Introduction

The Mississippi River and its tributaries drain more than half the continental United States (fig. 1A). During the Pleistocene (from about 2.6 million to 12,000 years ago), the Laurentide Ice Sheet intermittently extended southward from Canada into the northern third of the Mississippi River basin. As the ice sheet retreated after each advance, glaciogenic deposits blanketed the land surface, including glacial till, glaciofluvial sand and gravel, slack water and lacustrine sediments, and eolian sand and silt. In large part, these deposits define the present landscape of the central United States and the northern part of the Mississippi Valley, including along and just south of the glacial limit (fig. 1A). In the unglaciated lower Mississippi Valley (area south of Thebes Gap on fig. 1B), glaciofluvial sediments and eolian silt are the dominant glaciogenic deposits. Full-glacial slack water or lacustrine deposits also are present, but exposures are limited (Rittenour and others, 2007, Grimley and others, 2009). Glaciofluvial sand and gravel form braided stream channel belts that are geomorphically expressed as terraces (Saucier, 1994; Rittenour and others, 2007). Eolian silt blankets many of the glaciofluvial terraces but also forms or caps high ridges within, and adjacent on the east to, the Mississippi Valley (fig. 1).

Chronostratigraphy of Lower Mississippi Valley Eolian Silt Deposits

Although silt deposits in the lower Mississippi Valley were first discussed by Lyell (1847), there was no agreement on whether these deposits were windblown or water-lain until the 1950s and 1960s (see discussion and references in Follmer, 1996; Rutledge and others, 1996; and Markewich and others, 2011). By the 1970s, a windblown origin for these silt deposits generally had been accepted as the primary mode of deposition (see discussion and references in Follmer, 1996; Rutledge and others, 1996), but individual silt deposits containing gravity- or water-modified intervals were found to be common (McCraw and Autin, 1989; Rutledge and others, 1990; Jacobs and Knox, 1994; Leigh, 1994; Markewich and others, 2011) and locally difficult to distinguish from purely eolian silt deposits (loess). Numerous field investigations in the 1980s and 1990s focused on the origin, chronostratigraphy, and climatic significance of silt deposits in the unglaciated lower Mississippi Valley. These studies traced the stratigraphic sequence of loess deposits with increasing distance from the Laurentide Ice Sheet glacial margin by looking for continuous or discontinuous deposits (fig. 1). For example, Alford and Miller (1985), Miller and Alford (1985), Miller and others (1985), and Pye and Johnson (1988) suggested that some loesses older than the Peoria Loess (fig. 2) were eroded or never deposited in the southernmost part of the Mississippi Valley from west-central Mississippi to Louisiana (fig. 1), and that other loesses were present only in the southernmost part of the valley.

Most investigations of loess chronostratigraphy in the central United States and the Mississippi Valley during the 1980s and 1990s utilized new or advancing isotopic and luminescence dating methods. Methods that were being used in the 1980s and 1990s to date loess in the lower Mississippi Valley included thermoluminescence analyses (Huntley and others, 1985; Pye and Johnson, 1988; Forman, 1989; Rodbell and others, 1997; Markewich and others, 1998b), amino-acid racemization of mollusks found in the loess (Miller and others, 1988; Clark and others, 1989; McCoy and others, 1990; Mirecki and Miller, 1994), electron spin resonance (Mirecki and Skinner, 1991; Skinner and Mirecki, 1993), and cosmogenic beryllium-10 (¹⁰B) analysis (Pavich and others, 1984; Pavich, 1993). Methods for dating organic material from loess deposits by radiocarbon (¹⁴C) analysis were also advancing, and proportional gas counting and accelerator mass spectrometer ¹⁴C analyses were also being applied to date organic material in loess (Rubin and others, 1993).

Chronostratigraphic data from the field investigations in the 1980s and 1990s indicated that the established late-middle Pleistocene and late Pleistocene loess stratigraphy of the central United States and northern Mississippi Valley could be extended into the lower Mississippi Valley as far south as west-central Mississippi and that age data were sufficient to correlate individual loess units with oxygen isotope stages and (or) marine isotope stages (fig. 1; Guccione and Rutledge, 1990; Follmer, 1996; Rutledge and others, 1996; Rodbell and others, 1997; Markewich and others, 1998b). From oldest to youngest, this loess stratigraphic sequence (fig. 2) is the Loveland Loess (MIS6, from about 185 ka to about 125 ka), Roxana Silt (MIS4, from about 70 ka to about 60 ka and MIS3, from about 60 to about 30 ka), and Peoria Loess (MIS2, from about 30 ka to about 12 ka) (age ranges for the marine isotope stages based on the LR04 stack in Railsback and others, 2015).

