Rock magnetic properties of a soil developed on an alluvial deposit at Buttermilk Creek, Texas, USA



[1] The evolution of magnetization within a floodplain soil begins with initial deposition of magnetic particles during sedimentation and continues via subsequent alteration and growth of iron-bearing compounds by pedogenic and biologic processes. Measurements of soil magnetic properties capture information about the developmental history of the soil and are a convenient method by which to investigate environmental change and pedogenesis. Using a range of magnetic measurements, a comprehensive scenario for soil development was constructed for floodplain sediments at the Debra L. Friedkin site, an important archeological site near Buttermilk Creek, Texas. Floodplain deposits have traditionally been avoided for soil magnetism studies because it is thought that the episodic input of sediment would form soils characterized by discrete sedimentary units rather than a continuous record of pedogenesis. We demonstrate that alluvial deposits can sometimes carry a straightforwardly interpretable magnetic signal similar to those typically seen in loess deposits. Smooth variation of rock magnetic parameters as a function of depth also leads us to conclude that the soil at this site is largely undisturbed and that the age of lithic artifacts found within the soil may be interpreted within stratigraphic context.

1. Introduction

[2] Magnetic measurements of soils and paleosols are becoming increasingly useful tools for studies of past environmental conditions, sedimentology, and pedogenesis. Magnetic susceptibility values of paleosols have been used to estimate variables such as paleoprecipitation and paleotemperature [Maher, 1998; Maher et al., 2002, 2003]. Rock magnetic measurements enable estimations of the composition, concentration, and grain size distribution of magnetic minerals in a sample. These parameters can then be used to better understand soil development. Changes in the composition and concentration of magnetic minerals, e.g., changes in the amount of magnetite (Fe3O4) versus goethite (α-FeO·OH) or hematite (α-Fe2O3) as a function of depth, can be interpreted as the result of the formation or dissolution of Fe-bearing compounds during pedogenesis [Singer et al., 1996]. Rock magnetic measurements have the added advantage of being relatively rapid and inexpensive relative to other soil characterization techniques.

[3] Magnetic minerals in soil originate from a variety of biotic and abiotic mechanisms and are rarely the result of steady state processes. Measurements that focus on the concentration of magnetic minerals, such as magnetic susceptibility, are frequently central to soil studies [Fine et al., 1989, 1992; Singer et al., 1996; Lu, 2000] because concentration fluctuations are indicative of changing external forces. Wetter and/or warmer climates generally facilitate production of more magnetically soft Fe-bearing minerals, such as magnetite, than do cooler, drier climates [Maher, 1998; Geiss et al., 2008]. Forest and grass fires are known to produce surface layers enriched in magnetite [Oldfield et al., 1981]. Flyash from nearby coal-fired power plants can have a similar effect [Kapicka et al., 1999], and agriculturally tilled soils are typically homogenous as a result of repeated plowing. Biologically mediated production of magnetite can also be an important source of magnetic particles in soils. Magnetotactic bacteria produce magnetite that is single domain (SD) and occasionally pseudo-single domain (PSD) in size [Balkwill et al., 1980; Guyodo et al., 2006]. Dissimilatory iron-reducing bacteria (DIRB) produce grains that are predominantly in the superparamagnetic (SP) size range [Moskowitz et al., 1993].

[4] Soil parent material, or regolith, has a strong influence on soil magnetic properties. Much of the previous research on soil magnetism has focused on soils derived from wind-blown loess. Loess has the advantage of being a well-mixed mineral assemblage with relatively homogeneous magnetic properties [Heller and Evans, 1995]. Measurements that deviate from those of the parent loess can be readily interpreted as post-depositional changes related to pedogenesis [Singer et al., 1996]. Soils formed from more heterogeneous parent materials, such as floodplain deposits, are generally thought to be too complex for soil magnetism studies.

[5] Soil studies are further complicated when mixing occurs naturally within a soil. Vertisols, like the one in this study, undergo many wetting and drying cycles during their development, which causes the soil to form cracks as the constituent clay minerals expand and contract. These cracks may allow soil particles from higher in the soil profile to infiltrate downward, mixing the soil and potentially confusing the magnetic record of soil development.

