roots (i.e., rhizoliths; Klappa, 1980), which supports an interpretation of the vermicular carbonates as a product of the depth of leaching in a well-drained soil. This reflects the fact that most plant root systems do not develop vertical root structures in poorly drained soils (e.g., Pfefferkorn and Fuchs, 1991). The presence of rhizoliths in the calcrete facies is important because it represents the first trace evidence for macroscopic terrestrial life in the Quartermaster Formation.

X-ray diffraction analysis of the 001 d(hkl) peaks for dolomite and calcite in bulk samples indicate that many Whitehorse Group and Quartermaster Formation samples contain both types of carbonate cements, although some samples contain only dolomite or calcite. Cement concentrations of the collected samples range from 32–100% dolomite in the Palo Duro Canyon section, 26–100% dolomite in the Caprock Canyons section, and 38–100% dolomite in the Dickens section whereas the Highway 207 section is 100% dolomite. With the exception of the Highway 207 section, which contains only dolomite cement throughout the exposures, there is a stratigraphic trend in the distribution of carbonate mineralogy wherein cements from the salt-lake facies belt are dominated by dolomite, cements in the sand flat facies belt are mixed dolomite and calcite with increasing proportions of calcite upward through the stratigraphy, and cements in the fluvial facies belt are dominated by calcite. Note that these weight percent cement values are not tabulated in this manuscript, but are available to any interested parties upon request.

Based on yields of CO₂ generated by the differential reaction times of the carbonate cements with H₃PO₄ (Al-Aasm et al., 1990), the weight-percent of calcite and dolomite cements in samples from the (1) Palo Duro Canyon section range from 1.8 to 38.0 calcite and 1.0 to 21.5 dolomite, (2) Caprock Canyons section range from 0.1 to 27.7 calcite, and 1.0 to 29.7 dolomite, (3) Dickens section range from 0.1 to 15.7 calcite and 1.0 to 10.7 dolomite. Weight percent dolomite in samples from the Highway 207 section range from 1.9 to 20.9.

Petrographic and x-ray diffraction analysis of samples from the four composite stratigraphic sections suggests that dolomite is likely the dominant syn-depositional and early post-depositional cement in the salt-lake and sand flat facies belts, whereas calcite may have precipitated early in the fluvial facies belt. Because it is likely that calcite cements throughout these stratigraphic successions reflect post-depositional diagenetic conditions, an emphasis was placed on isotopic analysis of dolomite cements in order to develop a carbon-isotope chemostratigraphy as a means of correlation among sections, and to assess whether there is a substantial negative δ¹³C excursion of the carbonate cements, as has been found to characterize Permian-Triassic boundary stratigraphic successions around the world (e.g., Korte and Kozur, 2010).

