Detrital Zircon Geochronologic Constraints on Patterns and Drivers of Continental‐Scale Sediment Dispersal in the Late Mississippian

The Late Mississippian was a critical time interval in Laurentia's history, marking the transition from carbonate deposition on a stable platform, during the Early to Middle Mississippian, to extensive clastic deposition in the Pennsylvanian to Permian associated with the Laurentia‐Gondwana collision. In the U.S. midcontinent, Chesterian incised valley fill (IVF) systems that developed within a carbonate‐dominated platform provide new insights on the patterns and drivers of continental‐scale sediment dispersal during this transitional period. Here we report 1,037 new concordant detrital zircon U‐Pb ages from nine samples of Uppper Mississippian sandstone collected from cores in southwestern Kansas and from outcrops in northwestern Arkansas. The sandstones are characterized by major age clusters corresponding to the Grenville (900–1,300 Ma) and Taconic‐Acadian (350–500 Ma) orogenies and minor older age groups, suggesting derivation from the Appalachian region. A compilation of published detrital zircon ages from Early Paleozoic sandstones as well as from temporally equivalent units across North America, including from the Appalachian Foreland Basin, Illinois Basin, Arkoma Shelf, Ozark Dome, Black Warrior Basin, and Grand Canyon, suggests Chesterian sandstone age distributions are distinct from those of older sandstones and are consistent with a change to a dominantly Appalachian age signature that was roughly synchronous across the continent. Together, the new and compiled ages support development of a generally E‐W transcontinental sediment dispersal system in the Late Mississippian that was likely controlled by orogenesis on the eastern Laurentian margin, while local variations in the age signatures appear to be controlled by N‐S drainage networks, influenced by glacioeustatic fluctuations and local structures.

In the U.S. midcontinent, Late Mississippian (Chesterian) sandstones preserve the earliest record of this evolving Carboniferous setting. In Kansas, the sandstones are part of N-S-oriented IVF systems that developed across a broad carbonate platform that covered much of the cratonic interior, including in Kansas, Missouri, and northern Arkansas (Lane, 1978;Manger, 2014;Manger & Zachry, 2009;Montgomery & Morrison, 1999). The sandstones represent an abrupt change to mixed carbonate-siliclastic deposition during what is considered to be a relatively stable tectonic period in the midcontinent, one that predates the major and well-documented Pennsylvanian to Permian deformation across the region associated with the development of the Ancestral Rocky Mountains and of the Ouachita-Marathon fold-thrust belt (Adler et al., 1971;Leary et al., 2017;Montgomery & Morrison, 1999;Rascoe & Adler, 1983).
The presence of IVF systems and clastic detritus in a carbonate-shelf setting during the Late Mississippian raises two interesting issues. First, the provenance of the detritus is unknown. Although discrete sandbodies are preserved across the region, the northern reaches of these IVF systems in Kansas were eroded during Pennsylvanian deformation (Montgomery & Morrison, 1999). From the trend of the preserved sandstones, some workers have argued that sediments were sourced from local structural highs like the Central Kansas Uplift and Cambridge Arch to the north (Cirilo, 2002;Senior, 2012; Figure 1). However, Cambrian through Middle Mississippian carbonates would have covered much of the region and are more likely to have provided carbonate materials as locally derived sediments. The extent to which underlying Precambrian basement rocks would have been exposed as a potential source is unknown, although limited studies have pointed to highs like the Nemaha Ridge as a source of first-cycle zircons in age-equivalent units (Xie, Cains, et al., 2016;Xie, O'Connor, et al., 2016). A second and associated question relates to the role of Figure 1. (a) Distribution of major basement age provinces in North America that may have been sources for Paleozoic sediments (modified from Ojakangas et al., 2001;Park et al., 2010;Gehrels et al., 2011;Xu et al., 2017). (b) Basement structure of the midcontinent (modified from Rascoe & Adler, 1983). Contours are in feet (1 feet ≈ 0.3045 m). more distant clastic sources on the provenance of IVF systems and Upper Mississippian sandstones in the region. Gehrels et al. (2011) proposed a transcontinental sediment transport model whereby sandstones in the western part of Laurentia were sourced from the Appalachian orogen through a combination of westward flowing rivers, wind systems, and ocean currents. Although the Appalachian region is widely accepted as a source for Late Paleozoic sediments (Gehrels et al., 2011;Xie, Cains, et al., 2016;Xie, O'Connor, et al., 2016;Alsalem et al., 2017;Thomas et al., 2017;Bidgoli et al., 2018;Kissock et al., 2018;Jones et al., 2019;, the details of such long distance sediment transport require further study (see discussion in Thomas, 2011). For example, the role of recycling of Early and Middle Paleozoic strata along the hypothetical E-W route across Laurentia has not been fully explored. Sediment dispersal paths from the Appalachian region also require further investigation, as the N-S orientation of IVF systems in Kansas appears at odds with the E-W transport model. This suggests more complicated drainage systems and sediment routing from source to sink.
This study examines the provenance of Upper Mississippian (Chesterian) sandstones using detrital zircon U-Pb geochronology. The data are aquired from seven samples obtained from boreholes in the Hugoton Embayment, in southwestern Kansas, that offer a rare opportunity to examine IVF systems, now buried >1,500 m in the subsurface. We also report ages obtained from Chesterian sandstones in northwestern Arkansas that represent some of the earliest pulses of clastic sediments to Arkoma Shelf (Winkelmann, 2007;McGilvery et al., 2016;Xie, O'Connor, et al., 2016). The new ages are compared to published detrital zircon ages from Early and Middle Paleozoic sandstones as well as from age-equivalent units across North America to evaluate patterns and drivers of sediment dispersal during this critical time interval. The analysis also provides insight into the relative roles of orogenesis, eustacy, and local topograpy in transcontinental sediment transport.

