The Anadarko Basin (south-central USA) is the deepest basin on the North American craton and occupies a region largely surrounded by major, late Paleo-zoic plate-margin (Marathon-Ouachita-Appalachian) and intraplate (Ancestral Rocky Mountains) orogenic systems, albeit a distal arm of the latter. The Anadarko Basin hosts an exceptionally voluminous record of Pennsylvanian strata, and much of this fill has been attributed to erosion of the adjacent Wichita uplift—composed of granitic and rhyolitic rocks of Cambrian age—separated from the basin by a fault zone exhibiting 12 km of vertical separation. This work incorporates thin-section petrography (102 samples) and U-Pb detrital zircon geochronology of sandstone samples (12 samples) from core and outcrop of the Middle Pennsylvanian Red Fork Sandstone (and equivalents) as well as slightly younger Upper Pennsylvanian units (Tonkawa, Chelsea, and Gypsy sandstones) in order to interpret drainage pathways and evolution of those pathways toward and into the Anadarko Basin (Oklahoma) and evaluate the relative importance of the major provenance regions.

Our petrographic analysis indicates sandstones with arkosic compositions are limited to the region immediately adjacent to (north of) the Wichita uplift. All remaining samples, which reflect the vast bulk of sediment in the depocenter, including sediment on the northern and eastern Anadarko shelf, are litharenites. Analysis of kernel density plots of the U-Pb ages of detrital zircons together with multidimensional scaling analysis of the Middle Pennsylvanian samples indicate three groups of similar provenance: (1) samples dominated by Cambrian ages from locations directly adjacent to the Wichita uplift; (2) samples dominated by Neoproterozoic ages from locations along the northern shelf of the Anadarko Basin; and (3) samples dominated by Mesoproterozoic ages from locations along the eastern Anadarko shelf and the basin center. These samples are spatially discrete, indicating partitioning of drainage networks during the Middle Pennsylvanian, with two continental-scale fluvial systems entering the Anadarko Basin from the north (transversely) and east (axially). The lack of Cambrian ages in the depocenter and (northern) shelf samples indicate that the Wichita uplift supplied only limited sediment to the basin; sediment derived from the uplift was trapped in fringing fans directly adjacent to the uplift. In contrast to the patterns exhibited by the Middle Pennsylvanian samples, Upper Pennsylvanian samples exhibit more uniform U-Pb ages across the basin. This indicates the relatively rapid evolution of the Appalachian-derived northerly and easterly drainages into an integrated system that flowed axially across the (overfilled) mid-continent basins to the ultimate continental sink in the Anadarko Basin.

The Anadarko Basin (Fig. 1) contains an exceptionally thick record of Pennsylvanian strata that extends across parts of Oklahoma, Texas, Kansas, and Colorado (USA) (Perry, 1989). Rifting, magmatism, and crustal deformation events during the development of the late Proterozoic Southern Oklahoma Aulacogen produced complex basement geology underlying the region (Perry, 1989). Late Paleozoic reactivation of Southern Oklahoma Aulacogen structures created the Amarillo-Wichita uplift (Fig. 2; Price, 2016), and in most models this compressional event created load-induced subsidence during the Pennsylvanian (Brewer et al., 1983; Johnson et al., 1988; Perry, 1989), although other studies suggest significant strike-slip faulting formed some of the observed structures (Wickham, 1978; Liao et al., 2019; Turko, 2019).

To date, no large-scale provenance study has resolved sediment transport into the Anadarko Basin during this period of peak subsidence, despite extensive field-scale studies of structures, stratigraphic sequences, and reservoir characterization within the basin (Hentz, 1994; Peyton et al., 1998; Thomas et al., 2016). Understanding the paleogeography and provenance of fluvial systems entering the Anadarko Basin is essential for our understanding of late Paleozoic source-to-sink sediment routing across the North American craton during major orogenesis on multiple fronts, and these data provide constraints on basin formation mechanisms and basin filling patterns. This work attempts to resolve some of these issues with two objectives: (1) to determine the provenance of Middle and Upper Pennsylvanian sandstones in the Anadarko Basin, utilizing detrital zircon geochronology buttressed with petrographic framework composition; and (2) to use these provenance data to pinpoint sediment pathways into the Anadarko Basin. In doing so, we compare to provenance data from other Pennsylvanian basins in North America and provide constraints on the location of large-scale drainage systems within central Pangaea and suggest how our provenance data bear on models of basin formation.

Global and Regional Paleogeography and Climate

During Pennsylvanian time, Pangaean assembly occurred by complex, rotational suturing of Laurentia and Gondwana near the paleoequator (Hatcher, 2005; Kroner et al., 2016). This suture occurred along a complex belt of orogenic uplifts referred to as the Central Pangaean Mountain Belt and the associated foreland basins, which include the Appalachian, Black Warrior, Arkoma, and Fort Worth basins (Fig. 1; Thomas, 1977; Perry, 1989; Otto-Bliesner, 1993). Far-field stresses from this collision also reactivated subsidence and sedimentation in the intracratonic Michigan, Williston, and Illinois basins (Fig. 1; Klein and Hsui, 1987). Both axial and transverse transport through these basins have been proposed as means to deliver Appalachian-derived sediment to western North America, although exact transport pathways remain debated (Gehrels et al., 2011; Kissock et al., 2018; Chapman and Laskowski, 2019). In southern Laurentia (midcontinent USA), platformal carbonate deposits were incised by fluvial-deltaic systems, then both were covered by deep-basinal deposits as the Wichita uplift rose contemporaneously with accelerated subsidence of the Anadarko Basin (Tomlinson and McBee, 1959; Ambrose, 2011; Soreghan et al., 2012).

Studies of paleochannel orientations and framework mineralogy indicate sediment transport pathways into the Anadarko Basin during the Pennsylvanian from the north, east, west, and southwest, with inferences of sediment provenance from the Wichita uplift, northern Ancestral Rocky Mountains, Nemaha uplift, and Appalachian foreland basin (Fig. 1; Hentz, 1994; Lambert, 2006; Alsalem et al., 2018; Kissock et al., 2018). Detrital zircon geochronology of approximately coeval strata onlapping the Arbuckle Mountains and within the Arkoma foredeep suggests recycled pre-Ordovician continental-margin sedimentary rock and Appalachian orogenic rocks as potential additional provenance regions (Sharrah, 2006; Thomas et al., 2016). The Arkoma and Black Warrior basins to the east (Fig. 1) both show a strong recycled orogenic provenance signal matching the provenance of coeval strata in the Appalachian foreland basin, consistent with the concept of an interconnected axial drainage network in equatorial western Pangaea during late Paleozoic Pangaean suturing (Archer and Greb, 1995; Sharrah, 2006; Boothroyd, 2012; Yezerski, 2013; Xie et al., 2018; Kissock et al., 2018).