[6] In this study we examine a vertisol that developed on floodplain deposits from Buttermilk Creek in Bell County, Texas. The two main goals of this study are to test whether this soil is dominated by episodic sediment input or by pedogenic processes (as is the case with loessic soils) and to test whether vertical mixing has occurred within this soil profile that would cause archeological artifacts to be found far from their original stratigraphic location. This latter goal is particularly important because the Debra L. Friedkin site is the location of an ongoing archeological investigation, where researchers are examining the age of stone artifacts based on their position within the soil stratigraphy [Waters et al., 2011].

[7] The Debra L. Friedkin site contains lithic artifacts that unambiguously predate Clovis settlement [Waters et al., 2011]. Until recently, the Clovis people were generally believed to be the first humans in North America [Goebel et al., 2008], with occupation between 12.8 and 13.1 ka. The pre-Clovis artifacts from the Debra L. Friedkin site are known as the Buttermilk Creek Complex and date between 13.2 and 15.5 ka. While this is not the only site that contains pre-Clovis artifacts, its more than 15,000 artifacts make it the most archeologically rich pre-Clovis site in North America. This site is also important because it records nearly continuous occupation during the transition from the Buttermilk Creek Complex to Clovis artifacts. Because of this site's importance, a variety of age and soil characterization data have already been collected and are readily available [Waters et al., 2011]. These ancillary data provide an unusually complete context for a rock magnetic study of pedogenic processes in the area.

2. Site Description

[8] A set of 55 oriented samples was collected from the Debra L. Friedkin archeological site near Buttermilk Creek in southwestern Bell County, Texas (Figure 1). The bedrock in this part of Texas is composed of the Lower Cretaceous Comanche Peak, the Edwards Formation, and the Georgetown Formation. These units consist primarily of clay/mud, limestone, dolostone, and chert. Overlying these Cretaceous rocks are unconsolidated colluvium and floodplain deposits, the latter of which have been weathered to form vertisols that extend over much of the region. Archeological artifacts at the Debra L. Friedkin site were collected from a soil that had developed entirely within a sequence of floodplain deposits. The lowermost four samples have a different parent material (limestone regolith) than the rest of the soil profile. Consequently, they have significantly different magnetic properties. Measurements from these samples have therefore been left out of some figures in this paper. The sampling site is upwind and nearly 80 km away from the nearest coal-fired power plant, which minimizes the likelihood of any significant contamination from modern anthropogenic flyash. Cubic samples (2.5 cm × 2.5 cm × 2.5 cm) were collected continuously from the surface to the limestone regolith (to a depth of 137.5 cm), which allowed complete coverage of the soil profile. The samples contain negligible quantities of plant roots and smaller rock pieces, and care was taken to avoid sampling cracks. Because most artifacts were not collected from cracks, this sampling method enabled us to test the soil in areas analogous to those from which the artifacts were collected. Plastic sample boxes were sealed, labeled for measurement, and stored in a magnetically shielded room with a background field less than 100 nT.

Figure 1.

Map of Buttermilk Creek, the Debra L. Friedkin site (labeled “Block A”), and the surrounding area, adapted from Waters et al. [2011]. The black rectangles and diamonds are trenches that were dug and areas that were excavated for the archeologic study. T-1 and T-2 are river terraces. Elevations listed are meters above sea level.

[9] The depositional history of the sediments was determined using optically stimulated luminescence (OSL) dating (Figure 2) [Waters et al., 2011]. OSL dates generally estimate when the sediment was last exposed to sunlight and represent the age of sedimentation. Sediment accumulation occurred largely between about 9,000 and 30,000 years BP (Figure 2) at roughly 0.5–0.75 cm/century. The most significant deviations occurred around 16,000 and 12,500 years BP (accumulation rates of 2–3 cm/century). These times correlate roughly with the Oldest (14.7 to 15.1 ka [Stuiver et al., 1995]) and Younger (11.6 to 12.9 ka [Alley et al., 1993]) Dryas, respectively.

Figure 2.

Compilation of measurements from the soil column at the Debra L. Friedkin site including an illustration of the soil with marked soil horizons (Vertisol: A-Bss; Colluvium: 2Bk). The column of circles represents the samples taken for OSL dating. The names along the far left are those of the archeological complexes. Periods of rapid deposition are labeled and roughly correspond to the Younger [Alley et al., 1993] and Oldest [Stuiver et al., 1995] Dryas (also marked). LCC and HCC correspond to the low and high coercivity components, respectively, identified during unmixing of the ARM and IRM data.