In the lower Mississippi Valley, the Loveland Loess and Roxana Silt each have at least one stratigraphically significant paleosol (fig. 2). The Sangamon paleosol, with its distinct yellow-red hue, high chroma colors (for example, 7.5YR 5/4 and 5/6), and well-developed pedologic structure, characterizes the upper part of the Loveland Loess. The Farmdale paleosol, with its much less red hue, lower chroma colors (for example, 10YR 4/3 and 4/4), and weak pedologic structure, characterizes the upper part of the Roxana Silt. Commonly, an unnamed paleosol is present in the lower part of the Roxana Silt and is superjacent to the Loveland Loess. The Sangamon paleosol, the unnamed paleosol, and the Farmdale paleosol are aggradational paleosols that formed during alternating periods of loess deposition and pedogenesis (Wang and others, 2009; Markewich and others, 2011). Table 1 shows a timeline summary for loess deposition and paleosol development in the lower Mississippi Valley that reached as far south as west-central Mississippi.

Roxana Silt Paleosol/Loess Chronostratigraphy at Phillips Bayou

As a result of their investigations in the lower Mississippi Valley, Markewich and others (1998a, b; 2011) suggested that late MIS5 (about 115 to about 80–70 ka) Loveland Loess deposition occurred during alternating climate intervals that were cool to cold and windy, and hot and dry, with Sangamon paleosol development occurring during each of the hot and dry climate intervals. Markewich and others (1998a, b; 2010, 2011) also suggested that in the lower Mississippi Valley, MIS4–MIS3 was a transitional period of regional climate cooling, with rapid, relatively short-term (millennial to decamillennial) climatic oscillations. The overall climate was cool to cold and subhumid to humid; Roxana Silt deposition rates were low (0.10–0.25 meters [m] per 1,000 years; Pye and Johnson, 1988; Markewich and others, 1998b), and chemical weathering and pedogenesis were sufficient for development of soils with some horizonation and weakly developed structure as seen in the unnamed paleosol in the lower part of the Roxana Silt and the Farmdale paleosol in the upper part of the Roxana Silt (fig. 2). The Phillips Bayou study locality of Markewich and others (1998b, 2011) in the lower Mississippi Valley was a 30-m-high exposure on the eastern side of Crowleys Ridge, about 1 kilometer (km) south of the former village of Phillips Bayou, Arkansas (Phillips Bayou, fig. 1B). The exposure was about 11 km north of Helena, Arkansas, and adjacent on the west to the St. Francis River near its confluence with the Mississippi River. The St. Francis River is a small tributary of the Mississippi River that flows southward from southeastern Missouri into northeastern Arkansas along the east side of Crowleys Ridge (fig. 1B). The drainage basin of the St. Francis River is too small to be shown at the scale of figure 1B. A summary of the paleosol and loess chronostratigraphy at Phillips Bayou is in Markewich and others (1998b, 2011), and is illustrated in figure 2. The age data for the Roxana Silt at Phillips Bayou from Markewich and others (2011) are summarized in table 2. A photograph of the Phillips Bayou exposure with contacts indicated between the Loveland Loess, Roxana Silt, and Peoria Loess, as well as the paleosol/loess contacts within the Roxana Silt, is shown in figure 3. Ward and others (1993) described the complete pedostratigraphy of the Phillips Bayou thick loess section, including the about 17-m-thick Peoria Loess. The Peoria Loess was sampled by coring (Helena No. 2 core; Ward and others, 1993) upslope from the Phillips Bayou exposure because the full thickness of the Peoria Loess was not accessible from the Philips Bayou cut-face. The Phillips Bayou cut-face exposed the basal Peoria Loess, the Roxana Silt, the Loveland Loess, the Crowleys Ridge Loess, and alluvial deposits The section from the Loveland Loess through the basal Peoria Loess is shown in fig. 3. The loess section was described and sampled on vertical faces on a series of steps cut into the section. This was a necessary field decision because of the section height, slope angle, and amount of recent colluvial overburden that needed to be removed.