Geologic Background
The North American Craton (later the paleocontinent of Laurentia) occupied the center of the supercontinent Rodinia from the Late Proterozoic until breakup in the Precambrian-Cambrian. The multistage breakup of the supercontinent resulted in the opening of the Iapetus Ocean and the development of a passive margin along the southern and eastern margins of Laurentia (Aleinikoff et al., 1995;Park et al., 2010;Thomas et al., 2017). By the Ordovician, the majority of the craton was covered by shallow seas. Beginning in the Middle Ordovician, Appalachian orogenesis initiates and interrupts passive margin sedimentation along the eastern Laurentian margin. Elsewhere, in the interior of the craton, stable conditions continued with widespread deposition of shelf carbonates and intervening shales (Manger, 2014;Xie, O'Connor, et al., 2016). This pattern of deposition in the cratonic interior persisted until initiation of the Late Mississippian Alleghanian orogeny on the eastern margin of Laurentia, which increased the flux of clastics to the interior of the craton (Winkelmann, 2007;McGilvery et al., 2016;Xie, O'Connor, et al., 2016). Orogenesis continued on the margins of the craton throughout the Pennsylvanian, with development of the Appalachian-Ouachita-Marathon belt and associated uplifts, leading to the final assembly of the supercontinent of Pangaea in the Permian (Hatcher et al., 1989;Park et al., 2010;Thomas et al., 2017).
The Hugoton Embayment in southwestern Kansas is located in the center of the North American Craton and is considered as the shallow subsurface extension of the Anadarko Basin ( Figure 1). In the Early and Middle Mississippian, the entire midcontinent was covered by epeiric seas and characterized by deposition of Kinderhook, Osage, and Meremec series carbonates and shales on a broad and relatively shallow platform (Figure 2;Rascoe & Adler, 1983;Montgomery & Morrison, 1999). During the latest Meramecian, sea level regression led to widespread subaerial exposure and fluvial incision of the carbonate platform, resulting in the development of deep, linear valleys across the Hugoton Embayment (Montgomery & Morrison, 1999;Cirilo, 2002;Senior, 2012;Figures 2 and 3). These valleys were subsequently filled, during a Late Mississippian (Chesterian) transgression, with dominantly fineto very fine-grained sandstones. Near the center of the Hugoton Embayment, one such valley has been recognized in 3-D seismic, well log, and core data (Figures 2 and 3). The valley, more than 70 m deep in places, can be traced for over 160 km from Kansas into Oklahoma (Montgomery & Morrison, 1999;Cirilo, 2002;Senior, 2012). The valley fill is interpreted to have been deposited in an estuarine system, with more fluvial influence to the north and more tidal and marine influence to the south (Cirilo, 2002;Montgomery & Morrison, 1999;Senior, 2012), a pattern that is well exemplified by core and well log-based lithofacies from boreholes in three oilfieldsthe Pleasant Prairie South, Eubank, and Shuck fields-located along the axis of the valley (Figures 2 and 3).
In the Pleasant Prairie South oilfield, Mary Jones #2 is one of only two boreholes with core from the Chesterian IVF, which consists of upward-fining sequences that grade from conglomerate to sandstone and siltstone. The sandstones are mainly well-sorted quartzarenites. Sedimentary structures and features in the core include laminations, trough cross-beds, and carbonized organic material (wood fragments), indicating a fluvial origin, and mud drapes, indicating minor tidal influence (Cirilo, 2002;Dubois et al., 2014;Senior, 2012).
The only core available for direct observation and sampling of the Eubank field comes from the MLP Black borehole that reveals a succession of tidally influenced estuarine deposits, with five major facies including marine shale, transgressive conglomerate, intertidal sand flat, estuarine bar margin, and estuarine bar facies (Dubois et al., 2014b). The estuary bar sandstones are the main oil-producing reservoir in the field and consists of fineto very fine-grained quartzarenites and sublitharenites that are moderately to well sorted with subrounded to rounded grains (Dubois, Williams, Youle, & Hedke, 2014b;Montgomery & Morrison, 1999).
Unlike the northern oilfields, the Shuck Oilfield is located in a deeper part of the Hugoton Embayment, and the youngest rocks of the transgressive fill were deposited beyond the extents of the incised valley. The lithofacies across the field are similar to the Eubank Field, with major lithofacies identified as marine shale, transgressive conglomerates, valley margin conglomerates, salt marsh, tidal flat, and estuarine bar facies (Dubois, Williams, Youle, & Hedke, 2014c). The estuary bar facies consist of fineto very fine-grained, limey sandstone, interpreted to have been deposited in migrating sand bars in a subtidal estuary.

Geochronologic Provinces
Provenance analysis based on detrital zircon geochronology requires detailed knowledge of the age of potential source regions. Earlier geochronologic investigations provide a robust foundation for provenance analysis in North America (e.g., Gehrels et al., 2011;Sims, 1996;Van Schmus et al., 1993;Whitmeyer & Karlstrom, 2007). The North American Craton formed billions of years ago through the collision of microcontinental volcanic arcs and oceanic terrains (Hoffman, 1989). These collisions and accretionary events formed six major age provinces that may be sources of sediments. They include the Superior, Wyoming, Trans-Hudson, Penokean, Yavapai-Mazatzal, and midcontinental Granite-Rhyolite provinces (Whitmeyer & Karlstrom, 2007;Gehrels et al., 2011; Figure 1). In the following sections, we organize age provinces by geography.

Northern Provinces
The Archean basement of the Canadian Shield occupied the central part of the Laurentian Craton and was formed from microcontinent collisions in the Paleoproterozoic (1,960-1,800 Ma) that eventually formed a major orogenic belt (Whitmeyer & Karlstrom, 2007). The Superior and Wyoming provinces, the southernmost components of the Canadian Shield, are the closest Archean sources to the study area ( Figure 1a). The Superior province was formed by the collision of a 2,700-2,800 Ma arc with an Archean block of 3,500 Ma (Hoffman, 1989 and references therein). The Wyoming province mainly consists of 2,500-2,700 Ma granite and gneiss. The Trans-Hudson province, between Superior and Wyoming provinces, is an orogenic belt made up of 1,800-1,900 Ma metasedimentary rocks (Hoffman, 1989 and references therein). Similar age magmatic rocks are also distributed along the southern margin of the Superior province, in the Penokean province (Sims, 1996;Van Schmus et al., 1993).