3. Methods

[10] Magnetic measurements were conducted at the Institute for Rock Magnetism at the University of Minnesota. Natural remanent magnetization (NRM), anhysteretic remanent magnetization (ARM), and isothermal remanent magnetization (IRM) acquisition and demagnetization experiments were conducted using a 2G Enterprises superconducting rock magnetometer. A Schonstedt alternating field (AF) demagnetizer was used for stepwise AF demagnetization of NRM using a ten-step demagnetization regime with steps at 1, 5, 10, 15, 20, 30, 40, 60, 80, and 100 mT. The same instrument was used to impart an ARM with a DC bias field of 0.05 mT. The ARM was acquired in 27 steps using fields up to 200 mT. An IRM was acquired at room temperature in 30 steps using fields up to 1.17 T by passing samples through a stable field on a Princeton Applied Research vibrating sample magnetometer (VSM). Unmixing of IRM and ARM acquisition curves was performed using IRMunmix version 2.2 software byHeslop et al. [2002], and the results from selected samples were cross-checked using the unmixing routine ofEgli [2004]. The differences between the components isolated using the Heslop et al. [2002] and Egli [2004]routines were negligible. Room temperature hysteresis measurements were measured on the same VSM using peak fields of 1 T. Room temperature bulk susceptibility was measured using an AGICO Geofysika KLY-2 Kappabridge magnetic susceptibility meter with an applied field of 300 Am−1 and a frequency of 920 Hz. Frequency dependent susceptibility (χfd) was measured on a Magnon variable frequency susceptibility meter with an applied field of 300 Am−1. Low-frequency susceptibility (χlf) was measured at 450 Hz, and high-frequency susceptibility (χhf) was measured at 4500 Hz.

[11] A Quantum Design Magnetic Properties Measurement System (MPMS) was used to measure a variety of low-temperature magnetic properties. A room temperature saturation IRM (RT-SIRM) was imparted to samples using a 2.5 T field and was measured during cooling to 10 K and during warming back to 300 K. A low temperature SIRM (LT-SIRM) was imparted using the same field strength on samples that had been cooled to 10 K in the absence (zero-field cooling, ZFC) or presence (field cooling, FC) of a 2.5 T magnetic field. The behavior of these LT-SIRMs was measured during warming back to 300 K. In-phase and out-of-phase susceptibility was measured from 10 to 300 K using frequencies of 1, 6, 32, 178, and 997 Hz. The 178 Hz data were removed because this frequency is near a harmonic of the 60 Hz current used in the United States, which caused excessive background noise.

[12] Samples measured on the MPMS were concentrated magnetic separates extracted from the soil. To produce these separates, a slurry of soil and deflocculant was shaken overnight and circulated past a strong magnet at decreasing speeds for two to four days to extract as many magnetic particles as possible. The separated particles were dried, packed into gelatin capsules, and stored at 4°C to minimize alteration.

4. Results

4.1. Susceptibility

[13] The bulk magnetic susceptibility of soils is typically controlled by the concentration of magnetic minerals. At the Debra L. Friedkin site, measurements of mass normalized susceptibility decrease smoothly and rapidly with depth (Figure 2). We observe no sudden changes associated with discrete depositional units and find no evidence of buried paleosols. Similar magnetic susceptibility profiles are observed in modern soils developed from loess deposits [Maher et al., 2003; Vidic et al., 2004], and the observed magnetic susceptibility pattern is interpreted here as a sign of uninterrupted soil development.

[14] χfd was calculated as a percentage using the relationship χfd = [(χlfχhf)/χlf] * 100. χfd values range from 5 to 12% (Figure 2); the upper soil column has values that are somewhat higher than those from the lower soil column. This is indicative of an increased SP component near the surface, which is common in soils where pedogenic processes produce SP grains [Maher, 1986].

4.2. Magnetic Remanence

[15] The characteristic remanent magnetization (ChRM) has an average direction of D = 335°, I = 58° (α95= 5.4°) which is broadly consistent with the present-day magnetic field direction at the site (Figure 3). The modern IGRF value at the site is D = 005°, I = 60°, while the inclination expected at this latitude for a geocentric axial dipole (GAD) is 51°. As with the magnetic susceptibility measurements, NRM intensity decreases progressively with depth (Figure 2). Stepwise AF demagnetization reveals a consistent ChRM direction in all samples (Figure 3). Weak secondary remanent magnetizations are fully removed by 5 mT. The mean ChRM direction does not directly overlap either the IGRF or the expected GAD directions. However, this inconsistency is minor and is likely due to a slight, systematic misorientation of samples during collection. The most important observations from the demagnetization experiments are that the soil contains sufficient quantities of fine-grained magnetic material to record the ambient magnetic field and that thousands of years of wetting and drying have not greatly disrupted the ChRM.