Zircon crystals extracted from seven volcanic ash deposits (tephras) exposed in five sections within the salt-lake and sand flat facies belts were subjected to CA-TIMS (Chemical Abrasion-Thermal Ionization Mass Spectrometry) analysis. The populations of zircon crystals yield indistinguishable U-Pb zircon ages with 0.1% level precision and these cluster around 251.9 Ma (Figure 6, Tables 3,4, and Mitchell 2014). The biostratigraphically defined marine Permian-Triassic boundary (FAD Hindeodus parvus) has been interpolated to 251.902 ± 0.024 Ma at the type-section for the global stratotype and point (GSSP) in the Meishan section, Zhejiang Province, China, and the abrupt decline in δ¹³C_carb slightly prior to 251.950 ± 0.042 Ma lasting about 2.1–18.8 ka (Burgess et al., 2014). Within this context, there is an explicit expectation that the Permian-Triassic boundary should be present within the strata of the upper Whitehorse Group/Quartermaster Formation sequence and, furthermore, that the fluvial facies belt of the Quartermaster Formation is earliest Triassic in age, rather than Permian as has long been described and mapped up to this time. Considering that a significant negative shift in stable carbon isotope values of carbonate (e.g., Magaritz et al., 1988; Magaritz, 1989) as well as organic matter (Magaritz et al., 1992) appear to be a diagnostic feature of the chemostratigraphy of the Permian-Triassic transition, and that there is relatively well-documented global geomagnetic polarity time scale for the time interval that spans the Permian-Triassic boundary (e.g., Steiner, 2006; Henderson et al., 2012; Ogg, 2012; Szurlies, 2013; Hounslow and Balabanov, 2018; Pan et al., 2021), we consider below the stable carbon isotope chemostratigraphy of dolomite cements, as well as the magnetic polarity stratigraphy record from the upper Whitehorse Group and Quartermaster Formation strata to estimate more accurately the location of the Permian-Triassic boundary within these strata. We chose to use only the data from dolomite cements for this exercise of stable carbon isotope chemostratigraphy, because of the petrographic support (e.g., Figure 5C) for early formation of dolomite in the salt-lake and sand flat facies belts, given the likelihood that the Permian-Triassic boundary is most likely to reside in those facies belts that host the volcanic ash deposits (tephra/tuff) providing the U-Pb age determinations. We do not consider the stable carbon isotope data from calcite cements any further in this work, because of the petrographic evidence for these phases often including a secondary origin and a substituent cement in the salt-lake and sand flat facies. Calcite only appears as a primary cement in the upper parts of the sand flat facies and throughout the overlying fluvial facies belt.

Response to thermal demagnetization by the hematitic red beds we sampled indicates that the NRM in these rocks is readily interpretable and usually consists of one vector component or, at most, two vector components resolved during thermal unblocking (Figure 7). Samples from some sites contain a low intensity and relatively unstable first-removed component that is completely unblocked by ≈ 400°C. This component is typically north-directed and of steep positive inclination, consistent with the present local field. The stable endpoint component, with a direction that is northwest to north-northwest and shallow positive inclination, is isolated above 400°C to complete unblocking at about 680°C. The laboratory unblocking temperature spectrum for this component is typically very discrete, with a sharp decrease above 650°C. Typical NRM intensities are about 10 mA/m, with a few more strongly magnetized horizons having intensities of about 200 mA/m.

Response to other demagnetization methods further demonstrates that hematite is the principal remanence carrier in these rocks. AF demagnetization followed by thermal demagnetization treatment successfully randomizes the north directed low laboratory unblocking temperature remanence, leaving the ChRM of north-northwest declination and shallow positive inclination. This first removed component is likely a viscous remanent magnetization, at least in part residing in magnetite, again reflecting the modern geomagnetic field. Specimens initially subjected to AF demagnetization do not exhibit the relatively erratic behavior at low laboratory unblocking temperatures shown by those only thermally demagnetized. The first removed component in thermal demagnetization is also likely removed in AF demagnetization and is carried by the same assemblage of low coercivity grains. In progressive chemical demagnetization, specimens with strong calcite cement completely disaggregated after several seconds to minutes of leaching. On the other hand, those with considerable dolomite and quartz cement responded favorably to leaching, with all authigenic (pigment) hematite completely dissolved and specimens bleached uniformly white after about 12 total hours of leaching (Figure 8). Chemical demagnetization resulted in a decrease in intensity of about 50 percent with leaching times up to about 6 h. After substantial leaching of pigment hematite, a considerable fraction of the NRM intensity persisted after 10 steps (≈15 h) of treatment, implying that both disseminated hematite pigment and detrital hematite grains carry much of the ChRM in the Quartermaster strata. Notably, the direction of the remanence unblocked over a temperature interval well below about 580°C, and likely carried in detrital magnetite, is identical to the dominant remanence carried by hematite.

Progressive acquisition of isothermal remanent magnetization (IRM) data are consistent with the presence of both hematite and a low coercivity magnetic phase. Each specimen reached full saturation at about 2.0–2.9 T indicating hematite as the significant IRM carrier (Figure 9A). IRM curves are concave down, however, and reveal a low coercivity phase that is largely saturated by about 100 mT. Acquisition curves below ∼100 mT are steep positive, and the rate of IRM increase decreases as the field approaches ≈300 mT. The IRM acquisition data are consistent with demagnetization data that reveal a low coercivity phase, likely detrital magnetite, and both hematite pigment and detrital hematite in these rocks. Backfield measurements yield coercivity of remanence values between 100 and 140 mT.

