Detrital Zircon Provenance and Lithofacies Associations of Montmorillonitic Sands in the Maastrichtian Ripley Formation: Implications for Mississippi Embayment Paleodrainage Patterns and Paleogeography

: We provide new detrital zircon evidence to support a Maastrichtian age for the establishment of the present-day Mississippi River drainage system. Fieldwork conducted in Pontotoc County, Mississippi, targeted two sites containing montmorillonitic sand in the Maastrichtian Ripley Formation. U-Pb detrital zircon (DZ) ages from these sands (n = 649) ranged from Mesoarchean (~2870 Ma) to Pennsylvanian (~305 Ma) and contained ~91% Appalachian-derived grains, including Appalachian–Ouachita, Gondwanan Terranes, and Grenville source terranes. Other minor source regions include the Mid-Continent Granite–Rhyolite Province, Yavapai–Mazatzal, Trans-Hudson / Penokean, and Superior. This indicates that sediment sourced from the Appalachian Foreland Basin (with very minor input from a northern or northwestern source) was being routed through the Mississippi Embayment (MSE) in the Maastrichtian. We recognize six lithofacies in the ﬁeld areas interpreted as barrier island to shelf environments. Statistically signiﬁcant di ﬀ erences between DZ populations and clay mineralogy from both sites indicate that two distinct ﬂuvial systems emptied into a shared back-barrier setting, which experienced volcanic ash input. The stratigraphic positions of the montmorillonitic sands suggest that these deposits represent some of the youngest Late Cretaceous volcanism in the MSE. Gondwanan Terranes Taconic (490–440 Acadian (450–320 and Alleghanian (330–265 Ma). A detailed discussion of the North American igneous terranes with related citations can be found in Appendix A.


Introduction
Time constraints on the initiation of the present-day Mississippi River drainage system have benefitted from recent detrital zircon (DZ) studies that establish its emplacement by the Maastrichtian [1][2][3]. This has important implications for the paleogeography and tectonic history of North America, particularly in the Mississippi Embayment (MSE), where seismicity continues to the present. Here, we provide new data from the Ripley Formation in northern Mississippi to support Gulf-of-Mexico-directed flow from eastern North America during the Maastrichtian.

Mississippi Embayment Sedimentation
The MSE is a ~259,000 km 2 south-plunging syncline that extends from southern Illinois (IL) into parts of Alabama (AL), Arkansas (AR), Kentucky (KY), Louisiana (LA), Mississippi (MS), Missouri (MO), Tennessee (TN), and Texas (TX) ( Figure 1A,C). The MSE is underlain by the Reelfoot Rift zone, a Neoproterozoic to lower Cambrian aulacogen that formed during the breakup of Rodinia. Despite its age, the Reelfoot Rift zone is currently the most seismically active structural province in the eastern United States [8][9][10] and it is linked to the New Madrid Seismic Zone, in the northwestern and central MSE. The New Madrid Seismic Zone is characterized by Late Proterozoic to Early Paleozoic seismogenic faulting [8,10,11] that is still active, as indicated by historical epicenters oriented linearly along the southeastern margin of the Reelfoot Rift [12]. Our initial approach to the project was to determine the age of the ash that weathered to montmorillonite, with a secondary goal of determining DZ provenance to improve interpretations of paleodrainage patterns. We also undertook a lithofacies analysis to interpret the depositional environments of the montmorillonitic sand.

Mississippi Embayment Sedimentation
The MSE is a~259,000 km 2 south-plunging syncline that extends from southern Illinois (IL) into parts of Alabama (AL), Arkansas (AR), Kentucky (KY), Louisiana (LA), Mississippi (MS), Missouri (MO), Tennessee (TN), and Texas (TX) ( Figure 1A,C). The MSE is underlain by the Reelfoot Rift zone, a Neoproterozoic to lower Cambrian aulacogen that formed during the breakup of Rodinia. Despite its age, the Reelfoot Rift zone is currently the most seismically active structural province in the eastern United States [8][9][10] and it is linked to the New Madrid Seismic Zone, in the northwestern and central MSE. The New Madrid Seismic Zone is characterized by Late Proterozoic to Early Paleozoic Modified from Swann and Dew [23], Walker et al. [24], and Dockery [25].
The previously mined bentonites were laterally extensive and pure enough to be considered an economic resource in the 1940s [21,22]. The bentonite outcrops described by Priddy [21] were based on a coarse sand located either within or above the CSM, which graded into a less silty, less calcareous clay that then graded into a dark blue/gray, silty, calcareous clay. The bentonite reached a maximum thickness of ~2.7 m in northern Pontotoc County and thinned to ~0.6 m in the southeast [22]. The lack of exposed bentonites within the entire upper Ripley was interpreted as the result of erosion before deposition of the uppermost upper Ripley member [21].

Sample Collection and Preparation
Samples were collected at the two locations where expansive clay was observed in Pontotoc County, MS ( Figure 1B). The northern site is in northeast Pontotoc County, off Tuscumbia Road (34.34917°N, 88.90722°W) ( Figure 1B). The southern site is in a sand quarry southeast of the city of Pontotoc ( Figure 1B), near state highway 41 (34.20417°N, 88.97778°W). The outcrop is ~9 m below the K-Pg boundary, determined by pacing and correcting for slope. The two sites are ~17.4 km apart from each other as the crow flies.
Samples from the southern site were collected from two trenches that were excavated ~15 m apart, one to the northeast (NE) and one to the southeast (SE) in a weathered high wall from previous sand mining within the upper Ripley ( Figure 1B). Detailed stratigraphic sections were measured and lithofacies were characterized at each location. Bulk samples at the southern location were collected in ~1 m intervals with additional samples collected from the expansive clay-rich interval. Northern samples were collected from sections exposed by previous mining activity ( Figure 1B).
Samples were oven dried and pulverized in the University of Mississippi Geochronology laboratory using a disk mill. The southern site zircons were initially separated using a centrifugal bowl that separated sediment based on grain density and slightly on grain size as finer particles were washed out of the sample in addition to lighter density particles. The northern site samples were separated using a Jasper Canyon Research (JCR) zircon-concentrating table. After washing, the heavy fractions were gathered and dried before undergoing heavy liquid separation. Lithium Modified from Swann and Dew [23], Walker et al. [24], and Dockery [25].
The previously mined bentonites were laterally extensive and pure enough to be considered an economic resource in the 1940s [21,22]. The bentonite outcrops described by Priddy [21] were based on a coarse sand located either within or above the CSM, which graded into a less silty, less calcareous clay that then graded into a dark blue/gray, silty, calcareous clay. The bentonite reached a maximum thickness of~2.7 m in northern Pontotoc County and thinned to~0.6 m in the southeast [22]. The lack of exposed bentonites within the entire upper Ripley was interpreted as the result of erosion before deposition of the uppermost upper Ripley member [21].