Figure 3.

Equal area plot of ChRM directions, which yield a mean direction of D = 335° I = 58° (α95 5.4°). The modern IGRF direction at this site is D = 005° I = 60° (black star), while the inclination expected at this latitude for a geocentric axial dipole is 51° (yellow circle). N = 26. Inset is from the sample at 21 cm depth and is a typical vector component diagram illustrating the stable paleomagnetic behavior observed during stepwise AF demagnetization. Both axes are listed in 10−8 Am2.

[16] The S-ratio provides a comparison between the concentrations of magnetically soft to magnetically hard minerals [King and Channell, 1991]. We applied a 1 T and an antiparallel 300 mT magnetic field to the samples. The remanent magnetization was measured after each applied field and compared using S = −IRM−0.3T/IRM1T. S-ratio values range between 0.9 and 1 in the upper 37 cm of the soil profile (Figure 2), which indicates that magnetically soft minerals such as magnetite or maghemite (γ-Fe2O3) (or their titanium-substituted equivalents) dominate in this interval. S-ratios decrease in a linear fashion until a depth of ∼120 cm, which indicates increasing relative abundances of high-coercivity minerals such as hematite and goethite. The final decrease in S-ratio at the base of the profile occurs across the transition into the limestone regolith.

[17] Acquisition and demagnetization of IRM and ARM can also be used to mathematically “unmix” coercivity spectra into lognormally distributed components [Heslop et al., 2002; Guyodo et al., 2006]. IRM acquisition measurements indicate trends similar to the S-ratio measurements. The sample at 20 cm illustrates the upper soil profile behavior and reaches saturation by 300 mT (Figure 4). This indicates that magnetically soft minerals such as magnetite or maghemite are the primary magnetic remanence carriers. By contrast, the sample at 120 cm does not reach saturation even after exposure to a 1.2 T field, suggesting that magnetically hard minerals, such as hematite or goethite, hold a portion of the remanence. IRM acquisition measurements were processed using IRMunmix by Heslop et al. [2002]and were best fit using two components: a low-coercivity component (LCC) with a mean of 1.5 log units (32.6 mT) and a relatively narrow dispersion parameter of 0.31 log units, and a high coercivity component (HCC) with a mean of 2.0 log units (116 mT) and a larger dispersion parameter of 0.75 log units. Similarly, ARM acquisition data were unmixed into a HCC with a mean of 1.4 log units (25 mT) and a dispersion parameter of 0.23 log units, and a LCC with a mean of 1.3 log units (17.2 mT) and a dispersion parameter of 0.37 log units. The characteristics of the IRM-LCC and the ARM-HCC are similar.

Figure 4.

Two representative soil samples with magnetic mineral assemblages that are dominated by (a) magnetically ‘soft material’ (e.g., magnetite and/or maghemite) that saturates by 300 mT, and (b) magnetically ‘hard material’ (e.g., hematite and/or goethite) that does not saturate by 1200 mT. “IRM Unmix 2” [Heslop et al., 2002] was used to determine lognormally distributed components within the sample. One component has a narrow distribution and likely represents the detrital and pedogenic contribution; the other has a more broad distribution and likely represents accumulated hematite/goethite.

[18] When interpreting these components, it is important to note that the population of grains that records an ARM is not necessarily the same population of grains that carry an IRM. ARM unmixing results encompass only grains with coercivities lower than the maximum AF, in this case 200 mT. In contrast, the IRM experiment captures the remanence held by grains with larger coercivities (up to 1.2 T). Both the IRM and the ARM unmixing results indicate that the samples are dominated by a component with a mean coercivity of ∼30 mT and narrow dispersion. This result is similar to a component that Egli [2004] identified as a combination of detrital magnetite and extracellular or pedogenic magnetite that is SP to SD in size. The concentration of all components decreases progressively with depth in a manner similar to the magnetic susceptibility and NRM intensity (Figure 2).