Thermal demagnetization of three-component IRM data (Figure 9B) typically show a small drop of intensity of the intermediate component beginning around 200°C and ending by 300°C, as well as a small drop in intensity of all components around 580°C indicating a range of magnetite chemistry and grain size. Also of interest in these plots is a common crossover of intensity by the maximum and intermediate components, with both components eventually unblocked by 680°C. This likely indicates a strong contribution of coarse grained (detrital) hematite to the remanence of these rocks.

Anisotropy of magnetic susceptibility (AMS) data from collections obtained from selected northwest Texas localities sampled in this study yield a well-defined fabric that is typical of fine-grained detrital sedimentary rocks that have preserved an original depositional fabric (Figure 9C). Minimum susceptibility axes are clustered in a near vertical orientation with maximum and intermediate susceptibility axes generally scattered along the perimeter. The anisotropy parameter P (degree of anisotropy, P = Kmax/Kmin) shows little to minor dependence on bulk susceptibility (Km) for each collection. The AMS shape parameter T (T = [lnF-lnL/lnL + lnF]) shows a consistent oblate fabric, not dependent on the magnitude of the P parameter.

The monitoring of bulk rock magnetic susceptibility as a function of continuous heating and cooling (Figure 9D) reveals the presence of only a very minor concentration of a cubic magnetic phase (i.e. magnetite and/or maghemite of inferred detrital origin) and the dominance of hematite, with the clear resolution of a phase losing magnetic susceptibility by about 680°C. For experiments performed in air, the heating/cooling curves are irreversible, with an increase of typically about 50 percent in bulk susceptibility upon cooling, indicating the formation of new magnetic material in the heating process. In addition, there is only slight evidence of the presence of a cubic phase. For experiments in argon gas, on the other hand, the heating curves show a slight Hopkinson peak below about 550°C. Upon cooling, there is a 3–4 fold increase in bulk susceptibility, and a very well-developed Hopkinson peak, indicating the production of a cubic phase of higher susceptibility. The results of all of the rock magnetic experiments described above, supporting the presence of both abundant authigenic (pigmentary) hematite, as well as a mixture of hematite and magnetite as detrital phases, are supported by and consistent with petrographic observations.

In principal components analysis (PCA) (Kirschvink, 1980) the last several (at least three and usually six or more) measurement steps were used for PCA, as identified from the most well-resolved segment of the orthogonal demagnetization plot (Zijderveld, 1967). PCA directions from specimens from stratigraphic levels of normal polarity are generally north-northwest seeking (∼335°) with shallow positive inclination (∼+20°). Horizons of reverse polarity are characterized by southeast seeking declinations (∼155°) with shallow negative inclinations (∼−10°). Specimen directions from individual stratigraphic levels (sites) show minimal to minor dispersion of ChRM directions and are usually sufficiently well-grouped to provide polarity determinations.

Virtual geomagnetic poles (VGPs) were calculated for each stratigraphic layer sampled using the estimated site mean direction from individual layers (beds). Before calculating the VGP, each individual site mean direction was corrected for inclination shallowing with a flattening factor of f = 0.59 which is commonly assumed to be valid for most hematitic red beds (Bilardello and Kodama, 2010; Domeier et al., 2012). A formation mean direction calculated from each sampling site (reverse sites transposed through the origin) is D = 340.0° I = + 26.0° k = 21.18 α95 = 5.8° n = 30 distinct, independent stratigraphic layers. VGPs yield a grand mean paleomagnetic pole located at pLat = +62N, pLon = 123E, A95 = 5.8^o. These results are slightly higher latitude than previous paleomagnetic pole work on Whitehorse Group/Quartermaster Formation strata (Molina-Garza et al., 1989; 2000). These two studies, however, did not correct results for potential inclination shallowing. The estimated paleomagnetic pole is also at a slightly higher latitude than paleomagnetic poles for the North America apparent polar wander path for the latest Paleozoic (Van der Voo, 1993; Domeier et al., 2012; Torsvik et al., 2012). If specimens with a larger mean angular deviation (MAD, >20°) are excluded from this analysis, the formation mean direction, corrected for inclination shallowing, is D = 340.1° I = +16.2° k = 20.42 α95 = 6.2° n = 28 horizons, providing a slightly revised paleomagnetic pole at pLat = +61N, pLon = 122E, A95 = 5.8^o. The estimated mean inclination of about 16^o corresponds to a paleolatitude of about +8^o.