Late Paleozoic paleoclimate was dominated by Gondwanan glaciation (Montañez and Poulsen, 2013). Accordingly, glacial-interglacial fluctuations during the Pennsylvanian drove eustatic sea-level variations of 40–120+ m (Soreghan and Giles, 1999; Rygel et al., 2008); in epeiric-sea basins such as the Appalachian, Anadarko, and midcontinent basins, these fluctuations caused large lateral facies shifts (Heckel, 1994; DiMichele et al., 2010). These coupled eustatic and climatic fluctuations are recorded by, in part, extensive development of incised paleovalleys and lithologically diverse cyclothem deposits in basins across North America (Heckel, 1986; Perry, 1989; Archer and Greb, 1995; Cecil et al., 2003; Kissock et al., 2018; Fielding et al., 2020). Incised paleovalleys are well studied in the Anadarko Basin and were major sediment pathways (and now petroleum reservoirs) throughout the studied strata (Brenner, 1989; Doyle and Sweet, 1995; Andrews, 1997; Lambert, 2006; Fitzjarrald, 2016). For this study, both fluvial and coeval subaqueous fan deposits are utilized to characterize sediment sources for the Anadarko Basin.

Tectonic History of the Anadarko Basin

The regional geology of southwestern Oklahoma was shaped by four tectonic events, each recorded by a distinct lithologic sequence (Perry, 1989). The first event was the formation of the late Proterozoic Southern Oklahoma Aulacogen, which began with extension-driven intrusion of the mafic Glen Mountains Layered Complex and Roosevelt Gabbros, eruption of the Carlton Rhyolite, and intrusion of the Wichita Granite Group, all in response to the breakup of Rodinia ca. 535 Ma (Powell et al., 1980). These igneous events are collectively referred to as the Wichita Mountains Igneous Suite. The second event was early Paleozoic thermal subsidence, recorded by ~4 km of passive-margin initial clastic and then predominantly carbonate sedimentary strata (Ham, 1973; Ball et al., 1991). The third event was Mississippian–Pennsylvanian compression, resulting in formation of the Wichita uplift and adjacent Anadarko Basin (Perry, 1989). The Anadarko Basin comprises an ~10-km-thick package of strata extending across parts of Oklahoma, Kansas, Colorado, and Texas, with ~6 km of strata deposited during the Pennsylvanian–Permian alone (Fig. 2; Witt et al., 1971). The uplift is considered the easternmost extension of the Ancestral Rocky Mountains orogeny and on trend with the former Southern Oklahoma Aulacogen and related Precambrian rift structures extending northwestward (Soreghan et al., 2012; Price, 2016). The fourth event includes minor uplift pulses in the Permian and subsequent subsidence and burial of both the Anadarko Basin and Wichita uplift that ultimately preserved Permian paleorelief (Soreghan et al., 2012; Price, 2016).

Study Area and Stratigraphy

The study includes Pennsylvanian fluvial, deltaic, and subaqueous sandstone strata sampled from outcrop and core within the Anadarko Basin of Oklahoma and northwestern Texas (Table 1; Fig. 3). Stratigraphic names vary between outcrop and subsurface and among subregions. Thus, e.g., “Red Fork Sandstone” (samples RF1–RF6; Table 1; Fig. 3) applies to subsurface sandstone of various depositional origins correlated on the basis of stratigraphic position below a marker bed termed the Pink limestone; the Red Fork is part of the Cherokee Group on the Cherokee platform (Fig. 3). The Red Fork crops out in eastern Oklahoma and on the Cherokee platform but is termed the Boggy Formation (sample BG1) of the Krebs Group in the Arkoma Basin and eastern Oklahoma (Fig. 3). Additionally, we sampled the Chelsea sandstone (sample CH1), which crops out in eastern Oklahoma and lies stratigraphically above the Cherokee Group. We sampled two Upper Pennsylvanian units, the Tonkawa sandstone (sample TK1; subsurface, Anadarko Basin) and Gypsy sandstone (sample GY1; outcrop, Cherokee Platform), as a preliminary test for changes in provenance between the Middle and Late Pennsylvanian Epochs. Finally, we sampled three subsurface units defined by well operators as “Granite Wash,” which is an informal term assigned to poorly sorted, commonly conglomeratic clastics that occur adjacent to the Wichita uplift in the subsurface (Fig. 3) and are part of the Marmaton Group and overlying Shawnee Group (Lyday, 1985; Johnson et al., 1988; Boyd, 2008; Mitchell, 2011). Based on previous work (Boyd, 2008; Mitchell, 2011), the Granite Wash sample labeled GW1 (Table 1) is of Late Pennsylvanian age, whereas the other two samples (GW2 and GW3) are of Middle Pennsylvanian age.

Potential Detrital Zircon Source Regions

To interpret provenance of Pennsylvanian strata in the Anadarko Basin by U-Pb geochronology of detrital zircons, the age range and location of potential source rocks present at the time of deposition need to be identified. While not all detrital zircons are assumed to originate from first-cycle weathering of age-correlated basement rock, these age populations can also be compared to sedimentary rocks in neighboring basins and to lower Paleozoic rocks that may have been eroded and deposited as recycled sediments in the Anadarko Basin. The potential source terranes are outlined below with pertinent studies referenced for the age range as well as evidence of late Paleozoic exposure; these terrane designations are similar to, and in some cases summarized from, other recent studies utilizing these same techniques (e.g., Gehrels et al., 2011; Kissock et al., 2018; Soreghan et al., 2018; Wang and Bidgoli, 2019; Thomas et al., 2020; Allred and Blum, 2021).

Archean Cratons and Paleoproterozoic Orogens (>1825 Ma)

The ultimate source regions for U-Pb ages >1825 Ma include the Superior and Wyoming cratons (3600–2500 Ma) and the Trans-Hudson and Penokean orogenic belts (1800–1900 Ma; Hoffman, 1989; Van Schmus et al., 1996), both located north of the Anadarko Basin (Fig. 4). The Superior and Wyoming cratons were covered in the late Paleozoic, although lower Paleozoic strata contain abundant Archean grains (Chapman and Laskowski, 2019) and were exposed regionally; detrital zircons from these sources are also documented in upper Paleozoic sandstone units east and west of the Anadarko Basin (Becker et al., 2005; Sharrah, 2006; Link et al., 2014; Kissock et al., 2018; Alsalem et al., 2018). Basement ages of 1950–2250 Ma, however, are not common in the North American craton. Potential sources for detrital zircon grains in this age range are likely the Trans-Amazonian and Eberian orogenic belts (1950–2250 Ma) that were exposed during exhumation of Gondwanan terranes along the Central Pangaean Mountain Belt or recycled from lower Paleozoic units within peri-Gondwanan terranes or strata of the Arkoma Basin (Sharrah, 2006; Neves, 2015; Thomas et al., 2017).

