Defining Geographic Soil Bodies by Landscape Position, Soil Taxonomy, and Cluster Analysis

fying landscape attributes which are believed to contain similar soils (Hudson, 1992). “Landscape position” can The key Soil Taxonomy is based upon the idea that certain properbe considered to be a geographic approach to classificaties can be used to define populations of soils from the soil continuum. tion, in which combinations of surficial and stratigraphic The soil mapping paradigm is that similar soil populations exist within landforms. High taxonomic variability has been reported within nuattributes are used to identify populations of related merous soil mapping units. The hypotheses in this paper are that for soil individuals within a landscape (Ruhe, 1956). Thus, large scale mapping (i) Soil Taxonomy creates classes that are only to the extent that taxonomic classes contain soil attripartially related to landform and (ii) more homogeneous soil classes butes that are correlated with landforms and/or other exist if different defining constructs are used. The objectives are to: identifiable surficial attributes within a particular soil (i) classify a sampling of soils within a 40-ha upland pasture using landscape, Soil Taxonomy is a useful mapping guide. both Soil Taxonomy and cluster analysis, (ii) identify the distinctness The high taxonomic variability reported for many soil and relationships of these samples to landforms, and (iii) compare map units (e.g., Powell and Springer, 1965; Wilding et the geographic distributions of soil classes identified by Soil Taxonal., 1965; Edmonds et al, 1985; Young et al., 1998) indiomy and cluster analyses. Ninety-four soil properties were measured cates that some map units contain numerous landforms, from 257 pedons along point transects. Cluster analysis identified three pedologically and geographically distinct groups. A single cluster or that taxonomic groups do not coincide with landform group was identified for soils within the “ridge” landform, whereas the groups, or both. The relevant question then becomes, “backslope” landform was a mixture of all three groups. A significant “Do taxonomic classes represent the relationships of relationship was found for soil attributes and slope profile curvature soil attributes with terrain attributes across the areal within the backslope landform, but predictive value was low. Soil extent of the soils?” Taxonomy produced 13 geographically indistinct classes that were Cluster analysis is a general term for a family of statispartially related to cluster groupings. Cluster analysis appears to be tical classification methods that group objects. The idea useful for revealing patterns of soil homogeneity and for identifying is statistically to minimize within-group variability while relationships among soil properties and landforms. Numerical analysis maximizing among-group variability in order to produce may be a helpful supplementary method for correlating soil surveys relatively homogeneous groups that are distinct from with large soil databases, or for defining those soil attributes which distinguish mappable bodies from the soil continuum. one another. Statistical groupings are unique to the data used (and to the data collection methods), and are statistically referenced to multivariate group centroids which have distinct group boundaries that are defined by S and temporally variable geomorphic and means and variances in multidimensional space. This pedogenic processes create different soils, but siminumerical approach would seem to be conceptually well lar soils can be identified and grouped with classification suited to the methods and objectives of mapping soils schemes. The characteristics and numbers of identified in the field. Bidwell and Hole (1964a) were among the soil groups depend on the kind of classification scheme first to recognize the potential of numerical methods, (Joel, 1926; Cline, 1949). Soil Taxonomy (Soil Survey including ordination, to refine soil classification. Rayner Staff, 1992), the classification system used within the (1966) suggested that cluster analysis be used to “pick National Cooperative Soil Survey (NCSS), is a hierargroups of profiles ... to act as standards with which to chical approach with clearly defined, invariant group compare other profiles.” Arkley (1971, 1976) advocated boundaries. The hierarchical approach is conceptually cluster analysis as an objective means of developing a unsatisfactory to some soil scientists, who argue that soil classification, and illustrated his philosophy with a similar pedons can be classified into separate classes on cluster analysis of a diverse soil dataset (Arkley, 1976). the basis of differences which may be unimportant for Virtually all applications of cluster analysis in soil scicertain interpretations or uses (Webster, 1968). For exence literature have been “hierarchical agglomerative” ample, Young et al. (1998) found that interpretive purity methods, as defined by Aldenderfer and Blashfield was high in floodplain soils whose subsurface textural variabilities resulted in a high number of taxonomic (1984). Webster and Oliver (1990) have illustrated clusclasses. ter analysis techniques using soil data. Several authors Soils are grouped in field mapping primarily by identi(Moore and Russell, 1967; Campbell et al., 1970) have considered and compared various clustering methodologies, including different multivariate space distance F.J. Young, GIS Lab, 306 Founders Hall, Lincoln Univ., 820 Chestnut measures. St., Jefferson City, MO 65102; R.D. Hammer, Soil Genesis, School of Cluster analysis has not been used extensively to creNatural Resources, Univ. of Missouri-Columbia, 302 ABNR Building, Columbia, MO 65211. Contribution from the Missouri Agric. Exp. ate new soil classification criteria, although it has been Stn. J. Series No. 12,451. Received 17 Dec. 1996. *Corresponding used to develop land-based ecological classification author ([email protected]). schemes. Lentz and Simonson (1987) used cluster analysis to identify six vegetation classes, then examined the Published in Soil Sci. Soc. Am. J. 64:989–998 (2000).