These data permit an estimate of composite polarity sequences for the Caprock Canyon, Palo Duro Canyon, Highway 207, and the Clarendon sections. The Palo Duro sequence is defined by a relatively thick normal polarity magnetozone beginning at the first exposed level that was capable of being sampled (1.9 m) above the top of the last massive bedded gypsum in the uppermost Whitehorse Group, which is exposed at the base of the sampled section. The normal magnetozone is followed by a relatively narrow stratigraphic interval of reverse polarity, with the remaining stratigraphic section defined by another normal polarity magnetozone, and thus a N-R-N inferred sequence. Samples from the Caprock Canyons section define a reverse polarity magnetozone at the base of the sampled section, which is in the upper part of the bedded massive gypsum sequence of the Whitehorse Group, followed by a relatively thick normal polarity magnetozone in the remainder of section where we were able to collect material for measurement. For the Caprock Canyons section we were able to sample several meters lower in Whitehorse Group strata than at the Palo Duro section. However, our sampling at Caprock Canyon did not continue far enough up section to identify the reverse polarity magnetozone found at the Palo Duro section. At the Highway 207 section, the uppermost part of the Whitehorse Group section that is exposed is of reverse polarity. The remainder of the exposed section, in the Quartermaster Formation, is entirely of normal polarity, and this includes horizons immediately above and below the volcanic ash deposit at that locality. We also sampled the Clarendon section (Figure 1) for geochronologic and magnetic polarity stratigraphy purposes. Similar to the Highway 207 section, the lower part of the Clarendon section exposes the uppermost part of the Whitehorse Group, with abundant karstification features, which is of reverse polarity. The rest of the exposed section is in the lower Quartermaster Formation, and the lower part of the formation, which contains the volcanic ash deposit referred to Table 2, is largely of normal polarity, with the very uppermost part of the exposed section being of reverse polarity.

In total, 260 analyses of dolomite cement δ¹³C and δ¹⁸O values were completed: 89 from the Palo Duro Canyon section (Table 7), 27 from the Highway 207 section (Table 8), 91 from the Caprock Canyons section (Table 9), and 53 from the Dickens section (Table 10). There is no clear parametric correlation between dolomite δ¹⁸O values and corresponding dolomite δ¹³C values among the sections (Figure 10).

Dolomite samples from the Palo Duro Canyon section were collected from rocks of the salt-lake facies belt (n = 58) and the sand-flat facies belt (n = 31). Dolomite δ¹³C and δ ¹⁸O values range from −8.0 to 3.2‰ and −6.6 to 3.1‰, respectively (Figures 10, 11). There is a shift in δ¹³C of dolomite from isotopically heavy values (∼1–4‰) near the base of the section to more negative values (∼−4‰) over a short stratigraphic distance within the Whitehorse Group strata (Figure 11A). Above the negative excursion, dolomite δ¹³C values remain between ∼−4 and −2‰ through the upper Whitehorse Group. Dolomite δ¹³C values shift toward generally more negative values, ranging from −4 to −8‰, across the upper boundary of the Whitehorse Group and through the Quartermaster Formation (Figure 11A). Dolomite δ¹⁸O values exhibit substantial, and somewhat erratic, stratigraphic variation between -6.6 and 1.7‰ in the Whitehorse Group at Palo Duro Canyon section (Figure 11B). However, there is a trend toward more positive dolomite δ ¹⁸O values (from −5 to 3‰) upward through the strata of the Quartermaster Formation in the Palo Duro section.