Eastern Provinces (Appalachian Region)
The Appalachian region is characterized by the superposition of multiple tectonic events that include the Grenville orogeny, Late Proterozoic to Early Paleozoic rifting and magmatic events, and the Paleozoic Appalachian orogenic series (Taconic, Acadian, and Alleghenian orogenies; Hatcher et al., 1989;Hauser, 1996;Aleinikoff et al., 1995;Hatcher, 2005;Cawood & Nemchin, 2001;Park et al., 2010;Thomas et al., 2017). The Grenville province (900-1,300 Ma) is the residual basement of a large orogenic belt that formed from continental collision during the formation of the Rodinia supercontinent (Tohver et al., 2006;Whitmeyer & Karlstrom, 2007). Synrift magmatic rocks are related to the breakup of Rodinia and a series of rifts and associated magmatic events in Late Proterozoic to Early Paleozoic (Aleinikoff et al., 1995;Hatcher, 2005;Lukert & Banks, 1984;Wehr & Glover, 1985). Taconic and Acadian orogenic terranes are also distributed along the eastern margin of North America (Park et al., 2010) and resulted from collisions between east Laurentia and an arc system and the Avalonia terrane in the Ordovician (~465-445 Ma; McLennan et al., 2001) and Devonian (~400-350 Ma), respectively. During these two orogenic events, volcanic and metamorphic activities were continuous (from ca. 350 to 500 Ma; Becker et al., 2005 ;McLennan et al., 2001). These terranes are inferred to have provided considerable sediment to the Appalachian foreland clastic wedge (McLennan et al., 2001;Becker et al., 2005).

Midcontinent Provinces
The Midcontinent consists of the Yavapai-Mazatzal (1,600-1,800 Ma) and Granite-Rhyolite (1,300-1,550 Ma) provinces. The Yavapai-Mazatzal province is a combination of two basement provinces. The Yavapai province is made up of 1,760-1,700 Ma arc rocks that were accreted to the North American continent at about 1,700 Ma, whereas the Mazatzal province is mainly composed of 1,800-1,700 Ma accretionary wedge rocks that were added to the southern Yavapai province in the latest Paleoproterozoic (Bowring & Karlstrom, 1990;Van Schmus et al., 1993). The adjacent Granite-Rhyolite province was distributed along the southeastern margin of the Laurentian Craton-from present-day northern Mexico to northeastern Canada-with an age range 1,550-1,300 Ma (Whitmeyer & Karlstrom, 2007).

Rift-related igneous and sedimentary rocks associated with the Precambrian Midcontinent Rift System and
Cambrian Southern Oklahoma Aulocogen also contribute to the fabric of the basement in North America (Cannon, 1994;Donaldson & Irving, 1972;Hauser, 1996). The Midcontinent Rift System is made up of bimodal igneous rocks that reflect a short-lived (1,115-1,085 Ma) pulse of back-arc extension and magmatism during the Grenville orogeny (Ojakangas et al., 2001). Cambrian intrusive and volcanic rocks, including gabbro and basalt, also developed along the Southern Oklahoma fault system (Arbuckle and Wichita uplifts in Figure 1b), collectively known as the 550-530 Ma Southern Oklahoma Aulocogen (Bowring & Hoppe, 1982;Hanson et al., 2013;Hogan & Gilbert, 1998;Thomas, 2014;Thomas et al., 2016;Wright et al., 1996).