Sample Collection and Preparation
Samples were collected at the two locations where expansive clay was observed in Pontotoc County, MS ( Figure 1B). The northern site is in northeast Pontotoc County, off Tuscumbia Road (34.34917 • N, 88.90722 • W) ( Figure 1B). The southern site is in a sand quarry southeast of the city of Pontotoc ( Figure 1B), near state highway 41 (34.20417 • N, 88.97778 • W). The outcrop is~9 m below the K-Pg boundary, determined by pacing and correcting for slope. The two sites are~17.4 km apart from each other as the crow flies.
Samples from the southern site were collected from two trenches that were excavated~15 m apart, one to the northeast (NE) and one to the southeast (SE) in a weathered high wall from previous sand mining within the upper Ripley ( Figure 1B). Detailed stratigraphic sections were measured and lithofacies were characterized at each location. Bulk samples at the southern location were collected in~1 m intervals with additional samples collected from the expansive clay-rich interval. Northern samples were collected from sections exposed by previous mining activity ( Figure 1B).
Samples were oven dried and pulverized in the University of Mississippi Geochronology laboratory using a disk mill. The southern site zircons were initially separated using a centrifugal bowl that separated sediment based on grain density and slightly on grain size as finer particles were washed out of the sample in addition to lighter density particles. The northern site samples were separated using a Jasper Canyon Research (JCR) zircon-concentrating table. After washing, the heavy fractions were gathered and dried before undergoing heavy liquid separation. Lithium Polytungstates (LST; specific gravity (SG) =~2.85) were used to separate lighter minerals from the heavy fraction based on density and were then run through a magnetic separator (Franz LB-1). The non-magnetic fraction was separated with methylene iodide (MI) (SG =~3.30) to yield zircon grains.
Zircon grains were handpicked under a microscope and > 120 grains per sample were mounted on a glass plate using adhesive tape. The selected zircon grains were embedded in a 2.54 cm diameter epoxy plug and polished for imaging approximately halfway through the zircon grains.

Scanning Electron Microscope Imaging
Zircons were imaged at the Mississippi State University Institute for Imaging and Analytical Techniques (I 2 AT) using a Zeiss EVO 5.0 high-vacuum Scanning Electron Microscope (SEM) and a Gatan miniCL cathodoluminescence attachment. Secondary Electron (SE) and cathodoluminescence (CL) images were collected and used to characterize grain zoning and classification (igneous, metamorphic, or detrital) ( Figure 3). Targets for U-Pb laser ablation (~20 µm) were selected on the CL images for each grain based on zonation. Some grains had multiple U-Pb sites based on the presence of a core and rim in the same grain that could possibly yield different ages.
Geosciences 2020, 10, x FOR PEER 5 of 34 Polytungstates (LST; specific gravity (SG) = ~2.85) were used to separate lighter minerals from the heavy fraction based on density and were then run through a magnetic separator (Franz LB-1). The non-magnetic fraction was separated with methylene iodide (MI) (SG = ~3.30) to yield zircon grains. Zircon grains were handpicked under a microscope and > 120 grains per sample were mounted on a glass plate using adhesive tape. The selected zircon grains were embedded in a 2.54 cm diameter epoxy plug and polished for imaging approximately halfway through the zircon grains.

Scanning Electron Microscope Imaging
Zircons were imaged at the Mississippi State University Institute for Imaging and Analytical Techniques (I 2 AT) using a Zeiss EVO 5.0 high-vacuum Scanning Electron Microscope (SEM) and a Gatan miniCL cathodoluminescence attachment. Secondary Electron (SE) and cathodoluminescence (CL) images were collected and used to characterize grain zoning and classification (igneous, metamorphic, or detrital) ( Figure 3). Targets for U-Pb laser ablation (~20 µ m) were selected on the CL images for each grain based on zonation. Some grains had multiple U-Pb sites based on the presence of a core and rim in the same grain that could possibly yield different ages.

Figure 3.
A range of detrital zircon morphologies and the zonation CL images grains from all 11 samples. The ~20 μm laser ablation spots are indicated by circles on the individual grains.

Uranium-Lead (U-Pb) Zircon Isotopic Analysis
U-Pb analyses for the initial samples from the southern site (SE140 and NE100) were performed on a Nu Plasma 3-D Multi-collector inductively coupled plasma-mass spectrometer (Nu-ICP-MS) at the University of Florida Department of Geosciences, Geochronology Laboratory. Zircon ablation occurs in a stream of He and is then mixed into an Ar gas flow that carries minute particles of zircon into the plasma stream, which ionizes them, allowing U and Pb isotopes to be measured simultaneously using the multi-collector. Ion collectors on the end of the Nu-ICP-MS simultaneously collect 204 Pb ( 204 Hg), 206 Pb, and 207 Pb ions and Faraday detectors collect 235 U and 238 U signals. A 20-second integration that clears the gas blank and the 204 Hg input precedes a 30-second period that clears the previous analysis. After the completion of Nu-ICP-MS analysis, isotopic ratios of 206 Pb/ 238 U, 207 Pb/ 235 U, and 207 Pb/ 206 Pb were provided from the Nu-Instruments Time Resolved Analysis software. To correct the fractionation and drift errors that occurred, corrections were calibrated against a known standard (FC-1 natural zircon standard; Duluth Gabbro; 1098 Ma; [26,27]). Analyses were arranged with 2 ablations of FC-1, followed by 10 unknowns, and then 2 more FC-1. Collected data were input into the CALAMARI 9.0 (© P. Mueller) reduction protocol for U-Pb geochronology spreadsheet.
GeoSep Services (GSS) in Idaho analyzed the sample TR-02 from the northern site. Isotopic analyses used a New Wave UP-213 laser ablation system along with an Agilent 7700x quadrupole inductively coupled plasma-mass spectrometer (LA-ICP-MS) at the Washington State University GeoAnalytical Lab. Material was delivered to the plasma source by a stream of He and Ar gas. Each

