Agriculture Reference
In-Depth Information
systems. In the United States, Dr. Guy Smith was charged with the task of developing a system,
and he decided to be guided by principles and concepts that deviated from existing molds.
Hierarchies
Soil Taxonomy is structured as a nested hierarchy, in which classes of the lower levels are an
integral part of and conÝned by the deÝning properties of the classes at higher levels. With the
established linkages between the levels or categories in a hierarchy, the user can appreciate and
remember the functional relations of the classes. Use of a hierarchy is facilitated by appropriate
design of the nomenclature of the categories and the classes therein, in which the name signiÝes
the position of the category and its membership in the system, including some deÝning properties.
The process of building up hierarchies is generally one of aggregation. Entities with similar
behavior or that result from similar sets of processes are clustered together; unique objects may
deserve a separate cluster, or another rationale may be used in the aggregation process. The rationale
of this aggregation process leads to the principles upon which the system is developed.
Hierarchies are conceptual. However, the system that is developed has a deÝned purpose, which
is the making and interpreting of soil surveys. Thus it is desirable that each of the categoric levels
has a meaningful relationship to the range of scales to which the information could be applied.
This linkage to scale adds value to the system and assists in deÝning the functions of the category.
It is also a check on the information content (categoric relevance) in the context of the scale
(cartographic accuracy) as conceived by Orvedal and Edwards (1941). Unlike the hierarchy, which
is a conceptual construct, scale is real and quantiÝable. In using agro-ecological information, Fresco
(1995) indicates that there is an optimal scale at which each process can and must be studied, so
that scale-sensitivity is built into the system, helping to link one category to another.
A fundamental problem in developing hierarchies of natural systems is that they are continua,
rather than discrete objects. Breaking a continuum using deÝned limits of selected properties
introduces subjectivity into the system, and is frequently the source of disagreement. This is
particularly the case with soils for which not only the absolute value of the property has to be
considered, but also the position in the soil where that feature is exhibited. The case for discussions
on limits is reduced with the use of scientiÝc rationale, the selection of properties that have a
number of co-related properties, or a property that exhibits an extreme value from the population.
As the information content increases in the lower categories, it is in the categories that properties
with narrow limits are introduced to deÝne the categories. In the higher categoric levels, concepts
about processes and factors, or conditions that determine the processes, can be used to provide
aggregation of the lower level units. Table 4.1 lists the major deÝning process(es) employed at the
Order categoric level in Soil Taxonomy. DeÝnition of the processes is given by Buol et al. (1980).
All classiÝcation systems employ such sets of concepts, and Table 4.2 attempts to show how other
classiÝcation systems (Russian by Shishov and Sokolov, 1990; WRB by Deckers et al., 1998; and
Brasilian by SPI, 1999) deÝne their higher categories using similar concepts. These are more
abstract concepts based on the understanding of the nature of the population (Arnold, 1990).
Selection of Ñcontrols of processes,Ò or sets of conditions, also needs good rationale.
A number of soil classiÝcation systems have evolved with time. Soil Taxonomy has six cate-
gories and the World Reference Base (WRB) (Deckers et al., 1998), which is being considered as
a classiÝcation system, has only two. The objectives of a system obviously determine the number
of categories. The question, however, remains: Having decided on the number of categories, have
the objectives been satisÝed? Authors of systems usually indicate that the number is adequate for
their purpose. It would be useful to have some guidelines for making this determination. The
decision relates to the purpose of the classiÝcation and the function of the categories, and to the
scale at which the use of the category is relevant. Recently, TriantaÝlis et al. (2001) developed a
procedure to classify soil information. With the use of appropriate algorithms, they attempted to
reduce the subjectivity of conventional soil classiÝcations in deciding the number of classes and