Samples and Methods
Seven core samples were collected from three boreholes, the Mary Jones #2, MLP Black, and Hitch Unit 8-3 boreholes, located in the axis of the IVF complex in the Pleasant Prairie South, Eubank, and Shuck oilfields, respectively ( Figure 3). Figure 3 shows the log characteristics, lithofacies interpretations based on the log properties, and generalized stratigraphy of the sampled boreholes. The figure also shows the relative positions of core and selected samples in each well. Core samples were selected from sand-rich intervals and were generally well-sorted, subrounded, fineto medium-grained quartzarenites ( Figure 3). We also collected two samples of age-equivalent units, the Chesterian Batesville and Wedington sandstones, from outcrops in northwestern Arkansas. The details of the core and outcrop samples are provided in Table 1.
Samples were processed using standard mechanical, electromagnetic (Frantz iodynamic magnetic separator), and heavy liquids (methylene iodide) techniques for separation and concentration of zircon. For core samples, an additional processing step was to wash and remove any coring fluid or hydrocarbon residues. Once separated, zircons from sample splits were evaluated under binocular microscope, mounted onto epoxy pucks (25-mm diameter) using double-sided tape (tape mount) for U-Pb spot analysis, and photographed for archive.
Laser ablation inductively coupled plasma mass spectrometry analyses were carried out at UTChron Geoand Thermochronometry Laboratories at the University of Texas at Austin and Isotope Geochemistry Laboratories at the University of Kansas. GJ1 with a 206 Pb/ 238 U age of 600.4 ± 0.65 Ma is the primary reference material used to address calibration drift and both downhole isotopic and elemental fractionation (Jackson et al., 2004) and was analyzed once for every five to eight unknowns. Pak 1 was used as a secondary zircon standard with thermal ionization mass spectrometry ages of 43.0 Ma for the analysis at University of Texas at Austin. The Plešovice and Fish Canyon Tuff zircons with the age of 337.13 ± 0.37 Ma (Sláma et al., 2008) and 28.402 ± 0.023 Ma (Schmitz & Bowring, 2001;Wotzlaw et al., 2013) reported from chemical abrasion thermal ionization mass spectrometry were analyzed as the secondary reference materials for the analysis at the University of Kansas.
U-Pb data reduction of the samples from Kansas and Arkansas was completed using Iolite (Hellstrom et al., 2008;Paton et al., 2011) and the open-source software package ET Redux (McLean et al., 2016), respectively. 206 Pb/ 238 U ages were used for grains younger than 850 Ma, and 207 Pb/ 206 Pb ages were used for grains older than 850 Ma. All ages are reported using 2σ absolute propagated uncertainties. Standard GJ1 yields a 206 Pb/ 238 U weighted average age of 601.8 ± 2.4 Ma (mean square weighted deviation (MSWD) = 0.75, n = 51). The Plešovice and Fish Canyon Tuff zircon references yielded weighted mean 206 Pb/ 238 U dates of 339.8 ± 7.3 Ma (MSWD = 0.91, n = 20) and 28.93 ± 0.71 Ma (MSWD = 1.4, n = 20), respectively, and fall into the uncertainty ranges of the reported ages. For zircon U-Pb ages older than 850 Ma, the discordance was calculated based on 206 Pb/ 238 U and 207 Pb/ 206 Pb ages. For zircon ages younger than 850 Ma, the discordance was calculated based on 206 Pb/ 238 U and 207 Pb/ 235 U ages. For the Kansas IVF core samples, grains with greater than 10% discordance or 5% reverse discordance were excluded. For the Arkansas outcrop samples, which were generally poorer in data quality, we use lower discordance (20%) and analytical error (15%) cutoffs for older grains (>850 Ma).
To compare the new detrital zircon data with published data sets, we use a range of visualization and statistical tools in our analysis. The age data are presented and qualititatively compared using probability density (Ludwig, 2008) and kernel density estimation plots (Vermeesch, 2012;2013;Vermeesch et al., 2016) and pie charts. More quantitative comparison of samples is accomplished through cumulative probability plots, which can reveal differences in the proportions of grains among samples (Gehrels et al., 2011), and through Kolmogorov-Smirnov (K-S) tests (Press et al., 1986), which are a measure of the statistical significance of differences between age distributions, with P values as the major criterion and 0.05 as the cutoff. A P value less than the cutoff suggests that age distributions of two samples are not the same at a confidence level >95% (Guynn & Gehrels, 2006). Multidimensional scaling (MDS) analysis (Vermeesch, 2013;Vermeesch et al., 2016) was also used to evaluate similarities and differences among samples through grouping of samples with similar U-Pb age distributions.

Results
Of the 976 detrital zircon grains analyzed from core samples from southwestern Kansas, a total of 865 grains yielded ages within the discordance and error criteria. The full suite of isotopic measurements, U-Pb ages (Supporting Information Tables S1 and S2), and associated Concordia plots ( Figure S1) are provided in the supplement. The detrital zircon age results are plotted as probability density and kernel density estimation plots (Ludwig, 2008;Vermeesch, 2012) in Figure 4, with the range of known age provinces highlighted.
In general, the samples show major groups of Middle to Late Mesoproterozoic (~1,300-900 Ma) and Paleozoic (~500-350 Ma) ages and more minor groups of Late Paleoproterozoic (~1,800-1,600 Ma), Early Mesoproterozoic (~1,550-1,300 Ma), and Archean (>2,500 Ma) ages. Visual inspection of the age spectra of the samples and K-S statistical comparisons (Table 2), with P values generally less than 0.05, suggest they are statistically indistinguishable. The MDS plots ( Figure S2) also group the samples together and suggest a high degree of similarity between age spectra. Outcrop samples from northwestern Arkansas yielded a total of 172 grains within discordance and error criteria. The age results, while similar to the Kansas sample in terms of the major age groups (~1,300-900 and~500-350 Ma), also contained abundant numbers of Late Paleoproterozoic (~1,800-1,600 Ma) and Early to Middle Mesoproterozoic (~1,550-1,300 Ma) ages and minor groups of Neoproterozoic to Early Cambrian (650-500 Ma) and Paleoproterozoic and older grains (>1,800 Ma). The details of the age results are described below.

Pleasant Prairie South Field, Mary Jones #2
Two samples of fine-grained sandstone were analyzed from the Mary Jones #2 borehole ( Figure 4). However, one grain from the Mary Jones 5229 sample resulted in a concordant age of 191.7 ± 3.6, much younger than the depositional age of the unit and likely the result of sample contamination, perhaps from rock cuttings derived from higher in the borehole being circulated downhole. Therefore, this age is excluded from our analyses. Overall, the two samples yield very similar age distributions (Figure 4a), with pronounced groups of Middle to Late Mesoproterozoic (ca. 1,300-900 Ma) and Paleozoic (ca. 500-350 Ma) ages. A total of 108 zircons from the Mary Jones 5229 sample yielded ages ranging from 364 ± 22 to 3,267 ± 14 Ma, with thẽ 1,300-900 Ma age group accounting for 65% and a subordinate~500-350 Ma age group accounting for 9% of the ages. The 120 zircons from the Mary Jones 5194 sample range in age from 359 ± 7 to 2,881 ± 20 Ma, with the same dominant population of grains of~1,300-900 Ma accounting for 60% and the subordinate cluster (~500-350 Ma) accounting for 15% of grains. K-S statistical comparison of the two samples yields a P value of 0.507 (>0.05; Table 2), suggesting that the two samples have similar age distributions (Gehrels et al., 2011;Press et al., 1986).