Dolomite δ¹³C and δ¹⁸O values from the Highway 207 section range from −10.3 to −1.0‰ and −4.0 to 2.8‰, respectively (Figures 10, 12). There is a negative carbon isotope shift from values of ∼−1‰ to ∼−8‰ over a short stratigraphic distance in the upper 2.7 m of the Whitehorse Group (Figure 12A). Moving into the Quartermaster Formation, there is an abrupt stratigraphic shift toward more positive values of ∼2‰ that is stratigraphically followed by a gradual upward trend toward more negative values of ∼−10‰ at the top of the Highway 207 section. With the exception of the one sample directly beneath the karsted surface exposed at the Highway 207 section, dolomite δ¹⁸O values exhibit minor stratigraphic variation and range from −4 to −1.5‰ through the salt-lake, sand-flat, and fluvial facies belts in the Whitehorse Group and Quartermaster Formation (Figure 12B). The dolomite sample directly beneath the karst surface has a more positive value than other samples in the Highway 207 stratigraphic section, at 2.8‰.

Dolomite δ¹³C and δ¹⁸O values from the Caprock Canyons section range from −10.9 to −2.8‰ and −4.9 to 3.7‰, respectively (Figures 10, 13). There is a large negative carbon isotope excursion from dolomite δ¹³C values of ∼−3 to ∼−11‰ in the lower 7.4 m of the Whitehorse Group strata (Figure 13A), and dolomite δ¹³C values in the overlying Whitehorse strata vary between ∼−4 and ∼−3‰. There is another negative carbon isotope shift (to values ∼−5‰) moving upward from the upper Whitehorse Group into the lower ∼10 m of strata in the Quartermaster Formation, above which there is a negative carbon isotope trend from −5‰ to ∼−9‰. Beginning ∼20 m above the base of the Quartermaster Formation, the negative trend reverses and moves toward more positive δ¹³C values in the upper 30 m of the Caprock Canyons section (Figure 13A). Above this is a general trend toward more positive δ¹³C values in the basal ∼15 m of the fluvial facies belt. Dolomite is not present in strata above the lower 15 m of the fluvial facies belt (Table 9). Like dolomites from the Palo Duro section, dolomite δ ¹⁸O values in the Whitehorse Group of the Caprock Canyons section exhibit substantial variation, ranging between ∼−5 and 1‰ (Figure 13B). Dolomite δ¹⁸O values in the overlying Quartermaster Formation exhibit an upward stratigraphic trend toward more positive values from ∼−5 to ∼4‰.

Dolomite δ¹³C and δ¹⁸O values from the Dickens section range from −8.5 to −3.3‰ and −6.0 to −0.5‰, respectively (Figures 10, 14). The δ¹³C values range from ∼−3.5 to −5‰ in the upper Whitehorse Group (Figure 14A). There is a negative δ¹³C trend from ∼−4 to ∼−7‰ in the lower ∼5 m of the Quartermaster Formation, above which there is an abrupt shift toward more positive values of ∼3‰ over ∼1.1 m. This positive excursion is followed by a gradual trend (over ∼27 m) toward more negative δ¹³C values. Dolomite δ¹⁸O values in the upper Whitehorse Group exhibit substantial variability (−6.0 and −2.4‰; Figure 14B). In the overlying Quartermaster Formation, δ¹⁸O values show an upward trend toward isotopically more positive values, ranging from ∼−4 to ∼−0.5‰, with a positive excursion of 3.8‰ about 5.5 m above the base of the formation.