Uranium-Lead (U-Pb) Zircon Isotopic Analysis
U-Pb analyses for the initial samples from the southern site (SE140 and NE100) were performed on a Nu Plasma 3-D Multi-collector inductively coupled plasma-mass spectrometer (Nu-ICP-MS) at the University of Florida Department of Geosciences, Geochronology Laboratory. Zircon ablation occurs in a stream of He and is then mixed into an Ar gas flow that carries minute particles of zircon into the plasma stream, which ionizes them, allowing U and Pb isotopes to be measured simultaneously using the multi-collector. Ion collectors on the end of the Nu-ICP-MS simultaneously collect 204 Pb ( 204 Hg), 206 Pb, and 207 Pb ions and Faraday detectors collect 235 U and 238 U signals. A 20-s integration that clears the gas blank and the 204 Hg input precedes a 30-s period that clears the previous analysis. After the completion of Nu-ICP-MS analysis, isotopic ratios of 206 Pb/ 238 U, 207 Pb/ 235 U, and 207 Pb/ 206 Pb were provided from the Nu-Instruments Time Resolved Analysis software. To correct the fractionation and drift errors that occurred, corrections were calibrated against a known standard (FC-1 natural zircon standard; Duluth Gabbro; 1098 Ma; [26,27]). Analyses were arranged with 2 ablations of FC-1, followed by 10 unknowns, and then 2 more FC-1. Collected data were input into the CALAMARI 9.0 (© P. Mueller) reduction protocol for U-Pb geochronology spreadsheet.
GeoSep Services (GSS) in Idaho analyzed the sample TR-02 from the northern site. Isotopic analyses used a New Wave UP-213 laser ablation system along with an Agilent 7700x quadrupole inductively coupled plasma-mass spectrometer (LA-ICP-MS) at the Washington State University GeoAnalytical Lab. Material was delivered to the plasma source by a stream of He and Ar gas. Each analysis cycle took~30 s and consisted of a 6-s integration with the laser shutter closed to collect background measurements followed by a 24-s integration with the laser shutter open. Analyses were separated by a 20-s delay and measured the following isotopes: 202 Hg, 204 (Hg + Pb), 206 Pb, 207 Pb, 208 Pb, 232 Th, 235 U, and 238 U. U-Pb age standards were used during analysis for calibration purposes. Laser ablations consisted of 2 ablations of the FC-1 standard, 10 ablations of the zircon sample, and 2 more FC-1 ablations [26]. FC-1 was used in order to correct the fractionation and drift errors that occurred, corrections were then calibrated against a known standard [26,27]. Collected analytical data were input into the GSS-ZrnUPb (© Geoseps Services, LLC) reduction protocol for U-Pb geochronology spreadsheet.
All remaining samples were analyzed by laser ablation ICP-MS using a Varian 810 Quadrupole ICP-MS coupled with a PhotonMachines Analyte.193 excimer laser at the University of Arkansas Trace Element and Radiogenic Isotope Lab. Data reduction methods are outlined in Shaulis et al. [28]. Laser ablations consisted of 2 ablations of the FC-1 and Plešovic standard, 15 ablations of the zircon sample, and 2 more FC-1 and Plešovic ablations [26,29]. Concentrations of U and Th are calibrated relative to the NIST612 standard zircon, which contains~765 ppm of U and 80 ppm Th.
U-Pb analyses were sorted based on discordance, and those between < 10% or > −5% reversely discordant were approved for inclusion in the final dataset. These discordance thresholds are arbitrary, but are a widely accepted range to remove damaged grains from the data set (e.g., major Pb loss or large amounts of common Pb). After sorting based on discordance, analyses were chosen for each sample. For ages < 800 Ma, 206 Pb/ 238 U ages are used and for grain ages > 800 Ma, 207 Pb/ 206 Pb ages are used. These ages were used to create probability density plots (PDPs) in Isoplot 3.75 [30] which show the number and relative probability of the grain ages along with the potential source locations. Cumulative age-probability plots (CDPs) were created using the analytical spreadsheet created by the Arizona Laserchron Center [31] and are a visual tool used to show the cumulative age data by graphically comparing the cumulative probability for each sample on a single graph.

Statistical Analysis
Statistical analyses were performed in order to compare samples using Kolmogorov-Smirnov (K-S) tests [32] and Multidimensional Scaling (MDS). K-S tests were performed using the analytical spreadsheet from the Arizona Laserchron Center [32]. The K-S test compares and differentiates the DZ age spectra and tests the null hypothesis that the samples are from the same population based on a > 95% confidence level, represented by the p-value [32]. These tests are based on the cumulative density functions (CDFs) that create the CDPs [31]. The K-S test compares the maximum vertical probability distances between the curves and a value based on the number of samples in the distribution and confidence level, which is referred to as the critical value [32]. If the distance between two sample CDFs is greater than the critical value, then the null hypothesis is rejected, and DZ samples did not originate from the same source (p < 0.05). Therefore, the K-S test only indicates if populations are statistically different rather than whether the populations are the same [33], and if the p-value is greater than 0.05, then the two DZ age spectra may share the same provenance.
The K-S test is not without its criticisms. p-values may be strongly dependent on sample size, which can sometimes limit their usefulness for detrital geochronology [34]. The K-S statistic can be considered less sensitive at the extreme ends of the distributions [35], and it can be biased by near-unimodal distributions. However, Vermeesch [36] argued that it is more reliable in distinguishing provenance similarities than other dissimilarity statistics. Lower K-S values suggest relatively low distribution dissimilarities and may be used as support for common source regions. This statistical technique provides quantitative support for visual curve matching.
To overcome the potential limitations of our K-S tests, we supplemented our analyses with non-metric MDS using K-S dissimilarity in the package provenance for R [37]. MDS generates a matrix of D statistics-the maximum differences between successive pairs of cumulative density functions-calculated as part of the K-S test. The matrix is used to graphically represent relative dissimilarities in Euclidean space, where each DZ population plots as a point and the distance between points represents their dissimilarity values, i.e., samples with high similarities plot near each other, and samples that are highly dissimilar plot further apart [34]. MDS has been used relatively recently to compare DZ datasets (e.g., [36,[38][39][40][41][42]) and it has proven to be particularly useful when visualizing the dissimilarity of a large set of samples [34,38,39]. Following the recommendation of Vermeesch [36], we include solid and dotted lines connecting points on our MDS plot to represent the least dissimilar and second least dissimilar samples, respectively [34,36].
K-S tests were performed to compare all of the samples at the southern location to each other as well as all of the samples at the northern location to each other (MDS was not used in these cases because of the small sample sizes). The cumulative southern and cumulative northern data sets were then compared using a K-S test and MDS. Finally, a K-S test and MDS were used to compare the Pontotoc samples to strata with hypothesized shared provenance: Cenomanian strata and modern sediments from the GoM [1], Cretaceous sands in the Forest City Basin (FCB), Iowa [17], the Mesoproterozoic Mid-Continent Rift [43], the Paleozoic Mid-Continent [44], strata in the Illinois Basin [3,45], Paleozoic strata in the Black Warrior Basin [46,47], and Appalachian Foreland Basin strata [48][49][50]. The K-S results are provided in the Supplementary Material.

X-ray Diffraction (XRD) Analysis
To identify expansive clay mineralogy, two samples were collected from each apparent bentonitic interval from the northern and southern Pontotoc sites for XRD analysis. All four samples were powdered in house at the University of Mississippi and then sent to Mississippi State for analysis. XRD analysis was performed at the I 2 AT using a Rigaku Ultima III X-ray Diffraction System. Please note that bentonites are defined as clay deposits dominated by smectite-group minerals, commonly montmorillonite, formed from alteration of volcanic ash [51]. For our purposes, the presence of any smectite-group minerals would be significant as indicators of Maastrichtian volcanism in the region, regardless of whether the targeted units can be classified as true bentonites.

Lithofacies
We recognize six lithofacies in the field areas. Because of the friable and erosive nature of the material, exposures are limited and lateral lithofacies variations and relationships cannot always be evaluated. The stratigraphic relationships between lithofacies are represented in Figure 4.

Sandy Clay Lithofacies
Very thickly bedded (180-385 cm), massive, poorly sorted, sandy gray to red clay with matrix-supported, very fine to coarse, but dominantly medium-sand-sized, subangular to subrounded quartz grains with rare very-fine-sand-sized muscovite grains and local concentrations of clay, some as mm-scale diameter spherical aggregates that spall out of freshly exposed surfaces (Figures 4 and 5A). Beds fine upward to silty clay and contain somewhat laterally extensive (traceable for tens of meters) lenses of less sandy gray clay with irregular boundaries, as if disrupted by bioturbation (Figures 4 and 5B,C). This lithofacies weathers to sloped surfaces with polygonal desiccation cracked patterns when dry (Figures 4 and 5D), which led us to hypothesize that this lithofacies contained expansive clays. The upper ~145 cm of the beds show angular blocky and

Sandy Clay Lithofacies
Very thickly bedded (180-385 cm), massive, poorly sorted, sandy gray to red clay with matrix-supported, very fine to coarse, but dominantly medium-sand-sized, subangular to subrounded quartz grains with rare very-fine-sand-sized muscovite grains and local concentrations of clay, some as mm-scale diameter spherical aggregates that spall out of freshly exposed surfaces (Figures 4 and 5A). Beds fine upward to silty clay and contain somewhat laterally extensive (traceable for tens of meters) lenses of less sandy gray clay with irregular boundaries, as if disrupted by bioturbation (Figures 4 and 5B,C). This lithofacies weathers to sloped surfaces with polygonal desiccation cracked patterns when dry (Figures 4 and 5D), which led us to hypothesize that this lithofacies contained expansive clays. The upper~145 cm of the beds show angular blocky and wedge ped structure, black ped coatings, and abundant grass roots, representing recent soil development at the modern surface. Spherical clay aggregates are found below the zone of soil development.