Paleoproterozoic Yavapai-Mazatzal and Mesoproterozoic Granite-Rhyolite Terranes (1825–1300 Ma)

The broad 1825–1300 Ma age range correlates with known ages of the mid-continent Yavapai-Mazatzal (1600–1800 Ma), Central Plains (1800–1600 Ma), and Granite-Rhyolite basement provinces (1600–1300 Ma) as well as plutons throughout these provinces (Hoffman, 1989; Van Schmus et al., 1996; Fig. 4). The Yavapai-Mazatzal terrane was exposed in the Ancestral Rocky Mountains (excluding the Amarillo-Wichita uplift) during Pennsylvanian–Permian time, and the Central Plains and Granite-Rhyolite terranes were exposed along the Transcontinental Arch during Cambrian–Mississippian time (Billo, 1985; Van Schmus et al., 1996). Late Paleoproterozoic–early Mesoproterozoic detrital zircons occur within Paleozoic strata of the Grand Canyon (Arizona) and basins of the Ancestral Rocky Mountains (e.g., Soreghan et al., 2002; Gehrels et al., 2011). Zircons of this age range also occur in Mesoproterozoic–lowest Cambrian arenites in the western U.S. and the Fort Worth basin in Texas, Ordovician strata of the Arbuckle Mountains (Oklahoma; Stewart et al., 2001; Thomas et al., 2017; Alsalem et al., 2018), and Lower Pennsylvanian strata in the Arkoma Basin (Sharrah, 2006).

Mesoproterozoic Grenville Basement (1300–925 Ma)

The widespread Grenville basement as well as basement of the Mid-continent Rift are of the 1300–925 Ma age range. The Grenville basement (900–1300 Ma) is a remnant of an orogenic belt produced during the formation of Rodinia, while the Midcontinent Rift (1100 Ma) is a failed rift of the subsequent breakup of Rodinia (Stein et al., 2018). Zircons of the Midcontinent Rift province tend to narrowly cluster in age centered on 1100 Ma, the time of the igneous activity. Zircons derived from the Grenville basement exhibit a broad age spread and are documented in ancient and modern fluvial sediments across North America and as far as northwest Canada (Eriksson et al., 2004; Becker et al., 2005; Moecher and Samson, 2006; Sharrah, 2006; Gehrels et al., 2011; Boothroyd, 2012; Kissock et al., 2018; Alsalem et al., 2018). The lateral extent of the Grenville basement rocks (Fig. 4), the zircon fertility of the Grenville basement (Dickinson, 2008), and the abundance of Grenville-aged zircon grains in Paleozoic strata (Thomas, 2011) greatly diminish the utility of these zircons in provenance studies.

Neoproterozoic Peri-Gondwanan Terranes (790–570 Ma)

The 790–570 Ma age range reflects rift-related igneous rocks and terranes caught up within the Central Pangaean Mountain Belt to the east and southeast of the Anadarko Basin. Two pulses of Neoproterozoic Iapetan rifting are preserved in anorogenic igneous suites in the central Appalachians (Aleinikoff et al., 1995). The Neoproterozoic age population also encompasses the basement ages of peri-Gondwanan terranes (Fig. 4), a term that includes the Avalonian, Carolinian, Suwanee, and Yucatan-Maya terranes accreted to southern Laurentia during the Paleozoic and exhumed during the Alleghanian orogeny (Murphy et al., 2004; Thomas et al., 2004; Sharrah, 2006; Mueller et al., 2014). This age population occurs in late Paleozoic basins across North America but is usually less abundant than populations from the Grenville, Granite-Rhyolite, and Yavapai-Mazatzal source regions (Thomas et al., 2004; Eriksson et al., 2004; Becker et al., 2005; Sharrah, 2006; Xie et al., 2018; Alsalem et al., 2018; Kissock et al., 2018).

Cambrian Igneous Rocks of the Wichita Mountains (570–500 Ma)

The Wichita Mountains Igneous Suite, with ages in the 570–500 Ma range, formed during the final stages of rifting of Rodinia in the early Paleozoic and was clearly an uplifted source in the late Paleozoic. Crystallization ages of the various plutons, dikes, and layered mafic rocks range in age between 525 and 545 Ma (Hanson et al., 2013, and references therein). Similar ages also occur in early Paleozoic igneous plutons in New Mexico and Colorado (574–427 Ma) that were unroofed in the Ancestral Rocky Mountains uplifts to the west (McMillan and McLemore, 2004; Soreghan and Soreghan, 2013).

Paleozoic Appalachian Terranes (500–300 Ma)

Zircons in the 500–300 Ma age group are sourced from successive orogens recorded in rocks of eastern North America. The orogens contain basement and coeval strata of three tectonic events that formed the Appalachian Mountains: Taconic (490–430 Ma), Acadian (420–350 Ma), and Alleghenian (330–270 Ma; Miller et al., 2000; Aleinikoff et al., 2002). In addition, Late Mississippian volcanic arcs formed during subduction of southern Laurentia beneath Gondwana, as documented in provenance studies of strata in Oklahoma and Arkansas (Shaulis et al., 2012) and from newly dated tuffs from the Ouachita Mountains (Oklahoma) and Midland basin (Texas; Tian et al., 2022). Additionally, Middle Pennsylvanian volcanic ash layers are documented in central and southern Appalachian coal strata (Lyons et al., 1997), the ages of which approximate the depositional ages of the study samples.

Rock Description and Petrography

The sampled core intervals (ten) and outcrop sections (three) were logged at centimeter scale, detailing lithology, grain size, contacts, and sedimentary structures, and are reported in Table 1. Petrographic samples from cores were taken from sandstone intervals at least 7 m thick from throughout the interval (seven to ten samples per core) for description and point counting. One to three representative thin sections were made from intervals sampled in outcrop; we used published petrographic descriptions for the Gypsy sandstone and Boggy Formation outcrops (Dyman, 1987; Doyle and Sweet, 1995). Point counts were made of 400 sand-sized (>62.5 μm) framework grains classified into 11 categories using the Gazzi-Dickinson method (Table 2; Dickinson, 1970; Ingersoll et al., 1984), with additional feldspar textures typical of the Wichita Mountains Igneous Suite units identified and tabulated (Price et al., 1996; Morgan and London, 2012).

Detrital Zircon Geochronology

We processed sandstone samples (~2 kg for core and as much as 5 kg for outcrop samples) to extract detrital zircon grains for geochronological analysis at the Arizona LaserChron Center (Tucson, Arizona) using processing methodologies described by Gehrels and Pecha (2014) and Pullen et al. (2018). Details of acquisition and data reduction are presented in the Supplemental Material1 along with the summary U-Pb data and are summarized here. After processing bulk samples, we epoxy mounted and polished the samples and then imaged them using cathodoluminescence; grains not excluded because of cracks or strong zoning were analyzed using a spot size of 30 µm after a cleaning shot of 40 µm. Standards were analyzed after every five unknowns; the primary standard was FC-1 (1099.0 ± 0.6 Ma; Paces and Miller, 1993), and SL-M (563.2 ± 3.2 Ma; Gehrels et al., 2008) and R33 (418.9 ± 0.4; Black et al., 2004) were used as secondary standards.