The lack of a parametric correlation among microspar dolomite cement δ¹³C and δ¹⁸O values in the salt-lake, sand-flat and lower fluvial facies belts, in conjunction with lack of petrographic evidence for different generations of dolomite cementation, suggests that dolomite δ¹³C and δ¹⁸O values in the Whitehorse Group and Quartermaster Formation may faithfully reflect early-burial environmental conditions and may be used to provide a chemostratigraphic correlation among the different stratigraphic sections in this study.

A weighted mean ²⁰⁶Pb/²³⁸U age of 251.95 ± 0.17 Ma on 25 zircons for the lower ash fall deposit collected from the uppermost part of the Whitehorse Group in the Palo Duro Canyon section, and zircons in the lower (n = 10) and upper (n = 7) tuffs in the Quartermaster Formation from Caprock Canyons sections yield ages of 252.19 ± 0.30 and 252.14 ± 0.43 Ma, respectively (Tables 2, 3; Figures 11, 13, 14). Similar, and statistically indistinguishable, zircon crystal U-Pb age estimates have been obtained from ash fall deposits in the Clarendon section (251.92 ± 0.31 Ma), in the lower (251.76 ± 0.20 Ma) and upper (251.80 ± 0.2316 Ma) Dickens section as well as the Highway 207 (251.74 ± 0.28 Ma) section (Tables 2, 3 and Figures 12, 14).

Weighted mean ²⁰⁶Pb/²³⁸U ages on zircon crystals from volcanic ash deposits intercalated with marine sedimentary strata from southern China (GSSP in Meishan) date the biostratigraphically defined Permian-Triassic boundary at 251.902 ± 0.024 Ma (Burgess et al., 2014). Assuming that the U-Pb zircon age constraints from the ash fall deposits in the upper Whitehorse Group and lower Quartermaster Formation represent true depositional events (in the sense that the zircon crystallization age closely approximates the age of deposition), the ages of these ashes, tightly clustering around 251.9 Ma (Tables 2,3), are analytically indistinguishable from that of the age constraints for the end-Permian extinction event, Permian-Triassic boundary, and δ¹³C_carb excursion that are well documented in southeast China (e.g., Burgess et al., 2014). Additionally, the end-Permian extinction event is characterized by a large and abrupt decrease in the δ¹³C values of carbonate and organic matter formed in both marine and terrestrial environments. Although the magnitude of the negative δ¹³C shift varies from ∼3–4‰ in open marine biogenic calcite (e.g., Gruszcynski et al., 1989) to greater than 20‰ in terrestrial organic matter (Krull et al., 2000; Krull and Retallack, 2000), the shift itself is a global phenomenon, as it has been documented in at least 40 different boundary sequences around the world (Korte and Kozur, 2010). This prominent negative excursion in δ¹³C has even been used to identify the approximate location of the PTB in strata that are not datable by other means (Holser and Magaritz, 1985; 1987; Holser et al., 1989; 1991; Magaritz, 1989; Magaritz et al., 1988; 1992). More recent high-resolution studies of marine carbonate δ¹³C values from southern China demonstrate that a negative δ¹³C excursion began prior to the end-Permian marine extinction event with sustained and more negative δ¹³C values continuing in post-extinction, Lower Triassic strata (Lou et al., 2011).

