Results and Discussion

WDC Contents and Chemical Analysis of WDC Extracts

A number of replicate WDC measurements for the project soils were conducted to obtain reliable WDC values that are important in discerning relevant trends with respect to various measured soil parameters and estimating the degree of error and uncertainty associated with such measurements (Table 6). The operational nature of the WDC measurement demands that particular care be taken in replicating clay extractions. However, in certain instances, a visual evaluation of the suspensions prior to pipetting for quantification is necessary to detect potential sampling anomalies. For example, the Humatas soil, both the A and Bt2 horizons of the Puerto Rican Ultisol, appeared well-flocculated in WDC measurements; however, the soils displayed large floc volumes that in some instances was above the 2.5 cm sampling-depth threshold for the WDC analysis. Of the 33 soil samples included in the S207 Project, the following eight soil samples were found to have WDC clay contents below the threshold of the pipette method based on visual interpretation: Consumo B, Humatas A and Bt2, Wayah A2, Alachula Btg, Hayesville Bt2 and C2, and Cecil Bt. In many instances, these soils were excluded from statistical analysis because of the inability to obtain reliable WDC measurements.

Organic Matter Characterization

Both emission and synchronous scan analyses were performed on dry, powdered samples. The fluorescence data showed no consistent differences between water-dispersible clay and whole soil samples from a given soil. The spectra were characterized by broad peaks indicative of a host of different forms of molecular structures within the organic matter. Emission peaks in the 400-480 nm range, which occurred in many spectra are characteristic of aromatic ring structures with various configurations of attached carboxylic and phenolic groups (Senesi et al., 1991). A number of spectra exhibited a broad, intense emission peak between 330 and 400 nm. Various benzene ring derivatives containing amine functional groups and hydrolysis products of nucleic acids (e.g. adenine and guanine) have fluorescence maxima in this spectral region (Guibault, 1990).

Elemental Composition of Water-Dispersible Clays (Tables 8-11)

Based on mineralogical analysis, the soils were placed in one of six categories: smectitic, biorthic-mixed, biorthic-illitic, amorphic, monorthic, and sesquimonorthic. The elemental interpretations were based on two groups from the sample set with the 2:1 Group representing the smectic, biorthic-mixed, and biorthic-illitic categories, and the 1:1 Group representing the amorphic, monorthic, and sesquiorthic categories . For comparison, these two groups were further divided into surface and subsoil horizons (Table 7). The major subsoil horizons for the 1:1 Group, argillic horizons for Ultisols, and oxic horizons for Oxisols, produced very little WDC and were excluded from comparison.

Table 7. Grouping of samples which produced WDC considered in the elemental composition analysis (After Walthall and Haigler).

Series	State	Classification
	1:1 Surface Horizons
Nipe	Puerto Rico	Acrudox
Cecil	Georgia	Kanhapludult
Orangeburg	Georgia	Paleudult
Dyke	Georgia	Rhodudult
Hayesville	North Carolina	Kanhapludult
	2:1 Surface Horizons
Foley	Louisiana	Natraqualf
Maury	Kentucky	Paleudalf
Memphis	Kentucky	Hapludalf
Memphis	Tennessee	Hapludalf
Decatur	Alabama	Paleudalf
Marlton	New Jersey	Hapludalf
Sharkey	Louisiana	Epiaquert
	2:1 Subsoil Horizons
Foley	Louisiana	Natraqualf
Maury	Kentucky	Paleudalf
Memphis	Kentucky	Hapludalf
Memphis	Tennessee	Hapludalf
Kirkland	Oklahoma	Paleustoll
Marlton	New Jersey	Hapludult

In general, greater correlations were observed between the elemental composition of the WDC and the coarse clay compared to the fine clay or whole soil. Potassium in the WDC was closely associated with the coarse clay fraction of both the 2:1 and 1:1 surface horizons, with the same trend expressed to a lesser extent for the 2:1 subsoil horizons (Table 12). Sodium measured for the WDC, assumed to represent the cation exchange capacity (CEC) of the clays, was similar to the coarse clay with the exception of the Natraqualf subsoil (Foley Series) from Louisiana in which 80% of the total clay was water dispersible with 66% of the total clay in the fine clay fraction. Calcium linked the WDC with the coarse clay fraction in the 2:1 subsoils with the Ca oxide ratio for WDC/CC displaying a value close to one compared to the varying ratio for WDC/FC. The Si to Al ratios for both surface horizons indicated that WDC represents the average of the two bulk fractions, CC and FC.