Protocols for filtering and inclusion of U-Pb ages for population analysis varies significantly among even recent studies. In this study, we reduced the raw data using the Arizona LaserChron Center’s internal data reduction program (E2agecalc; Pullen et al., 2018). Ages used in the analyses are based on 207Pb/235U ages when <800 Ma and Pb206/Pb207 ages when >800 Ma. Uncertainty filters were set at 10% for both 207Pb/235U and Pb206/Pb207 ages, but for the latter case only when the Pb206/U238 age exceeded 400 Ma. Maximum discordance cutoffs were set at >20% (normal) or >5% (reverse) using 206Pb/238U versus 207Pb/235U ages when <400 Ma, and Pb206/U238 versus Pb206/Pb207 ages when >400 Ma. We present detrital zircon age data for each sample as kernel density estimate plots using DensityPlotter (version 8.3; Vermeesch, 2012) overlain by histograms of the individual ages for each sample produced. The kernel density estimate of a probability distribution arranges (in this case) the U-Pb ages of a sample and stacks a “kernel” of a certain bandwidth over the data (Vermeesch, 2012). DensityPlotter determines an adaptive bandwidth that varies with the density of the ages and thus varies within and between samples (Vermeesch, 2012).

To further explore differences in U-Pb ages among the detrital zircon samples, we performed a multidimensional scaling analysis following the methods of Vermeesch (2019) and using the R-based “provenance” package described therein. The multidimensional scaling analysis uses the Kolmogorov-Smirnov pairwise distances among all samples and produces dimensional coordinates used as the axes for a scatterplot with each sample plotting within this two-dimensional space; the proximity of samples in the plot is a measure of their compositional similarity (Vermeesch, 2018, 2019).

Results are grouped spatially by detrital zircon sample (note that some cored intervals contain multiple modal petrography samples) based on the position within the Anadarko Basin: southwestern (proximal to uplift), the depocenter, northern and northeastern shelf, and eastern margin. Details of the sample locations, thicknesses, and depositional interpretations of the sampled intervals are presented in Table 1, whereas Table 2 summarizes the average modal compositions of the petrographic data as well as the U-Pb ages of zircon grains by source-terrane age ranges. Figure 5 presents a ternary plot (monocrystalline quartz–feldspar–total lithics, Qm-F-Lt) of the petrographic data, and Figure 6 presents the kernel-density-estimate and histogram plots of the detrital zircon age data. Full compositional data sets are presented in Tables S1 and S2 (see footnote 1).

Southwestern (Proximal) Margin of the Anadarko Basin

Samples proximal to the Wichita uplift (Fig. 3) include two cores of Middle to possibly Upper Pennsylvanian Granite Wash (samples GW1 and GW2). Conglomerate observed in these cores contains granitic and gabbroic pebbles of lithologic types identical to those of the Wichita Mountains Igneous Suite as well as clasts of massive or bedded carbonate.

Sample GW1 (Granite Wash)

The interval containing sample GW1 is predominantly coarse sandstones that record proximal alluvial deposition (Table 1). The average Qm-F-Lt petrographic composition of sandstone units in GW1 is 33%-59%-8% (Table 2; Fig. 5), and thin-section observations reveal abundant perthitic feldspar grains with diagnostic Wichita Granite Group granophyre textures (Morgan and London, 2012; Supplemental Material, see footnote 1). Sample GW1 contains detrital zircons that range in age from 450 to 1373 Ma, but >90% of the grains exhibit ages matching the proximal Wichita Mountains Igneous Suite (500–570 Ma; Table 2; Fig. 6).

Sample GW2 (Granite Wash)

Sample GW2 is from a coarse sandstone within a cored interval interpreted as subaqueous (e.g., fan-delta) deposits (Table 1) analogous to Granite Wash cores described by Mitchell (2011). The average Qm-F-Lt petrographic composition of sandstone in GW2 is 41%-48%-11% (Fig. 5). Thin sections contain abundant feldspar grains, including graphic and perthitic feldspars characteristic of the Wichita Granite Group. Sample GW2 contains detrital zircon ages that range from 410 to 2950 Ma. The dominant age grouping is from the Wichita Mountains Igneous Suite (73.3%), but ~13% of grains match Grenville-age basement; other individual source regions contributed <5% each (Table 2; Fig. 6).

Depocenter of the Anadarko Basin

Samples from the Anadarko Basin depocenter, all from core, were taken from intervals previously identified as Middle Pennsylvanian Granite Wash (sample GW3) and Red Fork Sandstone (samples RF5 and RF6) and Upper Pennsylvanian Tonkawa sandstone (sample TK1).

Sample RF5 (Red Fork Sandstone)

Sample RF5 was collected from a fine-grained sandstone within a cored interval that reflects deposition in a delta slope environment (Table 1), analogous to shelf-edge deltas described in the (coeval) Boggy Formation of the nearby Arkoma Basin by Andrews (1997). The average Qm-F-Lt composition for RF5 is 67%-3%-30% (Table 2). Detrital zircons in RF5 range in age from 337 to 3296 Ma and represent a mix of sources; the largest group matches Grenville basement ages (34%), but the sample also contains >10% each of Granite-Rhyolite basement (15.1%), Neoproterozoic peri-Gondwana sources (13.9%), Archean cratonal sources (13.1%), and Yavapai-Mazatzal terranes (10.8%), but <5% of grains match the age of the Wichita Mountains Igneous Suite (Table 2).

Sample RF6 (Red Fork Sandstone)

Sample RF6 occurs within a series of sandstone units 10–50 cm thick that are interpreted as subaqueous slope deposits (Table 1). The average Qm-F-Lt composition for RF6 is 69%-4%-27% (Table 2), and samples in this interval exhibit angular quartz grains in a silty matrix and trace granophyric feldspar. Detrital zircon grains sampled in RF6 range in age from 341 to 2831 Ma. The largest group matches Grenville basement ages (28.3%), but like sample RF5, this sample contains a mix of other sources, including Granite-Rhyolite basement (15.8%), Neoproterozoic peri-Gondwana sources (15.8%), Archean cratonal sources (11.3%), and Yavapai-Mazatzal terranes (10.9%), and 7.5% of Wichita Mountains Igneous Suite (Table 2).

Sample GW3 (Granite Wash)

Although drilling records identified the sample GW3 interval as “Granite Wash,” we have included this sample in the depocenter samples because of its position farther removed from the Wichita uplift. The cored interval of sample GW3 is inferred to reflect deposition in a subaqueous slope (Table 1), similar to coeval cores analyzed by Mitchell (2011). The average Qm-F-Lt composition for GW3 is 56%-5%-39% (Table 2). Sample GW3 contains detrital zircon grains that range in age from 344 to 3241 Ma, with the largest group matching Grenville basement (41.4%) and a mix of other sources, including Granite-Rhyolite basement (15.3%), Archean cratonal sources (14.2%), and Yavapai-Mazatzal terranes (10.3%; Table 2). Only 1% of the grains exhibit ages that match the proximal Wichita Mountains Igneous Suite.