[19] The ARM/IRM ratio is a useful measure of the relative concentration of fine-grained, single-domain magnetic material and was calculated using the ARM acquired with a 100 mT AF and a 50μT direct current bias field. The IRM was acquired after exposure to a 104 mT DC field. Higher ARM/IRM values indicate an increase in fine-grained SD-like particles [King et al., 1982; Geiss et al., 2004]. In this study, ARM/IRM values indicate a higher concentration of SD-like particles in the upper 50 cm of the soil profile. Below this region, the concentration of SD-like particles begins to slowly decrease, following the trend of the S-ratio data.

4.3. Hysteresis Measurements

[20] Hysteresis parameters are summarized in the Day plot [Day et al., 1977] shown in Figure 5. As with several other magnetic parameters, saturation magnetization (Ms) decreases progressively with depth (Figure 2). Mr/Ms and Hcr/Hc ratios have a generally increasing trend with depth. This is consistent with an increased relative abundance of magnetically hard minerals, such as hematite or goethite [Jackson et al., 1990]. The greater variability in the coercivity ratio is unsurprising because its response to variations in mineralogy is not linear. Two minima at ∼80 cm and ∼100 cm correspond to perturbations in other magnetic properties such as S-ratio and ARM/IRM.

Figure 5.

Day plot of hysteresis parameters. The data plot in the PSD range, which is common for soils of this type.

4.4. Low-Temperature Measurements

[21] Magnetic extracts were collected for low-temperature magnetic measurements from multiple samples throughout the soil profile. The sample collected from a depth of 25 cm exemplifies the variation of susceptibility as a function of temperature and frequency (Figure 6). A major magnetic relaxation occurs between 10 and 60 K, followed by a steady relaxation up to 300 K. The lower temperature relaxation has a notable frequency dependence of in-phase susceptibility between 10 and 60 K and a peak in the out-of-phase susceptibility that migrates toward lower temperatures as the frequency of the applied field increases. The higher temperature relaxation is indicated by the more constant frequency dependence in the in-phase susceptibility data between 100 and 300 K and by the broad peak in the out-of-phase susceptibility data at ∼200 K. The low-temperature susceptibility results inFigure 6 closely resemble those of a synthetic multidomain (MD) TM40 sample (Fe2.6Ti0.4O4) [Church et al., 2011], which has a major relaxation around 70 K as well as a broad peak in the out-of-phase susceptibility around 225 K. If some form of titanomagnetite is present within the soil at the Debra L. Friedkin site, then presumably it was carried as part of the detrital load within the floodplain deposits.

Figure 6.

The (a) in-phase and (b) out-of-phase AC susceptibility for the soil at a depth of 25 cm indicate magnetic relaxation occurring primarily between 30 and 60 K, which suggests the presence of titanomagnetite with no sign of the Verwey transition.

[22] No evidence of the Verwey transition is seen in the low-temperature susceptibility data. By contrast, the Verwey transition is seen as a prominent drop in remanence at ∼110 K in measurements of the RT-SIRM during cooling (Figure 7). Because the Verwey transition is masked by small amounts of titanium substitution [Kakol et al., 1994] or high degrees of oxidation [Özdemir et al., 1993], it is clear that some stoichiometric magnetite is present, despite the lack of a Verwey transition in the susceptibility measurements. Upon continued cooling below the Verwey transition to 10 K, the remanence increases slightly. During warming back to 110 K, the remanence is not reversible. This has been interpreted as an indication of PSD maghemite [Smirnov and Tarduno, 2002]. On continued warming to room temperature there is no sign of the Verwey transition. This is typically inferred to indicate that SD grains hold the remaining remanence and that the difference in remanence before and after temperature cycling is representative of the MD portions of the magnetic mineralogy. RT-SIRM measurements of synthetic magnetite do not have the negative slope exhibited by samples from Buttermilk Creek. We interpret this behavior to be due to the presence of a second magnetic mineral phase, most likely goethite [Maher et al., 2004].

Figure 7.

Typical RT-SIRM plot with a drop in magnetization at 110 K, which is characteristic of the Verwey transition in nearly pure magnetite. The Verwey transition is visible in the RT-SIRM plots of all samples measured. This sample came from a depth of 55 cm.