Provenance of Chesterian IVF Sandstones
Detrital zircon age distributions from Chesterian core samples from the Hugoton Embayment have pronounced age clusters of 900-1,300 and 350-500 Ma that correspond to the age ranges of the Appalachian Grenville and Taconic-Acadian orogenies, respectively. The bimodal pattern, observed in the histogram and probability density plots (Figure 4), along with the K-S tests (Table 2) and MDS analysis ( Figures S2  and S3) indicate that all seven samples are similar and were likely derived from the same provenance.
In Figure 4c, the age data from Chesterian core samples are combined into a single kernel density estimation plot (Vermeesch, 2012). In addition to the pronounced Grenville and Taconic-Acadian age groups, the age spectrum shows that zircons sourced from northern and midcontinent age provinces are rare or absent in the IVF sandstones. The low content of Yavapai-Mazatzal (1,800-1600 Ma) and Granite-Rhyolite (1,550-1,300 Ma) ages, in particular, excludes major derivation from local midcontinent basement. The results seem to suggest that structural highs like the Central Kansas Uplift and Cambridge Arch remained covered during the Late Mississippian, a conclusion also supported by structural analysis that suggests that the main phase of deformation and potential breaching of these structures occurred later, in the Pennsylvanian (Adler et al., 1971;Rascoe & Adler, 1983).
The strong Grenville and Acadian-Taconic age peaks observed in the zircon age spectra for the Chesterian IVF sandstones point to more distant or potentially recycled sources for sediments. Both the Grenville and Taconic-Acadian orogenies occurred in the Appalachian region, and comparison of our data with data from the Alleghanian Appalachian Foreland Basin, a proxy for sediment sources and likely zircon age distributions derived from the orogen, shows a high degree of similarity (Thomas et al., 2017 and references therein). The almost mirror image in age spectra between sandstones from the Hugoton Embayment and Appalachian signature derived from synorogenic clastic wedge sediments (Thomas et al., 2017), shown in Figure 4c, strongly suggests that the Appalachian region was the primary source of sediments. However, it is worth noting that the percentages of Yavapai-Mazatzal grains (1,800-1,600 Ma; up to 10%) in some samples are higher than in Appalachian sources and may point to some limited contributions from local basement uplifts such as Nemaha Uplift (Xie, O'Connor, et al., 2016). The high percentages of the Grenville grains may also reflect contributions from the Midcontinent rift system, which contains bimodal igneous rocks that formed during a short-lived Grenville-age pulse (1,115-1,085 Ma; Ojakangas et al., 2001).

Recycling of Early Paleozoic Sandstones
Although an Appalachian age signature is clear in the Chesterian IVF sandstones, zircons are durable and can be eroded and transported through multiple sedimentary cycles. Thus, it is necessary to also evaluate the possibility of recycled grains in our age data sets. Several pre-Mississippian sandstones that could be sources of zircons are present west and northwest of the Hugoton Embayment ( Figure 5). These sandstones include Cambrian and Lower Ordovician quartz arenites (e.g., Jordan sandstone; Runkel et al., 2007) and the widely distributed Middle Ordovician St. Peter sandstone, a supermature arenite (Konstantinou et al., 2014;Figure 5). A compilation of published detrital zircon geochronologic data from these Early Paleozoic sandstones is presented in Figure 5.
The age spectrum for Cambrian sandstones shown in Figure 5a is compiled from seven samples collected in Wisconsin, Minnesota, and Missouri that have similar age distributions (Konstantinou et al., 2014;Li, 2016). The Archean component (>2,500 Ma) dominates the age spectrum whereas other ages groups, particularly the Grenville (900-1,300 Ma), are limited. By comparison, the zircon data from five samples of Early Ordovician sandstone are bimodally distributed, with both Archean (>2,500 Ma) and Grenville (1,300-900 Ma) ages strongly represented (Figure 5b; Konstantinou et al., 2014;Li, 2016). The pattern, with strong Archean and Grenville age groups, is repeated for the Middle Ordovician St. Peter sandstone, the most well studied and areally extensive of the Early Paleozoic sandstones (Konstantinou et al., 2014;Ibrahim, 2016;Figures 5c-g). It should be noted that there is some geographic variability in the age spectra for the St. Peter sandstone, particularly for zircons <2,000 Ma and most notably for the Grenville age group (Figures 5c-g). Comparison of these age spectra with the compiled age distribution for the Taconic synorgenic wedge (Gray & Zeitler, 1997;Cawood & Nemchin, 2001;McLennan et al., 2001;Eriksson et al., 2004;Park et al., 2010;Thomas et al., 2017) suggests that the Appalachian area was an important source of sediments and Grenville-age zircons during the Early to Middle Ordovician; however, its influence and contributions were not uniform.
Overall, the compilation suggests that the Superior province in the north was an important and continuous provenance area for Early and Middle Paleozoic sedimentary rocks that is largely absent in the Appalachian foreland and in the Chesterian samples from this study. This difference precludes the possibility of recycling of Early Paleozoic sedimentary rocks as a dominant source of sediments to the Hugoton Embayment and Arkoma Shelf. The compilation also demonstrates that Grenville grains are rare or absent in the Cambrian sandstones and that transportation of sediments from the Appalachian area initiated in the Early Ordovician, and its dominance as a sediment source persisted through the Paleozoic.

Comparison with Age-Equivalent Units Across North America
In contrast to the Pennsylvanian, with intense tectonic activity, widespread clastic sedimentation, and generally more complicated local environments (Sloss, 1963;Manger & Zachry, 2009;Thomas, 2011;Manger, 2014;Xie, O'Connor, et al., 2016;Alsalem et al., 2017;Xie, Buratowski, et al., 2018, Xie, Anthony, & Busbey, 2018, the Late Mississippian was a relatively stable period with only limited clastic flux, particularly in the midcontinent (Manger & Zachry, 2009;Manger, 2014;McGilvery et al., 2016;Xie, O'Connor, et al., 2016). In the following sections, we compile and summarize published detrital zircon data from Middle to Late Paleozoic basins that contain Upper Mississippian clastic strata. We focus on sandstones from the Appalachian Foreland Basin, Black Warrior Basin, Illinois Basin, Ozark Dome, Arkoma Shelf, and Grand Canyon and use the data to trace changes in age distributions and provenance across Laurentia in Late Mississippian.