Carbonaceous Sand Lithofacies
The complete thickness of this lithofacies could not be observed because it is present at the base of the measured section and is not fully exposed. Beds are at least 60 cm thick and consist of mostly massive, poorly sorted, fine-to-medium, subrounded to subangular quartz sand grains within a dark gray clay matrix (Figures 4 and 6A) with sporadic, thin, discontinuous beds of organic-rich sand and sporadic, irregularly shaped lenses of white sand with orange rims, possibly representing bioturbation. Also present are undulatory, thick lamina of resistant iron cemented sandstone. There is an upward decrease in clay content and the uppermost bounding surface is sharp, undulatory, and marked by a layer of resistant Fe-cemented sandstone (Figures 4 and 6B).

Carbonaceous Sand Lithofacies
The complete thickness of this lithofacies could not be observed because it is present at the base of the measured section and is not fully exposed. Beds are at least 60 cm thick and consist of mostly massive, poorly sorted, fine-to-medium, subrounded to subangular quartz sand grains within a dark gray clay matrix (Figures 4 and 6A) with sporadic, thin, discontinuous beds of organic-rich sand and sporadic, irregularly shaped lenses of white sand with orange rims, possibly representing bioturbation. Also present are undulatory, thick lamina of resistant iron cemented sandstone. There is an upward decrease in clay content and the uppermost bounding surface is sharp, undulatory, and marked by a layer of resistant Fe-cemented sandstone (Figures 4 and 6B).

Wavy-Bedded Sand Lithofacies
Very thickly bedded (> 200 cm), poorly sorted, orange-brown, silty fine sand that is dominantly subangular to subrounded quartz grains. Wavy bedding is the dominant sedimentary structure (Figures 4 and 6C), with rare lenticular bedding. Wavy bedding consists of poorly sorted (mud-rich), slightly darker sand interbedded with better-sorted, slightly lighter colored sand. No body fossils or trace fossils were observed. Lower boundaries of this lithofacies were not observable in the field, but upper contacts were gradational into the massive sand lithofacies.

Massive Sand Lithofacies
Very thickly bedded (220 to 1660 cm), massive, tan to orange, friable, poorly sorted, fine-to-medium, dominantly quartz, subangular to subrounded sand. Gray to white clay is present as matrix and as local resistant lenses, laminae, thin beds, and rare rip-up clasts. Iron-oxide-cemented thin beds, thick laminae, and concretions with concentric fabric are present towards the bases of beds, particularly at the sharp, undulatory basal contact with the carbonaceous sand lithofacies (Figures 4 and 6D). Basal contacts are gradational where underlain by the wavy-bedded sand lithofacies. Upper contacts are sharp where overlain by the sandy packstone lithofacies and gradational where overlain by the sandy clay lithofacies. Neither organic matter nor body fossils were observed.
Two types of trace fossils are abundant in localized horizons. Type 1 traces (Figures 4 and 6E) are exposed only in two dimensions on trench walls, but consist of horizontal to subhorizontal and vertical shafts that intersect. Shafts are up to~3 cm wide and are walled by 0.5 to 1 cm thick yellow mud and sand that appear smooth to pellet-like and resistant. Shaft fill is brown sand that is darker and coarser than the surrounding matrix. Shaft diameters vary along their lengths and some appear to contain rounded enlargements. Type 2 traces (Figures 4 and 6F) are horizontal, subhorizontal, and subvertical, resistant, cylindrical tubes of consistent diameter that range from 1 to 3 cm in diameter. Only short segments of the tubes are exposed so it is unknown whether they intersect. Tubes are walled with < 0.5 cm thick walls, some of which have iron oxide cement. Fill is concentric laminae of mud and fine sand, some white and some the same color as the matrix. The two types of trace fossils are not found together in the same horizons.

Ripple-Laminated Sand Lithofacies
Thick bedded (up to 50 cm), friable, ripple-laminated, light brown, poorly sorted, fine-to-medium, subrounded to subangular quartz sand (Figures 4 and 6G). Ripple lamination is subtle and difficult to see without proper lighting. Ripples are asymmetrical, but postdepositional slumping and microfaulting make it difficult to determine current direction. This lithofacies grades upward into the sandy clay lithofacies and grades laterally into the wavy-bedded sand lithofacies. No fossils were observed.

Zircon Geochronology
DZ provenance studies are possible because DZ ages correspond to crystallization ages of the individual cratonic elements that amalgamated over billions of years to form the continents in their current configurations [52].  A total of 649 U-Pb analyses were collected that fit the criteria across 11 samples (Figure 8). Age analysis shows an overall age range of 2870 Ma to 305 Ma. All U-Pb analyses are available in the Supplementary Materials (Table S1). The potential source age range percentages for the southern and northern samples are summarized in Table 1.  A total of 649 U-Pb analyses were collected that fit the criteria across 11 samples (Figure 8). Age analysis shows an overall age range of 2870 Ma to 305 Ma. All U-Pb analyses are available in the Supplementary Materials (Table S1). The potential source age range percentages for the southern and northern samples are summarized in Table 1.