Sample TK1 (Tonkawa Sandstone)

Sample TK1 (2964 m deep, study interval 23 m thick) was collected from a normally graded fine- to medium-grained sandstone. The cored interval is interpreted as a subaqueous-fan deposit (Table 1), consistent with previous interpretations by Fitzjarrald (2016). The average Qm-F-Lt composition for TK1 is 62%-2%-36% (Table 2), however there is a wide scatter in the relative amount of Qm to Lt (Fig. 5), with the latter composed mostly of foliated metamorphic rock fragments (Supplemental Material, see footnote 1). Sample TK1 contains detrital zircon grains that range in age from 370 to 2918 Ma. The majority of zircon grains ages match Grenville basement (54.8%), but 16% of the grains match Granite-Rhyolite basement ages (Table 2). The sample contains only 1% of grains with ages matching the ages of proximal Wichita Mountains Igneous Suite.

Northern and Northeastern Oklahoma (Anadarko Shelf and Cherokee Platform)

In northern and northeastern Oklahoma, Red Fork Sandstone units were sampled (samples RF2, RF3, and RF4) from cores on either side of the Nemaha uplift, which approximately separates the northern Anadarko Shelf and the Cherokee Platform (Fig. 3), as well as from outcrops of the slightly younger Middle Pennsylvanian Chelsea sandstone (sample CH1) and Upper Pennsylvanian Gypsy sandstone (sample GY1).

Sample RF2 (Red Fork Sandstone)

Sample RF2 was sampled from a sandstone near the base of the cored interval, which is interpreted as a submarine slope deposit (Table 1), possibly in proximity to the Cherokee carbonate platform (Ball et al., 1991). The average Qm-F-Lt composition for RF2 is 67%-1%-32% (Table 2) with a variable amount of poly-crystalline quartz. Detrital zircon ages in RF2 range from 308 to 3628 Ma, with the two largest groups exhibiting ages matching Neoproterozoic peri-Gondwanan terranes (33.3%) and Paleozoic Appalachian terranes (20.0%). Only 15.0% match Grenville basement ages and 13.8% match Archean craton ages.

Sample RF3 (Red Fork Sandstone)

Sample RF3 comes from a thick, sharp-based sandstone unit that likely records deltaic deposition (Table 1), matching descriptions of Boggy Formation cores interpreted in this region by Andrews (1997). The average Qm-F-Lt composition for sandstones within the RF3 core is 76%-3%-21%, with foliated metamorphic rock fragments as the primary lithic component. Detrital zircons in sample RF3 range in age from 326 to 3005 Ma, with 38.1% of the grains matching Neoproterozoic peri-Gondwanan ages and 19.4% of the grains matching Archean cratonal ages, but only 11.9% of ages match Grenville basement ages (Table 2).

Sample RF4 (Red Fork Sandstone)

Sample RF4 was collected from near the base of a cored interval composed entirely of sandstone that likely formed in a fluvial channel (Table 1), similar to analogous facies described in other Red Fork Sandstone cores by Andrews (1997). The average Qm-F-Lt composition for RF4 is 67%-2%-31% (Table 2). Sample RF4 contains detrital zircons with ages between 328 and 3260 Ma, and—similar to samples RF2 and RF3—the largest age grouping matches Neoproterozoic peri-Gondwanan terrane ages (34.2%), with relatively few grains matching Grenville basement (16.0%); additionally, 20.5% of grains in RF4 match Archean craton ages (Table 2).

Sample CH1 (Chelsea Sandstone)

Sample CH1 comes from a sandstone outcrop interpreted as fluvial (Table 1; Govett, 1959; Cole, 1968, and references therein). Representative thin sections from this work and Dyman (1987) for CH1 exhibit an average Qm-F-Lt composition of 71%-3%-26%. Sample CH1 contains detrital zircon grains that range in age between 347 and 3547 Ma but is dominated by ages matching Grenville basement (37.7%) and Paleozoic Appalachian terranes (24.7%) with only 8.1% of detrital grains matching Neoproterozoic peri-Gondwana terranes (Table 2).

Sample GY1 (Gypsy Sandstone)

Sample GY1 comes from a sandstone outcrop interpreted as fluvial channel and overbank deposits (Doyle and Sweet, 1995). Gypsy sandstone samples exhibit a homogeneous Qm-F-Lt composition of 82%-2%-16% (Table 2; Dyman, 1987; Doyle and Sweet, 1995). Detrital zircon grains in GY1 range in age from 392 to 2949 Ma but are dominated by grains matching Grenville basement (45.3%) with subequal numbers of grains matching Yavapai-Mazatzal and Granite-Rhyolite basement ages (16.8% each), but few grains (3.3%) match Neoproterozoic peri-Gondwana terrane ages (Table 2).

Eastern Margin

One well-core interval of the Red Fork Sandstone and one outcrop of the correlative Boggy Formation (samples RF1 and BG1, respectively) were sampled along the eastern margin of the Anadarko Basin.

Sample RF1 (Red Fork Sandstone)

Sample RF1 comes from a sandstone within a cored interval interpreted as a marginal-marine setting (Table 1), as posited by Andrews (1997). The average Qm-F-Lt composition for RF1 is 78%-1%-21%, and thin sections exhibit little variation in composition and grain size. Detrital zircon grains in RF1 range in age from 376 to 2706 Ma, with the majority (52.8%) of grains matching Grenville basement ages and another 18.7% matching the basement ages of the Granite-Rhyolite province (Table 2).

Sample BG1 (Red Fork Sandstone–Boggy Formation)

Sample BG1 (elevation 215 m, study interval 10 m thick) is from a sandstone outcrop inferred to record fluvial deposition (Table 1; Andrews, 1997). Thin sections of BG1 reveal an average Qm-F-Lt composition of 81%-1%-18% (Table 2; Dyman, 1987). Sample BG1 contains detrital zircons with ages between 301 and 2822 Ma, with the majority of grains (54.2%) matching ages of the Grenville basement and another 12% matching Paleozoic Appalachian terrane ages (Table 2).

Spatial and Temporal Trends in Provenance

Detrital Zircon Age Trends

Table 2 and visual inspection of the kernel density estimate plots (Fig. 6) reveal differences among samples in the relative proportions of zircons from the different source types (age groupings) outlined above. These differences indicate spatial and temporal changes in the provenance of the sedimentary systems that transported sand into the Anadarko Basin. Figure 7 is a plot of the multidimensional scaling analysis done on all of the zircon samples in this study. Figure 7A suggests that three main groups exist based on their proximity and grouping within the plot; the nature of these groupings and their spatial and temporal distribution are discussed below.

The first group (group A; Fig. 7A) contains two samples (GW1 and GW2) farthest removed from all other samples in the multidimensional-space plot. These samples are from wells located most proximal to the Wichita uplift (Fig. 3). Both samples are dominated by detrital zircons exhibiting a very narrow age spectrum (Fig. 7B) that corresponds with the ages of the Cambrian igneous rocks exposed within the adjacent Wichita uplift. The most proximal sample, GW1, records fluvial deposition and contains <10% grains with U-Pb ages older than the igneous rocks of the Wichita uplift; hence it exhibits the most distinct detrital zircon age spectrum and plots to the extreme right within the multidimensional-scaling plot. Sample GW2 records subaqueous deposition and contains ~30% grains with U-Pb ages older than the local Wichita igneous rocks but ~70% grains reflecting derivation from the local Wichita uplift.