Strata of the Whitehorse Group and Quartermaster Formation sampled at five different section localities in northwest Texas yield well-defined magnetization components of NNW and shallow positive inclinations, and their antipodes, which we interpret as being of normal and reverse polarity, respectively. These magnetization components are mainly carried by hematite of both detrital and authigenic origin. Notably, the stratigraphic intervals within which the prominent ash fall deposits, which have yielded high precision U-Pb zircon age data consistent with their emplacement at the time of the Permian-Triassic boundary, are either directly identified to be of normal polarity or are closely bounded by normal polarity magnetozones (Figure 15). At face value, based on the high precision U-Pb zircon age data, the pattern of terrestrial sedimentation in western Pangea preserved in strata of the Whitehorse Group and Quartermaster Formation, could be essentially time equivalent to the pattern of marine sedimentation and extinction, which, based on magnetic polarity data from the Meishan and Shangsi sections in SE and Central China, respectively, is thought to have occurred within a normal polarity chron of about 700 ka duration (Li and Wang, 1989; Yin et al., 2007; Szurlies, 2013, Glen et al., 2009; Huang et al., 2011; Pan et al., 2021). At the Palo Duro Canyon, Dickens, and Clarendon sections we have identified reverse polarity magnetozones above the normal polarity magnetozone containing the studied ash fall deposits, and at the Caprock Canyons, Highway 207, and Clarendon sections, we have identified reverse polarity magnetozones below this normal polarity magnetozone. On this basis, there is a plausible means to correlate these sections to proposed global magnetic polarity time scales and/or specific sets of polarity data for the time interval across the Permian-Triassic boundary (Steiner, 2006; Hounslow and Muttoni, 2010; Ogg, 2012; Henderson et al., 2012; Szurlies, 2013; Hounslow and Balabanov, 2018; Glen et al., 2009; Huang et al., 2011; Pan et al., 2021; Figure 15). Many magnetic polarity time scales referenced here differ in their placement of the Permian-Triassic boundary in the chron record and in the approach used to develop each polarity time scale, as discussed in Gastaldo et al. (2021), Hounslow and Muttoni (2010) and Hounslow and Balabanov, 2016) indicate the Permian-Triassic boundary at the very base of a ∼700 ka duration chron of entirely or almost entirely normal polarity. These two estimates of the magnetic polarity time scale, as well as that of Steiner (2006) represent a composite of a number of magnetic polarity stratigraphic sections from numerous localities, some of which have more robust age control than others. On the other hand, Ogg (2012), Henderson et al. (2012), and Szurlies (2013) place the Permian/Induan boundary, and the span of marine extinction events, within a comparable ∼700 ka duration normal polarity chron. Notably, the magnetic polarity time scale of Szurlies (2013) is based on data from the largely non-marine Germanic Basins and has complete stratigraphic superposition. The period-level time scales of Ogg (2012) and Henderson et al. (2012) are also composites from a number of sources. The recent study by Pan et al. (2021) documents the presence of a primary magnetization in the carbonate rocks at the Meishan section and, because of the detailed level of sampling, provides a magnetic polarity record that can be tied to the numerous high-precision age U-Pb zircon age determinations for the section. Notably, according to their work, the Changhsingian is dominated by normal polarity and that the normal polarity magnetozone within which the Permian-Triassic boundary lies is at least 600 ka in duration.

With this as a background, we propose that the upper, reverse polarity magnetozone at the Palo Duro Canyon, Dickens, and Clarendon sections can be correlated with the first reverse polarity chron in the Triassic (Induan), the base of which is about 251.40 Ma, according to Pan et al. (2021). The normal polarity magnetozone containing the ash fall deposits that yield U-Pb zircon ages that cluster at ∼251.9 Ma correlates with the ∼700 ka long normal polarity chron within which the Permian-Triassic boundary, as well as the end-Permian extinction event, have been proposed by many workers to lie. In addition, the reverse polarity magnetozone identified in the upper part of Whitehorse Group strata is interpreted to correlate with the youngest reverse polarity chron in the Changhsingian.

Based on the carbon isotope record of dolomite cements in the Whitehorse Group and Quartermaster Formation, the large negative δ¹³C excursion of ∼8‰ near the bases of both the Palo Duro Canyon and Caprock Canyons stratigraphic sections, in the salt-lake facies, is interpreted to correspond to the pre-extinction negative isotope excursion (Korte and Kozur, 2010) whereas the subsequent sustained upward pattern of relatively negative δ¹³C values is consistent with the post-extinction, Early Triassic continuation of relatively negative marine calcite δ¹³C values that has been documented in strata of southern China (Lou et al., 2011). Therefore, the carbon isotope chemostratigraphy and U-Pb zircon age estimates for the ash fall deposits in the uppermost Whitehorse Group and lower Quartermaster Formation indicate that the upper ∼20 m of Whitehorse Group strata, at most sections, and the entire Quartermaster Formation are earliest Triassic in age, contrary to previous interpretations of these strata as Ochoan (Changhsingian), late Permian (e.g., Gradstein et al., 2012).