U-Pb Geochronology of Southern Samples
The seven southern samples (n = 376) have detrital age spectra of mostly Grenville-aged grains (80.1%, n = 301) (Figure 9). Acadian-aged grains make up the next dominant group with 10.6% of the grains (n = 40) coming from the 450-320 Ma age range. The remainder of the age groups, in descending order, are: MCGR (4.5%, n = 17), Gondwanan Terranes (2.9%, n = 11), Alleghanian (1.1%, n = 4), Superior (0.5%, n = 2), and Taconic (0.3%, n = 1) (Figure 9). The southern samples did not yield any results from the Yavapai-Mazatzal or Trans-Hudson/Penokean provinces (Table 1; Figure 9). Although detailed sample descriptions of sample separates were not recorded, heavy mineral separation yielded high amounts of kyanite within the southern samples. Geosciences 2020, 10, x FOR PEER 14 of 34 Sample NE-S. This sample was collected from the wavy-bedded sand lithofacies (Figure 4). U-Pb zircon analyses (n = 25) for this sample range in age from 1372 Ma to 602 Ma. The PDP for sample NE-S contains a dominant double peak at ~1172 Ma and ~1081 Ma in the Grenville age range along with two other age peaks at ~602 Ma (Gondwanan Terranes) and ~1386 Ma (MCGRP) (Figure 8).
Sample NE0. This sample was collected at the base of the sandy clay lithofacies in the northeastern trench (Figure 4). U-Pb zircon analyses (n = 21) for this sample range in age from 1400 Ma to 440 Ma. The PDP for sample NE0 has a dominant double peak at ~1167 Ma and ~1036 Ma in the Grenville age range along with two other age peaks at ~444 Ma (Taconic) and ~698 Ma (Gondwanan Terranes) (Figure 8). Sample NE-S. This sample was collected from the wavy-bedded sand lithofacies (Figure 4). U-Pb zircon analyses (n = 25) for this sample range in age from 1372 Ma to 602 Ma. The PDP for sample NE-S contains a dominant double peak at~1172 Ma and~1081 Ma in the Grenville age range along with two other age peaks at~602 Ma (Gondwanan Terranes) and~1386 Ma (MCGRP) (Figure 8).
Sample NE0. This sample was collected at the base of the sandy clay lithofacies in the northeastern trench ( Figure 4). U-Pb zircon analyses (n = 21) for this sample range in age from 1400 Ma to 440 Ma. The PDP for sample NE0 has a dominant double peak at~1167 Ma and~1036 Ma in the Grenville age range along with two other age peaks at~444 Ma (Taconic) and~698 Ma (Gondwanan Terranes) (Figure 8).
Sample NE100. This sample was collected from the middle of the sandy clay lithofacies within the northeastern trench ( Figure 4). U-Pb zircon analyses (n = 76) for this sample range in age from 1447 Ma to 326 Ma. The PDP for sample NE100 has a dominant double peak at~1176 Ma and~1033 Ma in the Grenville age range along with two other age peaks at~337 Ma (Alleghanian) and~428 Ma (Taconic/Acadian). Minor age occurrences also appear in the Gondwanan Terranes and MCGRP ranges (Figure 8).
Sample NE180. This sample was collected from the sandy clay lithofacies near the top of the northeastern trench (Figure 4). U-Pb zircon analyses (n = 70) for this sample range in age from 1410 Ma to 332 Ma. The PDP for sample NE180 has a dominant double peak at~1163 Ma and~1041 Ma in the Grenville age range along with two other age peaks at~357 Ma (Acadian) and~432 Ma (Taconic/Acadian). Minor age occurrences are present in the Gondwanan Terranes and MCGRP ranges (Figure 8).
Sample SE60. This sample was collected from the lower third of the sandy clay lithofacies (Figure 4). U-Pb zircon analyses (n = 57) for this sample range in age from 2667 Ma to 305 Ma. The PDP for sample SE60 has a dominant double peak at~1110 Ma and~1222 Ma in the Grenville age range along with three other age peaks at~337 Ma (Alleghanian) and~428 Ma (Taconic/Acadian), and~599 Ma (Gondwanan Terranes). Sample SE60 does show a minor age occurrence in the Superior age range (> 2500 Ma) ( Figure 8).
Sample SE140. This sample was collected from the sandy clay lithofacies in the middle portion of the southeastern trench ( Figure 4). U-Pb zircon analyses (n = 83) for this sample range from 1489 Ma to 333 Ma. The PDP for sample SE140 has a dominant double peak at~1040 Ma and~1182 Ma in the Grenville age range along with three other age peaks at~342 Ma (Alleghanian/Acadian) and~418 Ma (Acadian), and~1485 Ma (MCGRP) (Figure 8).
Sample SE200. This sample was collected from the top of the sandy clay lithofacies close to the base of modern soil overprinting (Figure 4). U-Pb zircon analyses (n = 44) for this sample range from 1337 Ma to 349 Ma. The PDP for sample SE200 has a dominant double peak at~1040 Ma and 1182 Ma in the Grenville age range along with three other age peaks at~350 Ma (Acadian) and 435 Ma (Acadian/Taconic), and~680 Ma (Gondwanan Terranes) (Figure 8).
Sample TR-01. U-Pb zircon analyses (n = 44) for this sample range in age from 2870 Ma to 338 Ma. The PDP for sample TR-01 has a large Grenville peak at~1115 Ma along but does not exhibit the standard Grenville double peak. Four large age peaks occur at ages less than Grenville, which differs from the other samples. The four peaks are at~338 Ma (Acadian),~446 Ma (Taconic),~561 Ma, and 612 Ma, which are both Gondwanan in age. Sample TR-01 does show a minor age occurrence in the Superior age range (> 2500 Ma) and contains the oldest grain found out of all 11 samples (Figure 8).
Sample TR-02. U-Pb zircon analyses (n = 86) for this sample range in age from 2762 Ma to 354 Ma. The PDP for sample TR-02 has a dominant double peak at~1086 Ma and~1188 Ma along with two peaks at~449 Ma (Taconic) and~2754 Ma (Superior) (Figure 8).
Sample TR-03. U-Pb zircon analyses (n = 62) for this sample range in age from 2684 Ma to 439 Ma. The PDP for sample TR-03 has a dominant double peak at~1057 Ma and~1219 Ma along with two peaks at~444 Ma (Taconic) and~618 Ma (Gondwanan Terranes). Sample TR-03 does show a minor age occurrence in the Superior age range (> 2500 Ma) and contains grains in the MCGRP age range (Figure 8).
Sample TR-04. U-Pb zircon analyses (n = 81) for this sample range in age from 1769 Ma to 448 Ma. The PDP for sample TR-03 has a subtle double peak at~1105 Ma and~1180 Ma along with an age peak at~461 Ma (Taconic) and contains grains in the MCGR age range (Figure 8).
The MDS plot ( Figure 10) shows two groups: 1) All Pontotoc samples + Paleozoic AFB [48] and Central Appalachian Basin [49] [45]. The SS is nearest neighbor (least dissimilar) to the Cenomanian Tuscaloosa Formation [1] and second nearest neighbor to the Paleocene Wilcox Group of Tennessee [1]. The NS is nearest to the Mississippian to P

Statistical Results
The K-S statistical analysis of the seven samples from the southern Pontotoc site yielded p-values > 0.05 (~0.06-1.0), thus, we accept the null hypothesis that the DZ age spectra are from the same population. The K-S analysis of the four samples from northern Pontotoc likewise yielded p-values > 0.05 (~0.6-1.0) that indicated that the DZ age spectra are statistically similar. Due to the statistical similarity, the seven southern samples were combined into one southern sample (SS) and the four northern samples were combined into one northern sample (NS). These DZ spectra were then compared to each other with a K-S test, which yielded a p-value of~0.0000041 indicating that the two datasets are not statistically similar to one another. The K-S statistical analysis of the detrital age spectra for the NS and the Albian-Cenomanian Dakota Fm of Finzel [17] and the Ordovician to Devonian strata of the Appalachian Foreland Basin (AFB) of Thomas et al. [50] are statistically similar (p = 0.09 and 0.05 respectively). The SS was found to be statistically similar (p = 0.66 and 0.16 respectively) to the Cenomanian Tuscaloosa Fm and the Paleocene Wilcox in Tennessee of Blum and Pecha [1]. Both the NS and SS are statistically distinct from all other Precambrian and Paleozoic strata in the mid-continent [43,44], the other Paleozoic sandstones in the Illinois and FCB [45], the Paleozoic Black Warrior Basin strata [46,47], the Cenomanian Woodbine sandstone [1], and the Paleocene Wilcox Group in Mississippi [1].
The MDS plot ( Figure 10) shows two groups: 1) All Pontotoc samples + Paleozoic AFB [48] and Central Appalachian Basin [49] [45]. The SS is nearest neighbor (least dissimilar) to the Cenomanian Tuscaloosa Formation [1] and second nearest neighbor to the Paleocene Wilcox Group of Tennessee [1]. The NS is nearest to the Mississippian to Pennsylvanian deposits of the AFB [50] and second nearest to the southern and central AFB samples [48].