Manganese appeared to be the only element to display a unique composition compared to the two bulk clay fraction. A comparison of the Mn oxide ratio for WDC/CC and WDC/FC was generally above 1, except for the glauconitic Ultisol from New Jersey (Marlton), indicating enrichment with respect to Mn in the WDC fraction.

X-Ray Diffraction and Thermal Analysis of WDC (Tables 14-16)

A variety of minerals were detected in water-dispersible (Table 1), fine- (Table 2) and coarse-clay (Table 3) fractions of the soil horizons sampled, exhibiting geographic groupings that relate to regional trends in parent materials (e.g., Mississippi Valley alluvium and loess vs. Piedmont saprolite) and weathering intensities (e.g., Puerto Rico vs. Oklahoma and Texas). Samples from soils forming in fine-textured, minimally-leached parent materials (e.g., Vertisols and vertic subgroups of Mollisols and Alfisols) (Fig. 1) tended to have high base status and to contain appreciable smectite; samples from more highly leached environments of the U.S., such as the Piedmont province, tended to contain a lot of kaolinite and hydroxy-interlayered minerals(HIM), along with some gibbsite and goethite; and a samples from Puerto Rico was dominated by gibbsite and goethite. Mica was present in significant amounts for loess-influenced soils of Kentucky, Louisiana, and Tennessee, and for a soil forming in glauconite-rich sediment of the New Jersey Coastal Plain. The smectite vs. 1:1 or HIM dominance proved to relate closely to dispersibility, which is discussed under water dispersible clay in this report.

Another significant consistent tendency revealed by the mineralogy data was that water-dispersible clay mineralogy commonly was more similar to the coarse clay than fine clay with respect to relative amounts of quartz vs. phyllosilicates (Fig. 1 and  Fig. 2). Quartz, as a component primarily of the coarse clay for most samples, is apparently relatively dispersible over the diverse range of samples represented. The least difference between fine- and coarse-clay mineralogy was observed for samples high in goethite and gibbsite. Even for these samples, however, the water-dispersible clay contained quartz.

All XRD, TGA, and DSC patterns can be viewed and customized plots generated on an interactive web site at Auburn University. To move directly to the XRD, TGA or DSC patterns, select the appropriate link below.

X-Ray Diffraction Analysis Database (XRD)
Thermal Gravimetric Analysis Database (TGA)
Differential Scanning Calorimetry Database (DSC)

Scanning Electron Microscopy to Determine the Morphology and Mineralogy of Readily Dispersible and Nondispersible Soil Fractions

Morphology and Mineralogy of Water-Dispersible Sands and Silts

Water-dispersible sand and silt from the 33 samples investigated were characterized by scanning electron microscopy (SEM) and energy-dispersive fluorescent X-ray (EDX) analysis to determine the morphology and mineralogy of selected particles particles. The full-text document describes the results of the SEM analyses in detail and includes many examples of interesting features found in the water-dispersible sand and silt fractions.

Statistical Evaluation of WDC and Solution Parameters

WDC as a percent of both the soil clay (WDC-% clay) and bulk soil (WDC-% soil) were compared to each factor using both a multivariant and pair-wise correlation method. Approximately 150 specific analyses conducted for each of the project soils were included in the statistical evaluation. The correlation matrices for each parameter were examined to identify clear trends, especially non-linear responses and obvious data outliers. Single-factor statistical correlations between the various solution parameters and WDC were generally poor, presumably due to the diverse nature of the samples included in the project.

Studies designed to show the influence of solution pH and cation and anion valence on soil dispersibility also observed complex trends that differed dramatically for soils with different mineralogies and organic matter contents. This suggests that identifying generalized trends with respect to clay dispersion for a diverse group of soils, such as those included in the S-207 project, may be difficult without first grouping soils based on other important criteria. Therefore, a multiple-factor approach that takes into account soil mineralogy and various soil solution factors should have much greater predictive capability. As an example, trends with respect to the elemental composition of the water-dispersible, fine, and coarse clay fractions of the project soils were more apparent when the soils were grouped according to their clay mineralogies and further divided into surface and subsurface horizons. The same designations adopted for elemental analysis were also used for other statistical comparisons.