The Farmdale paleosol in the upper part of the Roxana Silt was tentatively considered a Cryumbrept soil by Markewich and others (1998b). The Cryumbrept great group no longer exists in “Soil Taxonomy” (Soil Survey Staff, 1975); a currently used equivalent classification is a Humicryept (Soil Survey Staff, 1999). This classification indicates a soil that has minimal horizon development and formed when the mean annual soil temperature was higher than 0 degrees Celsius (°C) but lower than 8 °C at a 50 centimeter (cm) depth from the soil surface, and the mean winter and mean summer temperature was less than 6 °C difference at that same depth (Soil Survey Staff, 1999). Markewich and others (1998b) did not classify the unnamed paleosol in the lower part of the Roxana Silt, but noted that the unnamed paleosol had somewhat higher chroma, greater clay content, and stronger structure than the Farmdale paleosol, and possibly was welded to the underlying Sangamon paleosol.

Roxana Silt pedostratigraphy at Phillips Bayou is given in table 3 of this report and is as described by Ward and others (1993). Schoeneberger and others (1998, p. 38, Section 2, Profile/Pedon Description) defined soil structure as “the naturally occurring arrangement of soil particles into aggregates that results from pedogenic processes.” Where the term “structureless” is given as the soil structure grade, the soil is either massive or single-grained. The Farmdale paleosol that developed in the Roxana Silt is massive; thus, the term “massive” has been added to the original description of Ward and others (1993).

The Roxana Silt is pedologically modified throughout. The description by Ward and others (1993) for the Roxana Silt, including the Farmdale paleosol and the unnamed paleosol, is included here (table 3) only as reference for the scanning electron microscopy/energy dispersive X-ray (SEM/EDS) analysis images. The description uses the terminology of the 1993 edition of the “Soil Survey Manual” (Soil Survey Division Staff, 1993). The terminology and field methods used by Ward and others (1993) to describe the Farmdale paleosol and the unnamed paleosol were described in Schoeneberger and others (1998). All colors in the descriptions are for moist soil or sediment (Munsell Color Company, 1992). Horizon textures were field estimates, and particle size was subsequently measured by pipette and sieve analyses by the Kellogg Soil Survey Laboratory (KSSL), Lincoln, Nebraska. In the paleosol description, the horizon texture as determined from particle size analysis is given in brackets. Measured percentages of each size fraction and other soil characterization data for the Roxana Silt were determined by the KSSL. These laboratory data are available online in the National Cooperative Soil Survey Soil Characterization Database (National Cooperative Soil Survey, undated) using lab pedon numbers 92P1024 and 92P1025. Methods of soil analyses used by the KSSL were described in the laboratory methods manual (Soil Survey Staff, 2014). Although not included in the description, the Farmdale paleosol in the upper few meters of the Roxana Silt at Phillips Bayou shows evidence of soft-sediment deformation and is bioturbated (Markewich and others, 1998b).

SEM/EDS Investigations of the Roxana Silt at Phillips Bayou

At the Phillips Bayou exposure, samples of the Roxana Silt were taken for SEM/EDS analysis in 1992 (Ward and others, 1993) and 1994. Samples for SEM/EDS analysis of the Roxana Silt exposed in the Phillips Bayou quarry cut-face were taken originally from May 29 through June 5, 1992, by D.A. Wysocki. Markewich and others (1998a, b) discussed the initial results of this SEM/EDS investigation. The objectives of the SEM/EDS study were to consider the depositional environment, the post-depositional weathering intensity, and identify any discontinuities in the deposition of the Roxana Silt. Specifically, the SEM/EDS study was to identify changes in grain characteristics and morphology within the Roxana Silt for paleoenvironmental interpretation (Markewich and others, 1998a, b). Although the results were summarized in articles by Markewich and others (1998a, b, 2010), the mineralogical and morphological data from the SEM/EDS analysis were not published.

On June 22, 1994, J.B. Dixon and D.A. Wysocki took additional samples at the Phillips Bayou exposure to supplement the original sample set taken in 1992. All loess samples were examined and analyzed by G.N. White and J.B. Dixon in the facilities of the Soil and Crop Sciences Department, Texas A&M University, College Station, Texas. Observations and several images from their SEM/EDS analysis are presented in the Materials and Methods and the Results of SEM/EDS Analysis sections of this report. Depths for analyzed samples are the same as the horizon depths in the pedostratigraphic description of Ward and others (1993) (table 3). Soil horizon depths were measured from the top of the Farmdale paleosol at its contact with the overlying Peoria Loess. Horizon nomenclature and depths for samples of Roxana Silt examined for the SEM/EDS study are 2A1 (0–17 cm), 2Bt1 (96–140 cm), 2Bt2 (218–252 cm), 2Bt4 (359–387 cm), and 2ABt (594–623 cm). Samples from 0–387 cm depth are from the Farmdale paleosol that developed in the upper part of the Roxana Silt. The sample from 594–623 cm depth is from the unnamed paleosol that developed in the basal Roxana Silt. For discussion purposes, samples from 0 to 359 cm depth (2A1, 0–17 cm; 2Bt1, 96–140 cm; 2Bt2, 218–252 cm) are considered to be from the upper part of the Roxana Silt, and samples from 359 cm to 623 cm depth (2Bt4, 359–387 cm; 2ABt, 594–623 cm) are considered to be from the lower part of the Roxana Silt (fig. 2).