Appalachian Foreland
The Late Paleozoic Appalachian orogen is the principal source of clastic sediments transported westward into the Appalachian Foreland Basin and farther west into the interior of the craton (Gehrels et al., 2011;Park et al., 2010;Thomas et al., 2017). Nearest this presumed source area, Ordovician Taconic synorogenic clastic wedges show unimodal age distributions of Grenville grains (Figure 5h), whereas Devonian Acadian

10.1029/2019GC008469
Geochemistry, Geophysics, Geosystems synorogenic clastic wedges and subsequent Upper Mississippian clastic wedges show pronounced bimodal distributions of Grenville and Taconic-Acadian ages. The age distributions suggest the Appalachian orogen remained a stable sediment source region, yielding zircon grains with similar age distribution with negligible variation since Devonian (Figure 6).

Black Warrior Basin
The Black Warrior Basin, in the southern midcontinent, is the easternmost foreland basin to the Ouachita orogenic system. The basin resides near the intersection of this system with the southern Appalachian orogenic system and thus, the provenance of Chesterian sandstones within the basin is debated (Mars & Thomas, 1999;Xie, Cains, et al., 2016). Detrital zircons from the the Lewis sandstone, at the base of the Chesterian section, contain Taconic-Acadian and Grenville grains that define two prominent age peaks, consistent with major derivation from the Appalachian region (Xie, Cains, et al., 2016; Figure 6). The unit also contains two major clusters of Granite-Rhyolite and Yavapai-Mazatzal ages, suggestive of more local influence on the provenance of the sandstone.

Illinois Basin and Ozark Dome
The Illinois Basin is an intracratonic basin, separated from the Appalachian Foreland in the east by the Cincinnati Arch (Quinlan & Beaumont, 1984;Root & Onasch, 1999; Figure 7). The basin is located across the transcontinental drainage system proposed by Gehrels et al. (2011). Due to its proximity to the  Thomas et al. (2017) and references therein; Illinois Basin spectra are from Rothschild et al. (2016) and Kissock (2016); Ozark Dome data are from Li (2016); and Arkoma Shelf data are from Xie, Cains, et al. (2016), Pickell (2012, and this study. Appalachian orogen, sandstones developed in this basin should be sensitive to sediments sourced from the Appalachian foreland. For example, detrital zircon data of the Devonian Dutch Creek sandstone, a set of quartz arenites developed in a shelf environment, are characterized by major Archean (67%) and subordinate Grenville (16%) age clusters, suggesting that sediment may be recycled from the Ordovician St. Peter Sandstone (Wallenberg et al., 2016). The Lower Mississippian Borden siltstone is dominated by Granite-Rhyolite ages (1,550-1,300 Ma; 68% of grains) with a peak at 1,460 Ma that matches the age of the Wolf River Batholith in the north, a local source for sediments that was exposed during Early Carboniferous (Gregorich et al., 2018). By comparison, Late Mississippian (Chesterian) incised valley systems, similar to those in the Hugoton Embayment, are dominated by Taconic-Acadian and Grenville grains. The Aux Vases sandstone at the base of Chesterian is incised into St. Genevieve limestone and marks the first appearance of these Appalachian-dominated sediments (Rothschild et al., 2016).
Similar provenance shifts also occurred on the margin of the Ozark Dome, west of the Illinois Basin (Li, 2016). The Aux Vases sandstone, present on the eastern limit of the dome, is a medium-grained orthoquartzite, interbedded with limestone and shale (Thompson, 1995) that appears to be correlated with similar strata in the Illinois Basin. Detrital zircon U-Pb ages from the Aux Vases are dominated by Taconic-Acadian and Grenville grains that are distinct from the Upper Devonian Bushberg sandstone, which is characterized by pronounced Grenville and Archean age peaks (Li, 2016; Figure 6).

Arkoma Shelf
The Arkoma Shelf, south of the Ozark Dome, was also characterized by carbonate and shale deposition in the Late Mississippian, prior to Laurentia-Gondwana convergence and development of the Arkoma Foreland Basin (Manger & Zachry, 2009;Manger, 2014;Xie, O'Connor, et al., 2016). During this period, two sets of Chesterian sandstones developed and are interbedded with carbonates and shales, among which the Batesville sandstone is interpreted as the first major pulse of quartz-rich clastic detritus, derived from the north-northeast by coastal currents (Handford, 1995;McGilvery et al., 2016). The stratigraphically younger Wedington sandstone formed as a constructive delta that dispersed clastic sediments from northwest to southeast across a part of the shelf (Winkelmann, 2007;Xie, O'Connor, et al., 2016; Figure 7). Compilation of Chesterian data from this study and from Xie, Cains, et al. (2016) suggests that the Archean-dominated age distribution of Early Paleozoic sandstones (Pickell, 2012) gives way to the bimodal "Appalachian" age distribution in the Late Mississippian and suggests that the Arkoma Shelf experienced a provenance change similar to other sites in North America (Figure 6). 6.3.5. Grand Canyon Farther west, the age spectrum of Early Paleozoic and Devonian rocks from the Grand Canyon is dominated by the regional basement made up of the Yavapai-Mazatzal and Granite-Rhyolite provinces. In contrast, the age distribution of the upper Mississippian Surprise Canyon Formation, like other age-equivalent units across Laurentia, changes dramatically to a more bimodal distribution, characteristic of the Appalachian orogen ( Figure 6). These Chesterian units in the Grand Canyon also consist of nonmarine and marine sediments that fill erosional valleys cut into the Lower to Upper Mississippian Redwall limestone (Billingsley & Beus, 1985), similar to IVF systems in the Hugoton Embayment. 6.3.6. Summary As illustrated in Figure 6, the detrital zircon age distributions of pre-Mississippian (mostly Devonian) sandstones are variable, whereas Upper Mississippian sandstones across the continent have age spectra that mirror the signature of the Appalachian region. We speculate that these discontinuous and sporadically preserved sandstones may have been components of a transcontinental sediment dispersal system that initially developed in the Late Mississippian.