The second group (group B; Fig. 7A) consists of three samples (RF2, RF3, and RF4) clustered around the origin of the x-axis but in negative space along the y-axis in the multidimensional-space plot (Fig. 7A). These three samples all contain a high percentage of Neoproterozoic grains (Fig. 7B) that likely reflect derivation from peri-Gondwanan terranes (33%–38%); no other sample contains >16% of this age group. Consequently, in these samples, the fraction of grains likely reflecting Grenville basement is relatively small (12% to 16%), especially compared to the third group (Fig. 7B). Group B samples come from wells located along the northern Anadarko shelf and northeastern part of the Cherokee platform (Fig. 3).

The final group (group C; Fig. 7A) consists of the remaining eight samples (TK1, RF1, RF6, BG1, RF5, GW3, CH1, GY1); these samples form a slightly more diffuse group that plots on the negative side of the x-axis but in positive space along the y-axis and distinct from groups A and B. Overall, the group is distinguished by a large proportion of grains of Grenville basement derivation; however, this larger group appears to consist of two subgroups, with the first subgroup (C1; Fig. 7A) plotting farther from the origin on the y-axis compared to the second subgroup (C2; Fig. 7A). Examination of the relative proportions of age groupings suggests that the C1 subgroup contains ~45%–55% Grenville-age grains but the C2 subgroup varies from 28% to 41% of this same age grouping (Fig. 7B; Table 2). The major difference between group C and group B, in particular, is the relative lack of Neoproterozoic-age grains in group C (Fig. 7B), although subgroup C2 contains on average slightly more of these grains (8%–16%) compared to subgroup C1 (2%–9%). Spatially, group C represents samples from the eastern side of the Anadarko Basin as well as the keel of the basin (Fig. 3). In Figure 7B, we have also separated out the slightly younger Middle–Upper Pennsylvanian samples (CH1 and GY1) from the rest of group C because these two samples are located along the northeastern margin, similar in location to samples of compositional group B (Fig. 3) but with an age spectrum similar to that of group C. The spatial and temporal differences among these groups are discussed further in the section Implications for Paleogeography and Basin-Filling Models below.

Framework Mineralogy Trends

The point-count data on framework mineralogy of sandstones from the studied intervals indicate two distinct framework mineralogies: arkosic and quartzolithic (Fig. 5; Table S2 for numeric data, see footnote 1). Arkosic samples (all from samples GW1 and GW2) plot within the “basement uplift” field of Dickinson et al. (1983) and exhibit a quartzofeldspathic composition with specific granophyric feldspar grains typical of the Wichita Granite Group, consistent with their locations proximal to the Wichita uplift. All other samples exhibit quartzolithic compositions with variable percentages of monocrystalline quartz and polycrystalline quartz with metamorphic textures but 5% or fewer feldspar grains (Fig. 5). This composition plots in the “recycled orogen” field of Dickinson et al. (1983), typical of recycled North American sandstone units and sandstones derived from orogenic belts.

Thus, sandstones of group A are distinctive in terms of both modal mineralogy and detrital zircon age distribution, whereas sandstone modal composition of the other two groups (B and C) comingle along the Qm-Lt join but exhibit distinct detrital zircon age distributions. Notably, the sandstones within the Anadarko Basin as a whole plot in very different tectonic discriminant fields (Fig. 5), suggesting that care must be utilized in using these tectonic discriminant diagrams in isolation, especially with the presence of potential long-distance transport (e.g., Garzanti, 2019).

Implications for Paleogeography and Basin-Filling Models

As shown by both the framework mineralogy and detrital zircon age data (Figs. 5, 6, 7), the Wichita uplift formed a spatially restricted and volumetrically minor source for clastic influx into the Anadarko Basin, despite its proximity and high structural relief (Fig. 2). The Granite Wash unit has long been interpreted as fan-delta and submarine-fan deposits that accumulated along the southern margin of the Anadarko Basin and derived directly and exclusively from erosion of the igneous suite within the Wichita uplift (e.g., Edwards, 1959; Johnson, 1989; Mitchell, 2011), but our data demonstrate that minimal sediment derived from the Wichita uplift reached the keel of the Anadarko Basin. Sample GW1, from a core within the northern margin of the Wichita uplift, contains >90% zircons that match the age of the exposed Wichita Granite Group, and all sandstone samples from this interval exhibit arkosic compositions that include the unique granophyric feldspars of the Wichita granites. Furthermore, zircons from a sample analyzed by Thomas et al. (2016) from the Pennsylvanian–Permian Post Oak Conglomerate, which forms a local mantle on the Wichita uplift, exhibit nearly exclusive derivation from the Cambrian igneous rocks. However, sample GW2, located just a few kilometers north of sample GW1, contains only 73% zircons that match the Wichita igneous ages, and the additional petrographic samples analyzed from the same core interval as sample GW2 contain, on average, 5% less feldspar than the petrographic samples analyzed from the same core interval as GW1 (see Table S2, footnote 1). The remaining samples within the Anadarko depocenter (GW3, RF6, RF5) contain mere traces (1%–7%) of zircons with ages matching the Wichita igneous rocks, and the sandstone modal compositions are exclusively litharenites with <5% feldspar. Thus, based on our data (Fig. 7B), it appears the majority of sediment delivered to the Anadarko Basin during peak subsidence of the Middle and Late Pennsylvanian (Soreghan et al., 2012) was not derived from the deformed and uplifted block bordering the Anadarko Basin.

Instead, the provenance data indicate that sedimentation within the Anadarko Basin records primarily long-distance transport of siliciclastic sediment sourced from distal regions—primarily the greater Appalachian orogen to the east. Moreover, this far-traveled sediment entered the basin from at least two discrete but coeval pathways: from the north and from the east (in modern coordinates; Fig. 8). Early studies of core-based thickness trends and inferred geometries of Red Fork sand bodies suggested transport from the north (Anadarko shelf) and the northeast (Cherokee platform) and deposition at the margins and basin center by fluvial-deltaic and submarine-fan systems, respectively (e.g., Visher et al., 1971; Andrews, 1997). More recently, a number of unpublished studies of the Red Fork Sandstone employing industry seismic data and seismic attribute analysis and other techniques (e.g., Peyton et al., 1998) imaged some of these channel systems inferred in earlier works but also suggested transport from the east within the deep Anadarko Basin during the Middle Pennsylvanian. However, these previous studies could only speculate on the ultimate location of the drainage systems that debouched into the Anadarko Basin. The provenance results in the present study, however, provide these constraints and indicate that the Red Fork and correlative strata were deposited by at least two contemporaneous and large river systems, draining distinctively different source regions during the Middle Pennsylvanian, and likely tied to large continental-scale drainages (Fig. 8); the location, persistence, and origin of this drainage network has been the subject of numerous recent studies, most based on new and/or compiled detrital zircon studies with increasingly larger datasets (e.g., Archer and Greb, 1995; Gehrels et al., 2011; Chapman and Laskowski, 2019; Thomas et al., 2017; Kissock et al., 2018; Xie et al., 2018; Wang and Bidgoli, 2019; Leary et al., 2020; Thomas et al., 2020; Allred and Blum, 2021; Thomas et al., 2021; Lawton et al., 2021).