Materials and Methods

Samples were separated into sand (wet sieving), silt, and clay (centrifugation and sedimentation) fractions for further examination using a dilute sodium carbonate (Na₂CO₃) solution as a dispersant with no other pretreatments (Dixon and White, 1993, 1999). Carbonate minerals were not removed to allow examination of the surface textures of individual dolomite or calcite grains, if present. KSSL data show that the Roxana Silt 2A1 (0–17 cm) horizon has a calcium carbonate (CaCO₃) equivalent of 2 percent in the less than 2-millimeter (mm) size fraction, and that upon conducting the KSSL hydrochloric acid response test, reaction with hydrochloric acid responded slowly. These data indicate that the Roxana 2A1 horizon was enriched with dolomite by reworking from the overlying Peoria Loess. Optical grain counts for 300 grains conducted by the KSSL show abundant calcite in the Peoria Loess from 30 cm (11 percent calcite) and 160 cm (22 percent calcite) above the Roxana Silt contact. No grain count was conducted on the Roxana Silt 2A1 (0–17 cm) horizon. A grain count on the Roxana Silt 2A2 (17–43 cm) showed no calcite. Distinguishing dolomite from calcite is difficult in optical grain counts, so the KSSL grain counts include dolomite with calcite. Neither calcite nor dolomite was identified in the grain counts conducted for SEM/EDS analysis, which used the grain morphology and the energy dispersive X-ray (EDS) patterns.

Sand and silt grains were secured to a 10-mm diameter aluminum stub with carbon adhesive, coated with 100×10^–10 meter-sized gold (Au) particles to reduce charging, and examined using a JEOL 6400 scanning electron microscope (SEM) with a Tracor Northern EDS analysis unit. Samples were placed into the SEM and then examined with an operating voltage of 25 kiloelectron volts. Minerals were identified by using the grain morphology and the EDS patterns (table 4; Dixon and White, 1993,1999; White, 2008). Elements with an atomic number of 11 (sodium) or greater were assumed detectable by EDS. The morphology of sand- and silt-sized quartz and weatherable minerals was examined closely because these minerals aid in determining depositional environment and evaluating the post-depositional weathering regime. Grain counts were performed on samples to reduce bias in choosing representative grains for photography. In the sand size fraction, the number of counted grains ranged from 121 to 20. In the silt fraction, the number of counted grains ranged from 167 to 91. Many grains were heavily coated with iron (Fe) oxides, especially in the sand size fraction (fig. 4). These iron-oxide coatings hindered examination of grain morphology and mineral identification, which resulted in a low number of counted grains in the sand size fraction. The sand fraction for sample 2Bt1 (96–140 cm depth) had a total grain count of only 20.

Results of SEM/EDS Analysis

Grain counts and mineral identification performed on the samples during SEM/EDS examination showed the samples to be composed dominantly of quartz and feldspars with minor amounts of weatherable minerals (for example, pyroxene and amphibole), mica, and resistant heavy minerals (for example, rutile and sphene). The amounts of feldspars and quartz present in the samples are consistent across the sand and silt fractions, but the content of weatherable minerals, mica, and resistant heavy minerals is variable but somewhat higher in the silt fraction.

In both the sand and silt fractions, the upper part of the Roxana (0–359 cm depth, fig. 2) contains less quartz and more feldspar compared to the lower part of the Roxana (359–623 cm depth, fig. 2). Excluding the sample from the 2Bt1 (96–140 cm depth) because of the low total grain count, the sand fraction in the upper part has weighted average quartz and feldspar percentages of 65 percent and 34 percent, respectively. The sand fraction in the lower part has weighted average quartz and feldspar percentages of 76 percent and 21 percent, respectively. The silt fraction in the upper part has weighted average quartz and feldspar percentages of 54 percent and 34 percent, respectively. The silt fraction in the lower part has weighted average quartz and feldspar percentages of 72 percent and 25 percent, respectively. The higher quartz and lower feldspar values in the lower part possibly are due to incorporation of weathered quartz-rich material from the underlying Sangamon paleosol that developed in the Loveland Loess (fig. 2).