Spatial Analysis and Constraints on Sediment Distribution Pathways
K-S tests, MDS plots, and cumulative probability and kernel density estimation plots are used to evaluate similarities and differences between age distributions of the Chesterian IVF sandstones and temporally equivalent strata (Table 3; Figures S3, 8, and 9). As shown in these plots, detrital zircon ages from Late Mississippian sandstones across North America are similar and are dominated by Grenville and Taconic-Acadian grains (50-73%). The shift to these relatively uniform age distributions and provenance, especially when compared to Middle Paleozoic sandstones, seems to occur simultaneously across Laurentia (Figure 6), suggesting a shared provenance region, most likely in the Appalachian region (Gehrels et al., 2011;Xie, O'Connor, et al., 2016;Xie, Cains, et al., 2016). However, as illustrated in Figures 6, 8, and 9 and via statistical analysis (Table 2 and Figure S3), age distributions have some variability, suggesting that sediment dispersal from the Appalachian region was more complicated than would be expected from a simple river system draining the orogen from east to west.
In the presumed regional source area of the Appalachian Foreland Basin, the Late Mississippian temporal equivalents are synorogenic clastic-wedge deposits of the Mauch Chunk Group (Park et al., 2010;Thomas, 2011;Thomas et al., 2017). In Figure 9, it can be seen that Mauch Chunk samples (Park et al., 2010;Thomas et al., 2017), from different parts of the foreland basin, all contain large numbers of Grenville grains, commonly greater than 50%. However, variations exist in the relative proportions of Grenville and Taconic-Acadian age zircons that could be interpreted to reflect either recycling of sediments derived from the orogenic basement or sediments directly sourced from the Taconic-Acadian synorogenic magmatic and metamorphic rocks (Park et al., 2010;Thomas et al., 2017). For example, from south to north, the percentage of 1,600-1,800 Ma grains diminishes from 13% to 4% and to 2% and >1,800 Ma grains decreases from 13% to 7% and to 1% (Figure 9). These changes in the relative quantity of grains could suggest that the northernmost samples of the synorogenic clastic wedge, made up predominantly of Grenville grains and generally lacking grains older than 1,500 Ma, may be more representative of the "Appalachian signature" defined by Thomas et al. (Thomas et al., 2017;Figure 4). However, more detailed analysis of the roles of zircon fertility of the source rocks and mixing and dilution effects during transport must be accounted for. The data from the IVF sandstones in the Hugoton Embayment show a relatively large percentage of Grenville (57%) and Taconic-Acadian grains (16%), but only 8% of grains have ages >1,500 Ma (Figure 9). K-S tests suggest these samples are most similar to the northernmost samples of the Mauch Chunk Group in the Appalachian Foreland Basin and to samples of the Aux Vases sandstone from the margin of the Ozark Dome (the basal Chesterian sandstone interval also developed in the Illinois Basin). The ages from the IVF sandstones are also similar to age data from the sample of the Wedington sandstone analyzed in this study. These similarities are also illustrated in the MDS plot through grouping of samples ( Figure S3). It should be noted that published data for the Wedington sandstone from Xie, O'Connor, et al. (2016) also show large percentage of Grenville grains (52%) but fewer Paleozoic grains. Differences between our Wedington sample and that of Xie, O'Connor, et al. (2016) may relate to number of grains analyzed (N = 559 vs. N = 82 in this study). By comparison, the other Upper Mississippian samples show more complex age distributions, with fewer Grenville grains (32-46%) and higher and relatively uniform proportions of Yavapai-Mazatzal grains (15-19%), similar to samples from the southern Appalachians (Figures 8 and 9). These age distributions have composite characteristics that suggest influence from both the regional Appalachian provenance and the local midcontinent provenance associated with structural highs like the Nemaha Uplift and the Ozark Dome, which may have been exposed in the Late Mississippian (Xie, O'Connor, et al., 2016;Xie, Cains, et al., 2016).
The Early to Middle Mississippian was characterized by epeiric seas that covered much of the Laurentian interior (Manger & Zachry, 2009;Manger, 2014;Xie, O'Connor, et al., 2016). This is interpreted to have been a tectonically quiescent period on the eastern margin (between Acadian and Allegheny orogenies; Park et al., 2010). During this period, thick and extensive carbonate deposits developed across the midcontinent. This carbonate/shale blanket offers a relatively simple background for sediment dispersal in the Late Mississippian ( Figure 10). From the Late Mississippian onward, the collision between Laurentia and Gondwana resulted in intense deformation and uplift along the eastern margin of the craton (Hatcher et al., 1989). The resulting topographic gradient across Laurentia may have created the conditions necessary for the birth of one or more large-scale westard-flowing river systems. The development of these Late Mississippian river systems may have been similar to the birth of the Yangtze River and other rivers associated with the uplift of Tibet (Zheng et al., 2013;Wang et al., 2017Wang et al., , 2018. The domination of the Appalachian age signature in Upper Mississippian sandstones across the continent, and strong similarity between samples from the Hugoton Embayment, Illinois Basin, and Arkoma Shelf and northernmost Appalachian synorogenic wedge sediments, suggests that a relatively uniform transcontinental river system likely developed in the north. The existence of such a river system could explain why N-S-oriented, southward-flowing drainage systems across the midcontinent lack the Archean signature of the northern age provinces and of Early Paleozoic sandstones and instead contain sediments more consistent with a source in the Appalachian orogen (Figures 9 and 10).
The more variable age distributions documented for other Upper Mississippian units (e.g., Chesterian in Black Warrior Basin and Batesville sandstone on the Arkoma Shelf) may be a direct influence of the Southern Appalachians and restricted structural highs such as the Ozark Dome and the now buried Nemaha Uplift (Xie, O'Connor, et al., 2016;Xie, Cains, et al., 2016). Paleotopography associated with these structures may have blocked the formation of more direct drainages and favored development of more complicated sediment dispersal networks and patterns, such as has been documented for the Batesville sandstone (Handford, 1995), that share some similarity with strata from the Black Warrior Basin (Figures 8 Figure 9. Relative abundance of zircons age populations. The blue arrows represent the Mauch Chunk-Pottsville clastic wedge (Thomas et al., 2017). Abbreviations: ArB = Arkoma Basin; ARM = Ancestral Rocky Mountains; BW = Black Warrior Basin; CU = Cincinnati Uplift; NR = Nemaha Ridge; OD = Ozark Dome; SOA = Southern Oklahoma Aulacogen. and 9; Table 3). Likewise, the provenance of the Grand Canyon may reflect a combination of sediments derived from a transcontinental river system and local influence from the early uplifts associated with the Ancestral Rocky Mountains (Gehrels et al., 2011).