During the latest Mississippian and Early Pennsylvanian, subsidence within the Anadarko Basin initiated as the region transitioned from a stable carbonate shelf in the Mississippian into a very deep basin by Middle Pennsylvanian (Perry, 1989; Soreghan et al., 2012). Several studies suggest that during the late Mississippian, continental-scale drainages flowed eastward to the western margin of Pangaea (Gehrels et al., 2011; Chapman and Laskowski, 2019; Lawton et al., 2021), although Wang and Bidgoli (2019) suggested that incised valleys in the Late Mississippian may have tapped into this postulated eastwest drainage system and created sediment transport southward to the Illinois basin and the Hugoton embayment (Fig. 1)—the northwestern extension of the Anadarko Basin. Hollingworth et al. (2021) showed that Atokan (to Middle Pennsylvanian) sandstones within southeasterly directed channels in the westernmost Anadarko Basin contain detrital zircons that match local basement from the Amarillo uplift (Granite-Rhyolite terrane age as defined here) but also zircons derived from Yavapai-Mazatzal terranes, suggesting local erosion of the Amarillo-Wichita uplift but also early tectonism of the Ancestral Rocky Mountains (Leary et al., 2017) as well as the initiation of west-to-east paleoslopes on the eastern side of the Anadarko Basin.

East of the Anadarko Basin during the latest Mississippian to Early Pennsylvanian, the Arkoma Basin (Fig. 1) was undergoing rapid flexural subsidence induced by loading from the Ouachita fold-thrust belt to the south (Sutherland, 1988; Housekneckt, 1986). The Atokan section of the Arkoma Basin contains thick packages of deep-water clastics ultimately derived from two main source regions: (1) the Appalachian orogen drained by large-scale river networks flowing longitudinally from east to west, but also rivers flowing from north to south (Xie et al., 2018; Wang and Bidgoli, 2019; Allred and Blum, 2021; Thomas et al., 2021; Lawton et al., 2021); and (2) transverse rivers sourced in peri-Gondwanan terranes south of the Ouachita orogenic belt (Alsalem et al., 2018, 2021; Thomas et al., 2021; Lawton et al., 2021). Detrital zircon data on Atokan sandstones within the Arkoma Basin show a predominance of Grenville basement and Paleozoic grains (Fig. 7B), with smaller groupings of 1600–1800 Ma and 2600–2800 Ma grains (Sharrah, 2006; Thomas et al., 2021). By the end of the Atokan, the Arkoma Basin had mostly filled and was accumulating fluvio-deltaic sediment (Sutherland, 1988).

Our data set does not enable further constraints on dispersal patterns or provenance within the Anadarko region during the latest Mississippian to Early Pennsylvanian. But our provenance data clearly demonstrate that by Middle Pennsylvanian time, the Anadarko Basin formed the ultimate sink for at least two continental-scale fluvial systems that we suggest were linked to the Forest City, Illinois, and Appalachian basins (Fig. 8) based on comparison of our data set with previously published data from those basins (Figs. 7B and 7C). Group B samples (RF2, RF3, and RF4) are from the northern shelf of the Anadarko Basin and the northeastern Cherokee platform (Fig. 3) and lack the prominent Grenville population found in sandstone units farther east that were sourced exclusively form the Appalachian orogen (Fig. 7B; Gehrels et al., 2011; Thomas et al., 2017, 2020, 2021; Kissock et al., 2018). Instead, group B samples are characterized by a large percentage of Neoproterozoic and Paleozoic zircon grains (Fig. 7B). Kissock et al. (2018) documented a similar distribution of zircon ages within the youngest samples in the roughly time-equivalent Floris Formation in the Forest City basin (Fig. 1), and these younger samples (FCB in Fig. 7C) are the most similar to group B on the multidimensional plot (Fig. 7C). Kissock et al. (2018) interpreted a northern Appalachian source region with headwaters in southern New England (Fig. 8) as a source for the Neoproterozoic grains in the youngest Forest City basin samples and suggested that increasing sediment yield derived from that region during the Middle Pennsylvanian buried and surmounted the low-amplitude Wisconsin-Kankakee and Mississippi River arches (Fig. 1), forming a depositional slope stretching from the New England region of the Appalachian orogen through the Michigan and Forest City basins. Alternatively, Thomas et al. (2020) suggested a different routing of a fluvial system from the central Appalachian orogen where Neoproterozoic terranes also occur, through the Appalachian basin, into the northern Illinois basin and the Forest City basin with a possible secondary source farther north in the Appalachian orogen (Fig. 8). The detrital zircon spectra of group B samples in the Anadarko Basin indeed show similarities to those of the northern Illinois basin (Fig. 7C), suggesting this linkage from the Appalachians through the Illinois and Forest City basins and then continuing to the northeastern edge of the Anadarko Basin in Desmoinesian time. However, the older Forest City basin samples do not show a similar composition to group B samples of the Anadarko Basin (Fig. 7C).

By Middle Pennsylvanian time, sedimentation had similarly buried the Nemaha uplift from south to north; previous workers have traced Red Fork channel bodies across the Nemaha uplift with little deflection (Brenner, 1989; Andrews, 1997). If so, then we suggest that a large transverse drainage system with headwaters in the central and/or northern Appalachians ultimately terminated as a deltaic and linked subaqueous system that followed the trend outlined in seismic data along the northern Anadarko shelf by Peyton et al. (1998) and that shifted laterally across the buried Nemaha uplift (Fig. 8).

However, the detrital zircon age spectra of samples in subgroup C1 indicate that a second major river system with a distinctive provenance supplied sediment to the Anadarko Basin. The samples BG1 and RF1 lie within and along trend of the Arkoma Basin, which was overfilled by Middle Pennsylvanian time and provided a sediment pathway from the southern Appalachian basin to the Anadarko Basin (Sutherland, 1988). Subgroup C1 contains a large contribution from Grenville-age grains, consistent with ultimate sourcing in the main Appalachian orogen (Fig. 7B). Furthermore, this subgroup of C1 samples has a similar detrital zircon spectrum compared to younger (Desmoinesian) sandstone samples of the northern Arkoma Basin (Fig. 7B) as well as samples from the southern Illinois basin and southern Appalachian basin (Fig. 7B). All of these samples show compositional similarities in Figure 7C, but the older samples from the Arkoma Basin are slightly different in composition (Figs. 7B and 7C), which Thomas et al. (2021) suggested reflects a transverse source from the southern peri-Gondwanan terranes.