[23] Zero-field cooled (ZFC) and field cooled (FC) low temperature SIRM are measured upon warming to room temperature (Figure 8). The ZFC/FC behavior shown in Figure 8could represent a sample consisting of goethite and progressively unblocking SP maghemite or a sample of goethite and MD titanomagnetite. Given that the Verwey transition is observed in the RT-SIRM experiments for all magnetic extracts, magnetite is present and it is reasonable to assume that some of this material has been oxidized to form maghemite. The relaxation observed at ∼70 K in low temperature susceptibility data suggests that detrital titanomagnetite may be present in the soil. Thus, the ZFC/FC behavior of the soil is likely a combination of both these effects.

Figure 8.

Zero-field cooled (ZFC) and field cooled (FC) low temperature SIRM (2.5 T) measured on warming for the same sample as inFigure 7. There is no evidence of a Verwey transition, and the data are more suggestive of a sample containing a goethite component.

5. Discussion

[24] Previous soil magnetism studies have avoided alluvial deposits because it was considered unlikely that discrete pulses of sediment would form a continuous magnetic profile that could be readily interpreted in terms of pedogenic development and environmental conditions. However, our data demonstrate that alluvial deposits can, in some circumstances, provide a clear record of the neoformation and breakdown of magnetic particles that occur during pedogenesis.

5.1. Sediment Mixing

[25] Waters et al. [2011]investigated the degree of mixing at the Debra L. Friedkin site. They found multiple lines of evidence indicating minimal and localized soil mixing. First, archeological artifacts in the soil are present in all sizes throughout the soil profile. Sediment mixing would have sorted the artifacts based on size because smaller artifacts and sediment are more apt to fall into cracks. This sorting is not seen at our site. Additionally, forty time-diagnostic artifacts are in correct stratigraphic order. Second, sediment filled cracks make up less than 2–12% of the sediment volume and the sediment between the cracks is intact. These cracks are 0.5 to 3 mm in diameter at their widest points and are largely non-vertical. They are also easily identifiable in the soil profile, which makes it possible to sample from un-mixed regions (as is the case in this study). Third, the distribution of clay, organic carbon, and calcium carbonate, in addition to the color, structure, and pH of the soil indicate that it developed following a mechanics model of Vertisol formation, which involves little vertical mixing.

[26] In this study, we elaborate on a fourth line of evidence from Waters et al. [2011], the soil magnetic data. The magnetic data indicate that it is unlikely that a significant amount of mixing occurred throughout the soil profile, as it would have resulted in a much more homogenous magnetic profile. As particles sifted downward through the soil column, they would have mixed magnetic properties from higher in the soil with the more developed lower sections. This would have obscured the progressive change in magnetic properties observed in our data. The saturation magnetization and susceptibility data both illustrate a peak in magnetic particles at the top of the soil profile and a subsequent decrease with depth. Further, the consistency of the ChRM directions is indicative of minimal mixing, as continual mixing would have randomized the remanence directions. While it is possible that some mixing occurred within the sediment cracks, the bulk of the artifacts were collected outside of sediment cracks, so those areas were of primary interest. Rock magnetic studies are therefore useful for establishing the degree to which a site has been disturbed.

5.2. Alluvial Soil Magnetism

[27] Despite the fact that the parent material for this soil is deposited episodically, the pedogenic processes that both form new magnetic particles and break down detrital magnetic particles dominate the magnetic behavior of this soil column.

5.2.1. Enhancement of Upper Soil

[28] Magnetically soft SP to SD grains, like those present in the upper soil at the Debra L. Friedkin site, are often the result of both inorganic and biologic processes. Magnetic susceptibility is useful for measuring the concentration of these particles. The susceptibility peak near the top of the soil column represents an enhanced concentration of ferrimagnetic particles, such as magnetite and maghemite, which has been found in numerous other soil studies [Singer et al., 1996; Geiss et al., 2004; Guyodo et al., 2006] and is probably the result of a pedogenic conversion of iron oxyhydroxides to iron oxides.