XRD Results
XRD results reveal that a large percentage of the northern and southern samples are quartz with other minor minerals ( Table 2). The northern samples (TR-01 and TR-04) have a range of 56.9%-49.8% quartz and a high kaolinite presence ranging from 21.9%-17.2% (Table 2). Low amounts of expansive clays such as montmorillonite were measured in theses samples (6.2%-3.6%) ( Table 2). The southern samples contain 36.6%-46.0% quartz and several constituents with and without expansive properties, including montmorillonite and illite, respectively. The southern samples contain a much lower amount of kaolinite than the northern samples ( Table 2). The XRD analysis images are compiled in the Supplementary Files.

Sandy Clay Lithofacies
Clay-matrix supported sand grains represent a relatively low energy environment, while the subangular to subrounded quartz sand, muscovite, and presence of likely volcanic montmorillonite indicate proximity to a terrestrial source. The spherical clay aggregates are found below the zone of soil development and therefore are not pedogenic in origin. The variety of diameters and apparent lack of organic matter argue against a fecal pellet origin. The aggregates are most similar to flocculated clay balls reported from possibly turbulent, brackish water conditions [55]. XRD and SEM analyses of these aggregates may reveal more detailed evidence about their origin, but such analyses are beyond the scope of this paper. Given the sedimentological evidence, this lithofacies may represent an estuary or back-barrier setting. We favor an interpretation as a distal back-barrier lagoon, with the fining upward profile representing decreasing energy with gradual filling (e.g., [56]).
Differences in the clay mineralogy between the northern and southern samples can provide information about relative proximity to the paleoshore. Because kaolinite flocculates more readily in low salinities and montmorillonite flocculates more rapidly at high salinities [55,57], the difference in clay mineralogy can be related to the degree of fluvial input. The northern Pontotoc samples are kaolinite rich and the southern samples have a higher montmorillonite/kaolinite ratio, suggesting that the southern location was more proximal to the open ocean or experiencing relatively minor fluvial input, while the northern location was fed by a larger fluvial system or was more isolated from the open ocean.

Carbonaceous Sand Lithofacies
The clay matrix and carbonaceous material suggest a low energy location close to a terrestrial source. The upward decrease in clay content suggests an increase in energy over time. We interpret the Fe-cemented layers as diagenetic and non-diagnostic of depositional setting. The organic carbon content of this lithofacies and its position at the base of the measured section suggests that it represents the middle Ripley, which contains abundant lignitic sand and bedded lignite [21]. We interpret the paleoenvironment as a proximal back-barrier lagoon.

Wavy-Bedded Sand Lithofacies
Despite the abundance of sand, the alternating layers of more and less mud matrix indicate alternating energy during tidally influenced deposition. The cleaner sand represents higher energy during the dominant tide and the more mud-rich sand represents lower energy during the subordinate tide. Compared to typical instances of wavy and lenticular bedding, this lithofacies represents an overall relatively high-energy environment (e.g., [58]). The lack of pure mud and lack of subaerial exposure indicators suggests that deposition occurred in a shallow subtidal setting, likely a subtidal sand flat.

Massive Sand Lithofacies
The massive nature of this lithofacies is not particularly informative in terms of depositional environment, but lateral and vertical relationships with the sandy clay, wavy-bedded, and sandy packstone lithofacies, suggest a largely subtidal marine origin. Clay bedding and laminae potentially indicate isolated low energy periods. Rip-up clasts formed when a return to high-energy conditions eroded mud beds.
Where trace fossils are present, their diversity is low, supporting the relatively high-energy conditions indicated by the sedimentology [59]. Type 1 trace fossils ( Figure 5E) likely represent two-dimensional expression of three-dimensional boxwork burrow systems. Even though the traces are poorly preserved, the size, architecture, and wall with suggestions of pellets indicate that these are attributable to Ophiomorpha and are comparable to callianassid shrimp burrows within a shoreface environment [60][61][62]. Type 2 traces ( Figure 5F) are most similar to concentrically laminated Thalassinoides described by Goldring [63]. Thalassinoides can be found in a range of marine settings, but is compatible with a shoreface environment [64].

Ripple-Laminated Sand Lithofacies
Ripple lamination indicates lower flow regime conditions in relatively shallow water. The lateral relationship to the wavy-bedded sand lithofacies suggests that the ripples represent a shoaling sand flat environment (e.g., [65]). Timing of slumping and microfaulting is difficult to determine because even recent seismicity would cause deformation along offsets due to the friable nature of the sand.