Furthermore, age distributions of zircons from Middle Pennsylvanian strata of the Fort Worth basin (south of the Anadarko Basin; Alsalem et al., 2018) also resemble those of subgroup C1 of the Anadarko Basin (Fig. 7B). Alsalem et al. (2018) also noted that the Sabine uplift (south of the Ouachita fold-thrust belt; Fig. 3) likely formed a second-order source for sediment into the Fort Worth basin but otherwise inferred a long-distance fluvial system that originated in the southern Appalachian orogen and traversed the Arkoma Basin into the Fort Worth basin. We suggest that the same or a similar fluvial system entered the Anadarko Basin on its eastern edge but that the Sabine uplift was not an important source of sediment to the Anadarko Basin at least until the Permian (Soreghan and Soreghan, 2013; Soreghan et al., 2018; Thomas et al., 2021); this is also borne out based on the distance between the Fort Worth basin and subgroup C2 on Figure 7C.

Provenance of Middle Pennsylvanian samples within the depocenter of the Anadarko Basin resembles that of samples from the eastern shelf of the basin (both are part of group C on Fig. 7A), albeit the basin-center samples contain a slightly higher percentage of Neoproterozoic detrital zircons, which is a characteristic of the sandstones from the northern Anadarko and Cherokee shelves (group B). Thus, the basin-center samples (subgroup C2), all of which are interpreted as subaqueous fan deposits, likely reflect a mixing of sediment from fluvial-deltaic systems emanating from the eastern margin with headwaters in the southern Appalachians and fluvial-deltaic systems from the northern and northeastern shelf with headwaters in the northern Appalachians. Accordingly, we suggest that the subaqueous-fan sandstones were derived from a mixed shelfal sand source rather than point sources (incised valleys) because derivation from a single incised valley system should produce a composition that more faithfully reflects the detrital composition of either one or the other fluvial-deltaic system.

Finally, samples CH1 and GY1 (Cy on Figs. 7B and 7C), which are younger than the other Middle Pennsylvania samples, are similar in composition to the slightly older grouping from the eastern margin of the Anadarko Basin (subgroup C1) was well as the basin center (subgroup C2) yet are located in the northeastern part of the Cherokee shelf in the geographic location of the group B samples. Although our data are limited to these two samples, it appears that the fluvial systems flowing westward from the Appalachian orogen after the Middle Pennsylvanian were likely dominated by well-integrated transverse fluvial systems transporting sediment predominantly from terranes of the Grenville basement. This implies that the fluvial systems emanating from the east were more integrated with sources in both the northern and southern Appalachians as they entered the Anadarko Basin, a hypothesis that should be further tested with a broader spatial sampling of Upper Pennsylvanian sandstones.

The Middle Pennsylvanian in the Anadarko Basin was a time of accelerated subsidence and vast sediment input from two continental-scale drainages. But the large basement block to the south—the Wichita uplift, bordered by a fault zone with ~12 km of displacement (Perry, 1989)—supplied surprisingly minimal sediment to the basin. Furthermore, the minimal contribution from Yavapai-Mazatzal–aged grains in the basin center reflects relatively minor sediment influx to the western edge of the basin despite active Early–Middle Pennsylvanian orogenesis in the Ancestral Rocky Mountains (Leary et al., 2017; Hollingworth et al., 2021). The Middle Pennsylvanian also appears to record a significant shift in provenance in basins east of the Anadarko, including the Arkoma (Wang et al., 2022; Thomas et al., 2021), Forest City (Kissock et al., 2018), and Illinois basins (Thomas et al., 2020). Kissock et al. (2018) inferred a tectonic driver to the east; alternatively, perhaps the increased subsidence of the Anadarko Basin at this time initiated a reorganization of these fluvial systems, although this idea requires further testing.

Petrographic data integrated with detrital zircon results from Middle and Upper Pennsylvanian sandstone units in the Anadarko Basin exhibit spatially discrete populations derived from multiple and separate drainage networks entering the Anadarko Basin. One major result of this analysis is that the Wichita uplift—the major structural highland adjacent to the basin and long presumed to have sourced abundant sediment—instead provided spatially and volumetrically minimal sediment to the basin, restricted to the narrow clastic apron fringing the southern basin margin and dominated by an arkosic composition and narrow detrital zircon ages reflecting the local igneous basement. In contrast, the vast majority of the sediment that accumulated in the basin is quartz rich, with variable amounts of polycrystalline and metamorphic lithic fragments, and detrital zircon spectra dominated by Neoproterozoic to Mesoproterozoic ages inferred to reflect sources in the greater Appalachian orogen. However, distinct differences in the detrital zircon age distributions, which are not captured in the basic framework compositions, suggest the influence of two major and distinct fluvial systems entering the basin during the Middle Pennsylvanian: (1) one entering from the north or northeast, connected to fluvial systems passing through the Forest City basin and ultimately headed within the northern Appalachian orogen; and (2) a second fluvial system that passed axially through the Arkoma Basin to the east with ultimate headwaters in the southern Appalachian orogen. This latter system may have connected to the Fort Worth basin, but the actual pathway is poorly constrained. The zircon data from the keel of the Anadarko Basin reflect mixing of these two discrete contributions prior to deposition in subaqueous fans. Although the data are more limited, Upper Pennsylvanian sandstones in the Anadarko Basin reflect a more homogeneous detrital signature, derived predominately from the southern Appalachian orogen. Increased sedimentation emanating from the Appalachian region and reduced regional subsidence may have created a broad, well-integrated drainage system flowing axially through the Appalachian foreland ultimately to the Anadarko Basin by Late Pennsylvanian time. Finally, these results provide additional constraints on the underlying geodynamics of the Anadarko Basin because the bulk of the sediment delivered to the basin was derived from sources that entered the basin axially (from the southeast) and transversely (from the north) with minimal contributions from unroofing of the Wichita uplift bordering the southern basin margin despite the presence of a major (~12 km displacement) fault zone. This result challenges the view of the Anadarko Basin as a classic flexural foreland basin wherein the uplifted load induced subsidence and yielded significant sediment to the basin.

This research was funded by U.S. National Science Foundation grants EAR-1338331 and EAR-1053018. Financial support was provided by the University of Oklahoma Libraries’ Open Access Fund. M.C. Gilbert, J. Price, G. Morgan, and N. Suneson were invaluable resources for petrologic and geologic knowledge of the Wichita Mountains and Anadarko Basin regions. We thank V. Jordan and N. Fedor for sample retrieval and processing at the Oklahoma Petroleum Information Center, and G. Gehrels and personnel of the Arizona LaserChron Center for facilitating detrital zircon analysis. We thank reviewers on an earlier version of this manuscript for their contributions in improving the structure of the manuscript, and Ryan Leary and an anonymous reviewer as well as Associate Editor Nancy Riggs and Science Editor David Fastovsky for their comments and suggestions that greatly improved this version of the manuscript.

1Supplemental Material. Table S1: U-Pb geochronologic data and acquisition parameters for detrital zircon samples presented in this study. Table S2: Raw point count data for samples from this study used in the modal mineralogy analysis. Table S3: Metadata for detrital zircon datasets from published studies used in this study. Please visit to access the supplemental material, and contact with any questions.
Science Editor: David E. Fastovsky
Associate Editor: Nancy Riggs
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.