[29] The coercivity component common to the ARM and IRM acquisition unmixing analyses (∼30 mT mean with a narrow dispersion parameter) is probably a combination of grains formed in situ during pedogenesis and original detrital sediment [Egli, 2004]. As a vertisol, this soil would have experienced numerous wetting and drying cycles. These cycles would have caused reduction of iron in oxyhydroxides, like ferrihydrite (FeOOH•0.4H2O) and goethite, followed by oxidation of these ions into ferrimagnetic minerals, like magnetite or maghemite. Pedogenically formed grains like these are often SP to SD in size [Fine et al., 1995] with little size variation. The original detrital sediment presumably contains the PSD to MD titanomagnetite grains reflected in the low temperature variable frequency susceptibility data and in the ZFC/FC curves. The ARM-HCC and the IRM-LCC dominate the uppermost 30 cm of the soil profile. This is also the region previously identified using the S-ratio as being dominated by magnetically soft materials like magnetite and maghemite and as having a higher magnetic susceptibility. The ARM-LCC is most likely due to ultrafine pedogenic or extracellular SP magnetite grains [Egli, 2004]. The IRM-HCC is probably due to finely disseminated high-coercivity hematite or goethite grains, which dominate the IRM acquisition curves from deeper in the soil column.

5.2.2. Lower Soil Column

[30] In addition to the IRM acquisition, the S-ratio, Hcr/Hc, and Mr/Msall indicate an increase in hematite and goethite with depth. The concentration of magnetic particles (represented by the susceptibility) also drops significantly. By a depth of 62 cm, most (but not all) of the magnetic enhancement has been attenuated and magnetically hard components dominate. Because sediment accumulation and soil-forming processes have been roughly continuous over the life of the soil profile, the decrease in magnetic susceptibility with depth is probably the result of leaching of magnetic particles [Singer et al., 1996]. The magnetic signal of a loessic profile evolves over long periods of time, possibly over several hundreds of thousands of years [Heslop et al., 2000; Vidic et al., 2004]. In contrast, the oldest sediment in this column is only 24,000 years old, so evolution has apparently occurred more quickly here.

6. Conclusions

[31] The many wetting and drying cycles experienced by the soil at the Debra L. Friedkin site have converted iron oxyhydroxides like ferrihydrite and goethite into magnetite and maghemite, thereby enhancing the upper soil horizons in SP and SD magnetite, as seen within 30 cm of the surface. This uppermost 30 cm of material was deposited over the last 1000 years (Figure 2), which indicates that magnetic enhancement matures over centuries, with the subsequent oxidation and breakdown of magnetic particles subsequently taking over. As the upper soil layers are buried by new layers of sediment, the buried material is less affected by the pedogenic processes that formed the SP and SD magnetite. The magnetite and maghemite are then further converted to hematite or goethite, progressively decreasing the intensity of soil magnetization, which reaches a relatively steady state for material deposited before 9.2 ka. This is reflected in ARM and IRM acquisition experiments, which do not saturate at high fields; the S-ratio measurements, which decrease at these depths; and the Mr/Ms ratio, which increases with depth. These measurements are indicative of an increase in magnetically hard materials with respect to magnetically soft materials with depth.

[32] Though the processes at work in this site differ from those at work in loessic soil, the floodplain soil studied here exhibits many of the same magnetic mineral properties as loessic soils, and its constituent magnetic minerals contain information about the original sediment and the processes that have occurred since deposition. Typically, alluvial deposits are not used for environmental magnetic studies because episodic input of sediment is assumed to form discrete units of magnetic properties, rather than a continuous pedogenic signal. Furthermore, the wetting and drying cycles would cause cracking that can lead to progressive sediment mixing. Mixing would obscure the appearance of a uniform soil profile and could potentially displace artifacts within the soil column. However, the magnetic properties of the Buttermilk Creek soil are characteristic of an undisturbed site, where the soil appears to have developed gradually as it was deposited over time. All measurements indicate that the magnetic profile is dominated by typical biotic and abiotic pedogenic processes, and not by episodic alluvial inputs. Extensive dating of the soil column provides a framework through which we were able to determine that processes responsible for magnetic enhancement in the uppermost 30 cm of the soil profile occurred over a time span of centuries. Because of the undisturbed nature of the soil, artifacts within the soil can be interpreted to occur in primary archeological context, and the ages of the surrounding soil can provide a depositional time frame for the artifacts.


[33] This research was generously supported by a grant to JMF from the University of Minnesota. We are grateful to S. Webb for help with conducting measurements at the IRM. The Institute for Rock Magnetism is supported by a grant from the Instruments and Facilities Program, Earth Science Division, the National Science Foundation. The North Star Archeological Research Program established by Joe and Ruth Cramer funded fieldwork and later analysis at the site from 2006 to the present. Support was also provided by the Chair in First American Studies at Texas A&M University. This is IRM contribution 1102.