Sandy Packstone Lithofacies
The moderate energy depositional setting implied by packstone combined with an abundance of siliciclastic grains suggests an inner to middle shelf environment. The high diversity of fossils indicates normal marine conditions. The presence of Sphenodiscus, Exogyra, and Hardouinia indicate that this limestone belongs to the Chiwapa Member of the Ripley Formation [22].
Overall, the Ripley Fm in the field areas represents a barrier island to shelf system. This is similar to interpretations of the Ripley system to the east in Alabama [66].
North American zircon ages are well constrained according to the igneous province in which each grain originated (Figure 7). However, within the context of deep time, it is important to consider the possibility that sampled detrital zircons have been subjected to one or more cycles of weathering, transport, deposition, and lithification. Given that the youngest age from the Pontotoc zircons is Pennsylvanian, there is a possibility that most or all these grains have been recycled. It is also important to note that the four samples from northern Pontotoc were combined into a single population based on similar K-S results (p > 0.05) and the same approach was applied to the seven samples collected from the southern sampling locality (Figure 8). Despite the statistical similarity of the individual samples constituting the NS and the similarity of those constituting the SS, the K-S test comparing the NS and SS yielded a p-value of~0.01. Likewise, MDS shows that the NS and SS are highly dissimilar to each other when compared to other published DZ data ( Figure 10). This indicates that the NS and SS age spectra are different and possibly had slightly different sedimentary inputs. The main differences in the samples can be seen by comparing dominant age peaks in probability density plots (Figure 9). The most prominent DZ age peaks in the NS are a double peak at 1193 Ma and 1107 Ma, which is a characteristic age representing the Grenville (e.g., [52,68]). Age peaks of 447 Ma, representing Taconic ages (e.g., [69][70][71]), and 1426 Ma, representing Mid-Continent Granite-Rhyolite (e.g., [72,73]), are the second most significant peaks (Figure 9). In comparison, the two most dominant age peaks in the SS are at 1042 Ma and 337 Ma, which are Grenville and Acadian (e.g., [70,74]) in age (Figure 9). The next two most significant DZ age peaks in the SS are at 1182 Ma and 430 Ma, representing Grenville and Taconic ages, respectively (Figure 9).
Geosciences 2020, 10, x FOR PEER 20 of 34 two most dominant age peaks in the SS are at 1042 Ma and 337 Ma, which are Grenville and Acadian (e.g., [70,74]) in age (Figure 9). The next two most significant DZ age peaks in the SS are at 1182 Ma and 430 Ma, representing Grenville and Taconic ages, respectively ( Figure 9). There are four differences in the overall percentages of DZ ages between the NS and SS. The first is an 8.8% difference in Acadian (~450-320 Ma) sourced grains between the SS (1.8) and NS (10.6%; Table 1; Figure 9). Next, the NS shows a larger percentage of Taconic (~490-440 Ma) aged grains at 3.3% compared to the 0.3% in the SS (Table 1; Figure 9). Third, the MCGRP ages (~1600-1350 Ma) are more prominent in the NS (9.9%) than in the SS (4.5%; Table 1; Figure 9). Finally, the SS shows no age spectra for the Yavapai-Mazatzal (~1800-1600 Ma) and Trans-Hudson/Penokean provinces (~1900-1800 Ma) compared to the NS (1.5% and 0.7% respectively; Table 1; Figure 9).
The Grenville double peaks (Figure 9) represent the orogenic events that bound a period of Adirondian magmatism (1180-1080 Ma) [75]. The first orogenic event is the Elzeverian Orogeny (1230-1180 Ma), followed by the back-arc tectonics of the Adirondian magmatic period and finally the continental-continental collision of the Grenv      Figure 9), which are comparable to the Cenomanian (100.5-93.9 Ma) Tuscaloosa Sandstone DZ age spectra, the Paleocene Wilcox Group from Tennessee DZ age spectra obtained by Blum and Pecha [1] (Figure 11E), and the Cenomanian-Albian (~113-100.5 Ma) Dakota Fm from Finzel [17] (Figure 11D). Alleghanian-aged grains within the SS suggest a sediment source from the Appalachian-Ouachita fold-thrust belt and drainage basin that routed sediment from the Appalachians to the northeastern MSE [1] and further west during the Late Cretaceous to deposit in the FCB (Figure 13) [17]. Alleghanian deformation extends from the southern Appalachians into New England and, before the continental-scale paleodrainage shift to the MSE, the Southern and New England Appalachians could have been sourcing the Albian and Cenomanian strata in the FCB dated by Finzel [17]. Then the paleodrainage shift deposited the Cretaceous and Paleocene strata of this study and Blum and Pecha [1]. Similarities with the Thomas et al. [50] detrital spectra ( Figure 11C) indicate that the Upper Cretaceous and Paleogene strata in the MSE are likely sourced from the Ordovician and Devonian sandstones from the Taconic synorogenic clastic wedges of the southern Appalachians and show no source change during this time ( Figure 11C).
A large Acadian age peak (~337 Ma; 10.6%) occurs in the SS along with a minor Taconic peak (~430 Ma; 0.3%). This is the opposite of the NS, which has a larger Taconic age peak (~447 Ma; 3.3%) and no Alleghanian-aged grains. Acadian-aged grains are present in the Pennsylvanian Illinois Basin deposits ( Figure 12G,K; [3,45]), Albian and Cenomanian sandstone deposits in IA, AL, and MS ( Figures 12H and 11D; [1,17]), and Paleocene paleo-river deposits from the GoM coastal plain ( Figure 12H; [1]). Similar to the Taconic-aged grains, the Acadian-aged zircons may have been derived from recycling of Paleozoic strata from the AFB. It is interesting to note that the elongate kyanite grains found in the southern sample likely correlate with the detritus from the kyanite-staurolite amphibolite facies zones in the central-northern Appalachians, suggesting a more direct sediment source from the Appalachian Orogen or AFB [1,77]. Blum and Pecha [1] similarly noted a range in kyanite abundances in Cenomanian and Paleocene GoM samples.
All samples in Figures 11 and 12 show major to minor peaks in MCGRP ages (~1456-1354 Ma). The NS contains 9.9% MCGRP grains with an age peak of~1426 Ma, while the SS contains 4.5% MCGRP grains and contains two lower age peaks of~1354 Ma and~1490 Ma. The appearance of a high percentage of MCGRP grains in the Ripley Formation suggests two possible sediment sources: (1) Paleozoic intracratonic basins or (2) the exposed Precambrian rocks within the MCGRP [73]. Paleozoic basin sediments that have undergone recycling before deposition, leading to a wide age spectrum (Archean to Cenozoic) of grains present in a single sample. The only outcrops of the MCGRP from which Mesoproterozoic grains may have originated are in the St. Francois Mountains (southeast Missouri), the highest points of the Arbuckle Mountains (south-central Oklahoma), and the Spavinaw Creek area (northeastern Oklahoma) [73]. Cenozoic subsidence and Quaternary southern tilting of the MSE resulted from the underlying Reelfoot Rift and would have created a direct sediment route from surrounding areas to the MSE. However, in the Late Cretaceous, MCGRP aged grains from Oklahoma and Missouri would have been following the western drainage patterns as described by Blum and Pecha [1] and Finzel [17], making recycled sediment from the intracratonic basins the most likely source of MCGRP aged zircons ( Figure 13). different fluvial inputs into the same lagoonal system, while similarities between DZ age spectra resulted from shared source regions and sediment mixing. In this scenario, we interpret a southern fluvial system dominantly draining more easterly locations (mostly the AFB), and a northern fluvial system that drained a more westerly area, including the AFB, but with greater input from Mid-Continent, Trans-Hudson, Penokean, and Yavapai-Mazatzal provinces (Figure 13-inset). This also agrees with the lower salinity northern site representing a greater volume of fluvial input because of the correlation between drainage basin area and discharge [81].  [79]. Red lines indicate Late Cretaceous mid-continent paleodrainge and flow directions, modified from Blum and Pecha [1]. Inset: Generalized paleoenvironmental reconstruction of the region encompassing the northern and southern Pontotoc field sites (black square; see Figure 1B). Single-headed arrows represent fluvial input into back-barrier setting, thicknesses of the arrows represent relative volume of input based on salinity indicators from clay mineralogy. Dashed line represents approximate outline of subaqueous portion of deltas formed at river mouths. Double headed arrow represents mixing between the two fluvial sources.  [79]. Red lines indicate Late Cretaceous mid-continent paleodrainge and flow directions, modified from Blum and Pecha [1]. Inset: Generalized paleoenvironmental reconstruction of the region encompassing the northern and southern Pontotoc field sites (black square; see Figure 1B). Single-headed arrows represent fluvial input into back-barrier setting, thicknesses of the arrows represent relative volume of input based on salinity indicators from clay mineralogy. Dashed line represents approximate outline of subaqueous portion of deltas formed at river mouths. Double headed arrow represents mixing between the two fluvial sources.
Superior-and Wyoming-aged grains show prominent peaks in the northern Illinois Basin ( Figure 12K; [45]), FCB (Figures 12J and 11D; [17,45]), and the Paleo-Tennessee Paleocene Wilcox Group samples ( Figure 11E; [1]). A small population of Superior-and Wyoming-aged grains are observed in the Cenomanian Tuscaloosa sample of Blum and Pecha [1] (Figure 11F Figure 12G). The lack of Archean-aged grains in the Blum and Pecha [1] samples supports that Archean grains are recycled from lower Paleozoic strata [44]. Gleason et al. [80] suggested an Ordovician shift from Archean Superior Province and MCGRP ages to a dominant Appalachian-Grenville influx of sediment in western Oklahoma and central Arkansas. The Cenomanian Woodbine samples from Blum and Pecha [1] ( Figure 12I) show a large input of recycled sediment from the Ouachita fold and thrust belt and could therefore explain the Superior signature in the Ripley Formation samples as a result of sediment input from the same region [1,80].
An influx of recycled sediment from the Ouachita fold and thrust belt is also supported by the work of Finzel [17] since Albian-Cenomanian Dakota Fm samples are geographically close to the Superior region. Finzel [17] collected Albian-Cenomanian Dakota Fm samples from eastern Nebraska and Western Iowa that are statistically similar to our NS. If Cenomanian samples that are closer to the Superior and MCGRP regions do not contain strong age ranges from those provinces as well, then Superior and MCGRP age signatures are the result of sediment recycling and not direct sediment sourcing (e.g., [1,17,80]). In addition, there is a complete absence of Yavapai-Mazatzal and Trans-Hudson/Penokean ages in the SS and only a small combined population in the NS (2.2% of the sample). The overall lack of Paleoproterozoic aged grains in the NS and SS, as well as a lack of these grains in the Pennsylvanian-aged strata described in Kissock et al. [45], indicates that there was likely no direct input into the north and northeastern MSE during the Maastrichtian from the south and southwestern North American igneous provinces where these grains originally crystallized (Figure 7). The small age peak in the NS (Figure 9) is attributed to recycling of Paleoproterozoic sediment into distal sources that then drained into the MSE.
The SS is statistically indistinguishable from the Cenomanian Tuscaloosa Formation DZ ages according to K-S results (p = 0.718) and the two samples are nearest neighbors according to MDS results ( Figure 10). The SS is also statistically the same as the Paleo-Tennessee Paleocene Wilcox Group samples DZ ages [1] according to K-S results (p = 0.165) and are second nearest neighbors according to MDS results ( Figure 10). This statistical compatibility supports the Blum and Pecha [1] east-to-west drainage system extending from the Appalachian Mountains into the Western Interior Seaway (WIS) from the Permian through the Early Cretaceous. At the end of the Early Cretaceous, the retreat of the WIS rerouted all drainage from west to east and the southern river systems joined the ancestral Mississippi River and other Appalachian-sourced systems [1].
Similar to the SS, an Appalachian source for the NS is suggested by its minimum dissimilarities to the Southern and Central Appalachians and Mississippian-Permian strata of the AFB ( Figure 10). In this case, MDS results do not entirely agree with K-S results, which indicate that the NS is indistinguishable from Albian-Cenomanian strata of the Dakota Fm in western Iowa and eastern Nebraska (p = 0.083) and Ordovician-Devonian strata of the AFB (p = 0.052). This is not problematic, however, because all results support an Appalachian-Ouachita cordillera source that deposited sediment in the FCB in the Albian and Cenomanian, and shifted to MSE deposition in the Maastrichtian due to paleodrainage reconfiguration ( Figure 13) [1,17]. Since the NS and SS are similar to the detrital age spectra of Blum and Pecha [1], Finzel [17], and Thomas et al. [50], i.e., sourced from the Appalachian-Ouachita foreland basin, Ripley Fm sediments are likely sourced from the same areas (Figures 11 and 13).
The difference between NS and SS K-S and MDS results at first seems to argue against a correlation between the northern and southern sites. However, recall that many of the individual northern and southern samples are statistically indistinguishable from each other. Given the back-barrier depositional setting, we attribute differences in DZ age spectra to influence from different fluvial inputs into the same lagoonal system, while similarities between DZ age spectra resulted from shared source regions and sediment mixing. In this scenario, we interpret a southern fluvial system dominantly draining more easterly locations (mostly the AFB), and a northern fluvial system that drained a more westerly area, including the AFB, but with greater input from Mid-Continent, Trans-Hudson, Penokean, and Yavapai-Mazatzal provinces (Figure 13-inset). This also agrees with the lower salinity northern site representing a greater volume of fluvial input because of the correlation between drainage basin area and discharge [81].