When all soils were considered (Table 19), the highest absolute correlations with WDC-% clay were observed for CDB-ext. Al (-0.6234), CDB-ext Fe (-0.4184), AO-ext Fe (-0.4158) and delta pH_water (0.4491). Generally poor correlations were observed for Ex. Na percentage and SAR. For WDC on a % soil basis, soil pH, suspension (0.4724) and saturated paste SAR (0.8040) were positively correlated and CDB- ext. Fe (-0.2279) and CDB-ext. Al (-0.4896) were negatively correlated with WDC.

Table 19. Select pairwise correlations for WDC (Astatula C excluded).

Variable	Correlation	# Observations
	WDC (%Clay)
pH(CaCl2)	0.2491	30
pH(water)	0.2885	30
CDB-Fe	-0.4184	28
CDB-Al	-0.6234	27
Acidity	-0.4136	29
Ex. Al.	-0.04706	12
Ex. Na%	0.0931	30
%Clay	-0.5572	30
Dissolved Organic Carbon	0.6621	29
delta pH(CaCl2)	0.4491	28
delta pH(water)	0.4076	28
AO-ext Fe	-0.4158	30
	WDC (%Soil)
pH(CaCl2)	0.5510	30
pH(water)	0.5780	30
CDB-Fe	-0.2279	28
CDB-Al	-0.4896	27
NH4OAc Ca	0.5432	28
NH4OAc Mg	0.6889	29
Base Sat%	0.5806	29
Base Sat % NH4OAc	0.4106	30
Sat Paste SAR	0.8040	11
Susp SAR	0.4724	30
delta pH(CaCl2)	0.7459	28
delta pH(water)	0.7090	28
Susp pH	0.6530	30

1:1 Surface and Subsurface

For all 1:1 samples (Table 20), both the highly dispersive surface soils and the generally non-dispersive subsoils. Negative correlations between CDB-ext Al (-0.6601) and -Fe (-0.4788) and WDC were observed. Generally poor or negative correlations (abs. value < 0.5) were observed for soil pH, solution phase Na, SAR, CEC, electrical conductivity, bulk-soil organic carbon content. However, this most likely reflects the high correlation between solution phase Na concentration and total soluble salts as indicated by electrical conductivity. High correlation with WDC was also observed with soluble organic carbon (0.2µm filter pore size) which suggests that properties controlling colloid stabilization also influence organic solubility.

Table 20. Select pairwise correlations for WDC for 1:1 soils, all horizons.

Variable	Correlation	# Observations
	WDC (%Clay)
%Clay	-0.7529	16
CDB-Fe	-0.3620	15
CDB-Al	-0.6505	16
Dissolved Organic C	0.6872	15
Sat. Paste SAR	-0.51	5	corr EC
Susp. SAR	-0.1768	16
	WDC (%Soil)
CDB-Al	-0.6153	16
Base Sat%	0.3557	16
delta pH(CaCl2)	0.4014	16
delta pH(water)	0.3076	16
Sat. Paste SAR	-0.4587	5	corr EC
Susp. SAR	-0.0006	16	see above

1:1 Surface Soils

When statistical analysis was restricted to just the surface horizons, negative correlations were observed for Organic-C (-0.6608), CDB-Fe (-0.6190), and CDB-Al (-0.8254) (data table not included).

Less-Weathered Soils

As a percentage of total clay, WDC was fairly-well correlated with various soil pH measurements (Table 21), as well as suspension (0.6065) and saturated-paste SAR (0.4335) values.

Table 21. Select pairwise correlations for WDC for less-weathered soils, all horizons.

Variable	Correlation	# Observations
	WDC (%Clay)
pH(CaCl2)	0.4141	14
pH(water)	0.4640	14
CDB-Fe	-0.5658	13
CDB-Al	-0.6692	11
Sat Paste SAR	0.5667	6
Sol. Org. Carbon	0.6620	14
Susp SAR	0.4335	14
delta pH(CaCl2)	0.6670	12
delta pH(water)	0.6408	12
Susp. pH	0.6065	14
	WDC (%Soil)
pH(CaCl2)	0.5262	14
pH(water)	0.5603	14
CDB-Al	-0.4657	11
% Clay	0.4686	14
% Fine Clay	0.7470	14
CEC	0.3737	12
Ex. Acidity	-0.3795	13
Ex. Al	-0.7660	4
Ex. Na %	0.5373	14
Sat. Paste SAR	0.8332	6
Sol. Org. Carbon	0.5576	14
Susp. SAR	0.4216	14
ZPSE	-0.4101	12
delta pH(CaCl2)	0.7067	12
delta pH(water)	0.6856	12
Susp. pH	0.6662	14

Step-Wise Regression Analysis

Astatula C, the only Entisol sample in the study set, and the Decatur samples were excluded from statistical analysis. A set of 22 parameters were included in a step-wise regression analysis to develop a model for predicting the dispersive properties of soils from within the Southern Region. Suspension composition data were used instead of saturated-paste results because of the greater detection limits provided a greater range of values for comparison; however, the two measurements were highly correlated (r > 0.82) when both values were compared.