Relationship to IVF Systems
The recognition of somewhat similar provenance for Upper Mississippian strata across Laurentia certaintly suggests the establishment of major transcontinental sediment delivery systems in the Late Mississippian. However, IVF systems across the midcontinent suggest that the sediment distributions pathways and processes are more complicated than can be represented by a simple east-west transport model. The IVF systems in the Hugoton Embayment and Illinois Basin are similar, with N-S-oriented incised valleys and sedimentary fill that suggests north-to-south (or inland to offshore) transport of sediments. The NW-SE constructive delta of Wedington sandstone on the Arkoma Shelf (Price, 1981;Winkelmann, 2007;Xie, O'Connor, et al., 2016) may have been fed by a similar drainage. This pattern indicates that sediment delivery, particularly to the shelf, was not completed through a single step. The sediments were first delivered to the north, perhaps through a major east to west flowing river with its provenance in the orogen, and subsequently delivered to the south by other river systems and processes.
For the Illinois Basin, Smith and Read (2000) proposed that incised valleys were formed by changes in the magnitude of eustatic fluctuations associated with abrupt increases in ice volume, tied to the expansion of continental glaciers. For the Hugoton Embayment, the IVF systems are also believed to be a consequence of transgressive-regressive cycles associated with sea level fluctuations (Ross & Ross, 1988). Fluvial systems across the region incised deep valleys during low stands in sea level and transported sediments from the north, southward, to the continental shelf. Sediments preserved within the incised valleys are interpreted to have been deposited and/or reworked during the transgressive part of the cycle. These sea level changes may be ultimately linked to changes in atmospheric and oceanic circulation associated with the closure of the equatorial seaway between Laurentia and Gondwana and the uplift of the Alleghanian Plateau (Smith & Read, 2000). Therefore, sediment dispersal in North America is ultimately tied to the Laurentia- Figure 10. Mississippian paleogeography (from North American Key Time Slices ©2013 Colorado Plateau Geosystems Inc.) and schematic diagram of one possible configuration for sediment dispersal in Late Mississippian. The exact route is difficult to ascertain due to Pennsylvanian through Permian erosion, which affected much of the region. The extent to which the Transcontinental Arch and other structural highs acted as barriers to westward transport of sediment is also uncertain. Note that ocean currents and wind currents (not shown) would have also played an important role in sediment dispersal from the orogen. Abbreviations: CU=Cincinnati Uplift; NR = Nemaha Ridge; OD = Ozark Dome.
Gondwana collision and was established into two steps: (1) east to west transcontinental sediment delivery was driven by direct collision and uplift along the eastern margin of the continent and (2) north to south (or inland to offshore) sediment delivery through incised valleys was controlled by associated eustatic changes. This tectonic and depositional framework established in the Late Mississippian was inherited by Pennsylvanian and Permian depositional systems but under a more intense tectonic regime and likely with more complicated sediment dispersal systems.

Conclusion
In this study, we report 1,037 new concordant detrital zircon U-Pb ages from Uppper Mississippian sandstones collected from core in southwestern Kansas and from outcrops in northwestern Arkansas. Based on the new data and comparisons with detrital zircon data from Paleozoic sandstones and time-equivalent units across the US, we draw several key conclusions: 1. Detrital zircon U-Pb ages from samples of Chesterian IVF sandstones from the Hugoton Embayment are similar and include two pronounced age clusters of Middle to Late Paleoproterozoic (1,300-900 Ma) and Paleozoic (500-350 Ma), which correspond to Grenville and Acadian-Taconic orogenies, and suggest major derivation from an Appalachian source region. Other age groups like the Yavapai-Mazatzal grains (1,800-1,600 Ma) and high number of Grenville grains indicate subordinate contributions from the local basement uplifts and the Midcontinent rift system. 2. Late Mississippian sandstones preserved in and/or along the Appalachian Foreland Basin, Black Warrior Basin, Illinois Basin, Ozark Dome, Arkoma Shelf, Hugoton Embayment, and Grand Canyon were likely derived from a fairly uniform E-W transcontinental river system; however, complications exist within the system due to local structural highs. 3. Sediment dispersal in North America can be interpreted as a response to the Laurentia-Gondwana collision and was established into two steps. East to west transcontinental sediment delivery was driven by collision and uplift along the eastern margin of the continent, whereas north to south (or inland to offshore) sediment delivery through incised valleys (Hugoton Embayment and Illinois Basin), one of the dispersal patterns, was controlled by associated glacioeustacic sea level fluctuations.

Data Availability Statement
The data used are listed in the references, tables, and supporting information. The full suite of data is also available through the IEDA (Interdisciplinary Earth Data Alliance) EarthChem Library (www.earthchem. org) at https://doi.org/10.1594/IEDA/111376.