Conclusions
Lithofacies analysis of the Cretaceous Ripley Formation in Pontotoc County, MS indicates a barrier island to shelf system with volcaniclastic deposits preserved in a back-barrier environment. While relatively pure bentonites were mined from below the Chiwapa Sandstone Member of the Ripley Formation in Pontotoc County in the 1940s, two newly discovered sandy expansive clay deposits are located above the CSM, but are likely not minable. These new deposits, nevertheless, are important because they record evidence of some of the youngest known Late Cretaceous volcanism in the Mississippi Embayment. While our samples did not yield volcanic zircons, we were able to perform a DZ study of a northern sample population (NS) and southern sample population (SS) to determine provenance of the Ripley Formation. DZ age spectra ranged from 2870 Ma (Mesoarchean) to 305 Ma (Late Pennsylvanian) in age with primary concentrations from the Grenville (80%), and Acadian Orogeny (7%). The NS yielded higher amounts of Taconic (3.3%) and Mid-Continent Granite-Rhyolite province (9.9%) compared to the SS (0.3% and 4.5% respectively). The SS yielded significantly higher amounts of Acadian-aged grains (10.6%) relative to the NS (1.8%). This spectrum of ages with very limited amounts of Superior, Trans-Hudson/Penokean, and Yavapai-Mazatzal leads to the conclusion that the sediment input into the northeastern MSE during the late Maastrichtian was likely sourced from the AFB with minor input from the mid-continent region by routing sediment through the Reelfoot Rift trough at the apex of the ancestral Mississippi River. AFB sourcing is based on the statistical similarities to the DZ ages of Albian-Paleocene strata of the MSE and FCB and the upper Maastrichtian strata of Pontotoc County, MS, which had not previously been analyzed in northeastern Mississippi. The similar DZ ages of this study compared to Albian-Cenomanian strata of the FCB support an east-to-west drainage system that transported sediment across the continent before the early Paleogene.
The high abundance of heavy minerals, such as muscovite and kyanite, along with the large number of Acadian-and Alleghanian-aged grains within the sample also support that sediment sources during deposition of upper Ripley strata included the AFB.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-3263/10/2/80/s1, Table S1: Analytical results (isotopic ratios) and calculated U-Pb ages for DZ from samples of this study. Samples analyzed at Washington State University (WSU) following the procedures described in Chang et al. [82] and samples analyzed at the University of Arizona (UA) following the procedures described in Gehrels et al. [83] and Gehrels et al. [84]. Reduced isotopic ratios are corrected for drift and fractionation effects. Errors reported are corrected based on analytical error and on observed variation of analyzed standards. Table S2:   Acknowledgments: This paper is the product of a Master's thesis by E.J.V. at the University of Mississippi. We thank Charles Swann and the MMRI, Chayan Lahiri, and Poe Brothers Trucking, Inc. for assistance in the field as well as access to the sample locations. We would like to thank the Gulf Coast Association of Geological Societies for funding part of this research. We also thank Barry Shaulis from the University of Arkansas Trace Element and Radiogenic Isotope Lab, George Kamenov at the University of Florida Geochronology Laboratory, and Paul O'Sullivan from GeoSep Services for their help acquiring U-Pb analyses. We also thank two anonymous reviewers for their helpful comments that greatly improved the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.