Based on field studies of continental Permian-Triassic strata of the Karoo Supergroup exposed in the central Karoo Basin, in South Africa, Ward and others (2000) hypothesized that continental fluvial morphologies, at least at mid-latitudes, shifted from dominantly meandering systems in the latest Permian to braided stream systems in the earliest Triassic, on a relatively global scale, as a result of the mass extinction of plants, which led to increased soil erosion rates and sediment supplies to fluvial systems. These workers argued that the increased sediment supply overburdened continental fluvial systems, which then developed braided morphologies (Ward et al., 2000). In the late Permian to Early Triassic, the central Karoo Basin was located at a southern paleolatitude in Gondwana of about 45°. Although a braided fluvial system is not preserved at the Palo Duro Basin sites presented here, we argue that there is considerable field evidence for the existence of meandering stream systems in the Lower Triassic fluvial facies belt strata of the upper Quartermaster Formation, and that this information can be used to test the vegetation extinction/fluvial style hypothesis. In addition, although macromorphologic evidence for vascular plants is lacking until the uppermost fluvial facies belt strata near the top of the Caprock Canyons section, vascular plants were present within the basin based on the abundance of phytoliths within the sand-flat and fluvial facies belts of the Quartermaster Formation (Figure 5D). Future work will attempt to gather organic matter constituents from the phytoliths preserved in Whitehorse Group and Quartermaster Formation rocks and will further explore for strata exposed below the large negative δ¹³C excursion documented here in the lowermost-exposures of the Whitehorse Group. Nevertheless, the presence of Lower Triassic meandering fluvial deposits of Early Triassic age in the Quartermaster Formation, in conjunction with ubiquity of phytoliths in the sand flat and fluvial facies of the Caprock Canyons section, does not support the hypothesized global transition from Permian age meandering fluvial systems to Triassic age braided morphologies. Fluvial systems in the Palo Duro Basin, which were deposited at low (∼5–10°) northern latitudes in western Pangea, had enough streambank cohesiveness in this region to maintain meandering fluvial systems soon after the end-Permian extinction.

Numerous studies have cited evidence for low atmospheric oxygen levels in the earliest Triassic, which either caused, or were a consequence of processes responsible for the end-Permian extinction. Inferred low oxygen levels were based on reports of the presence of specific pedogenic minerals that form under anoxic conditions (Sheldon and Retallack, 2002), the sedimentary record of fusain (burnt fossil wood; Uhl et al., 2008) and vertebrate functional morphology (Retallack et al., 2003). However, we find the abundance of hematite cement as an early-formed authigenic phase, which clearly is an important carrier of a very-well defined remanence of Permian-Triassic age through most of the salt-lake, sand flat, and fluvial facies belts, to be inconsistent with the hypothesis of low oxygen levels for these depositional environments. If low oxygen levels had prevailed in the Early Triassic atmosphere, ferrous iron-bearing authigenic sedimentary minerals such as siderite, ankerite, berthierine, pyrite, magnetite or maghemite would be prevalent in these strata. The absence of these minerals in Whitehorse Group and Quartermaster Formation rocks argues strongly against an oxygen-minimum life-crisis during the earliest Triassic. Although there is strong evidence for oxygen-poor marine waters in Tethys and Panthalassa during Early Triassic time (e.g., Wignall and Twitchett, 2002), our results suggest that that terrestrial ecosystems were not affected in a deleterious way by low oxygen levels.

In sub-aqueous environments such as those in which Whitehorse Group and Quartermaster Formation sediments were deposited, the δ¹⁸O value of dolomite cements reflects 1) the composition of water from which the dolomite crystallized and 2) the temperature at which dolomite crystallization occurred. The temperature dependence of oxygen isotope fractionation between dolomite and water is:

(1) 1000 lnα_{dolomite-water} = 2.73X10⁶/T² + 0.26