In the sand and silt size fractions in all samples, the potassium feldspar percentage was greater than the percentage of plagioclase. The percentage of intermediate plagioclase was greater than the combined percentage of calcium plagioclase and sodium plagioclase. The percentage of weatherable minerals, mica, and resistant heavy minerals ranged from 5 percent to 0 in the sand size fraction and from 9 percent to 0 in the silt size fractions. Other minerals occur in the Roxana Silt paleosols, but their morphology and chemistry are not distinctive enough for mineral identification without corroborating X-ray diffraction data.

Examination of the quartz sand surface features using the system of Krinsley and Doornkamp (1973) showed many of the grains to have cleavage faces and striated conchoidal fractures (fig. 5), although many grains were coated with iron oxides, making detailed examination of surface morphology impossible (fig. 4). Rare, rounded quartz sand grains were also observed (fig. 6A), and a close-up of the surface of the grain in figure 6B shows an unusual texture of deep, elongated pits separated by irregular ridges. The pits appear to be following a crystallographic direction from left to right in figure 6B, possibly the result of dissolution along structure or stress features. This kind of pitted, ridged texture was not recorded by Krinsley and Doornkamp (1973).

Quartz morphology in the silt was similar to that observed in the sand, with sharp edges, prominent striated conchoidal fractures, and cleavage features (fig. 7A, B) that suggest a glacial origin (Mahaney, 2002). Primary surface features resulting from deposition processes may have been hidden on many grains by silica overgrowths (fig. 8A, B) or by quartz crystal faces (fig. 8C, D) that resulted from extensive quartz overgrowths.

Feldspar grains in the Roxana Silt generally showed more signs of weathering than did quartz grains. Although most potassium feldspar grains, especially those of sand size, showed only minor weathering (fig. 9A), some silt-size potassium feldspar grains showed surface pitting (fig. 9B). Intermediate plagioclase grains (fig. 9C, D) were more weathered than the potassium feldspar and somewhat more weathered than end-member (anorthite, CaAl₂Si₂O₈) plagioclase grains. In some cases, etching of the plagioclase grains appeared to be crystallographically controlled (fig. 9C).

Other minerals also were observed in the samples from the Philips Bayou section, especially in the silt fractions. Pyroxenes and amphiboles were common constituents of the silt fractions of the entire section. Most grains appeared relatively unweathered (fig. 10A), but some were partially or highly weathered (fig. 10B). Titanium oxides appeared as rutile (fig. 11A), which probably is a primary mineral, as well as in secondary mineral forms (fig. 11B). Other mineral forms of titanium included sphene (fig. 11C) and ilmenite (fig. 11D). Muscovite also was observed, and most grains appeared to exhibit the initial stages of delamination due to potassium loss (fig. 12).

Summary of SEM/EDS analysis

The primary morphology of quartz sand grains from the Roxana Silt is irregularly shaped grains with sharp, conchoidal fracturing, striations, and an angular outline. This type of morphology occurs with high frequency in glacial environments because of grain-to-grain contact and high shear stress found in these kinds of environments (Mahaney, 2002; Woronko and Pisarska-Jamroży, 2016). There are few well-rounded sand grains, which are formed as a result of repeated saltation either in water or during eolian processes. The lack of rounded grains in the Roxana Silt suggests that most grains did not undergo multiple cycles of transport and deposition prior to loess transport. The limited grain-rounding reflects the initial glacial origin (pressure fracturing and shearing) and relatively rapid deposition via meltwater throughout the Mississippi Valley during deglaciation.

There were no major differences in the feldspars and other minerals as a function of depth in the Roxana Silt, which suggests that this loess was derived from a consistent source. The lack of large differences in weathering intensity with depth suggests a relatively consistent rate of sediment addition with no long periods of surface weathering. There also is a minimal degree of weathering near the top of the Roxana Silt in the Farmdale paleosol. One explanation for this lack of weathering differences might be that the mixing of new material by pedoturbation reduced the degree of weathering differences with depth. Alternatively, the weathered upper portion of the Roxana Silt may have been eroded prior to Peoria Loess deposition. Feldspar grains (fig. 9) in the Roxana Silt display slightly more surface weathering than quartz (figs. 7, 8), and this feldspar weathering would yield some clay, but in an insufficient amount for noticeable argillans or plasma development, which is consistent with the limited chemical weathering associated with Farmdale paleosol pedogenesis that has been documented in northeastern Illinois (Jacobs and others, 2009). Abundant dolomite in the 2A1 horizon of the Farmdale paleosol possibly is the result of pedoturbation during deposition of the overlying Peoria Loess, which contains abundant dolomite.

Observations