The following step-wise regressions were performed:

• All Soils

WDC (% Soil) All soils excluding Heiden, Decatur, Astatula

WDC (% Clay) All soils excluding Heiden, Decatur, Astatula

• Ultisols and Oxisols

WDC (% Soil) All horizons

WDC (% Clay) All horizons

WDC (% Soil) Surface soils

WDC (% Clay) Surface soils

• Less-weathered soils (Alfisols, Entisols, Inceptisols, Mollisols and Vertisols)

WDC (% Soil) All horizons

WDC (% Clay All horizons

Table 22. Parameters considered in step-wise regression.

Parameters
%Clay		Complete
% Fine Clay
pH(H2O)
pH(CaCl2)
delta pH = pH(water) - pH(CaCl2)
CDB-Ext. Fe		2 TR values (Pomona, Heiden)
CDB-Ext. Al		4 TR values (Foley Bt, Booneville Ap, Astatula, Heiden)
Total Organic Carbon
NH4OAc Extractable Acidity		1 calcareous soil (Heiden
NH4OAc CEC
NH4OAc- Ext. % Base Saturation
Sol. Organic Carbon (0.2µm pore size)		1 missing value (Hayseville C2)
Susp. pH
Susp. EC (µS cm-1)
Susp. Na+ mg L-
Susp. K+ mg L-1
Susp. Mg2+ mg L-1
Susp. Ca2+ mg L-1
Susp. SAR
PZSE		2 missing values (Wayah Ah, Heiden)
water delta PZSE = PZSE - pH(water)		2 missing values (Wayah Ah, Heiden)
CaCl2 delta PZSE = PZSE - pH(CaCl2)		2 missing values (Wayah Ah, Heiden)

Table 23A. Step-wise regression results (WDC % soil) for all soils.

Step	Parameter	R2	Cp
1	Delt water pH	0.5564	19.709
2	NH4OAc CEC	0.6502	12.468
3	CDB-Al	0.7089	8.682
4	Susp pH	0.7564	5.9964

Table 23B. Step-wise regression results (WDC % clay) for all soils.

Step	Parameter	R2	Cp
1	CDB-Al	0.438
2	%Clay	0.5005
3	CDB-FE	0.5844

Table 23C. Step-wise regression results (WDC % soil) for Ultisols, all horizons.

Step	Parameter	R2	Cp
1.	CDB-Al	0.3280
2.	Base Sat NH4OAc	0.7067
3.	NH4OaC CEC	0.8261
4.	% Fine Clay	0.8637
5.	Susp Ca (mg/L)	0.9136
6.	Delta water pH	0.9341
7.	CDB Fe	0.9500

Table 23D. Step-wise regression results (WDC % clay) for Ultisols, all horizons.

Step	Parameter	R2	Cp
1.	CDB-Al	0.4522
2.	Susp SAR	0.5637
3.	pH water	0.7084
4.	CDB-Fe	0.7726
5.	Susp Ca (mg/L)	0.9136
6.	Delta water pH	0.9341
7.	CDB Fe	0.9500

Table 23E. Step-wise regression results (WDC % soil) for Ultisol surface horizons.

Step	Parameter	R Square	Cp
1.	CDB-Al	0.6229
2.	Base Sat NH4OAc	0.8229
3.	Susp SAR	0.9397
4.	Susp Ca (mg/L)	0.9829

Table 23F. Step-wise regression results (WDC % clay) for Ultisol surface horizons.

Step	Parameter	R Square	Cp
1.	CDB-Al	0.8714
2.	CDB-Fe	0.9877

Table 23G. Step-wise regression results (WDC % soil) for less-weathered soils, all horizons.

Step	Parameter	R Square	Cp
1.	% Fine Clay	0.7371
2.	Susp SAR	0.8690