42
ecosystems (see Chapter 23) include fire (Christensen et al., 1989; Baker, 1992a, b; Turner et aI.,
1994) and grazing (McNaughton, 1985), and for
aquatic systems (see Chapter 26) include flooding
(e.g., Poff and Ward, 1989; Sparks, 1995) and fluvial geomorphologic processes (Ligon et aI., 1995).
As a consequence of the dynamic nature and hierarchical structure of ecosystems, there are likely to
be limits to the predictability of natural patterns and
processes at different scales (Costanza et aI., 1993).
There are three general steps in conducting
ecosystem characterization: (1) development of a
hierarchical land classification of ecological units,
including both a geographical or regional definition and a biological definition of the classes
(Austin and Margules, 1986; Mackey et aI., 1988;
Belbin, 1993); (2) definition of the relevant ecological properties (e.g., processes) of these units;
and (3) a method of allocating areas to the ecological units (mapping). Data resulting from ecosystem characterization can be used to assess the ecological conditions of land areas and to provide the
framework for making regulatory and land management decisions (see Chapter 1; also see an example for regional conservation planning in Chapter 20).
Although much is known about ecosystem function and landscape dynamics, many questions need
to be answered, including how to integrate population and landscape ecology, how to interpret the
ecological effects of spatial patterns and changes
in these patterns, and how to identify controls on
ecological processes at different spatial scales
(Turner et al., 1995). Ecosystem characterization
provides a basis for answering many of these questions and for overcoming two major challenges to
broad-scale ecological studies (Turner et aI., 1995):
assembling spatial databases over large areas (see
Chapters 5 through 7) and offering alternative approaches to traditional experiments.
3.3 Delineating Ecosystem Units
Two main aspects of ecosystem characterization
are classification and regionalization-mapping.
Characterization shares with classification the fact
that it is an abstraction (Zonneveld, 1994; Rowe,
1996). During the characterization process, we distinguish guiding principles, properties, and diagnostic characteristics (Zonneveld, 1994). Guiding
principles are rules for choosing and calibrating
properties and assessing the hierarchy of an ecological classification. These rules are derived based
on intrinsic characteristics of the system (e.g.,
Ecosystem Characterization and Ecological Assessments
ecosystem function), relationships with the surroundings, or the genesis of the system, depending
on the purpose of the characterization. The general
guiding principle of characterization is the convergence of evidence (Zonneveld, 1994), which means
that properties are chosen that result in convergence
on an object (e.g., an ecological unit) or class of
objects and lead to contrast of this object or class
of objects with others. Properties are all the attributes of the object to be characterized; diagnostic
characteristics for a class (or ecosystem) are selected from the suite of properties. Diagnostic characteristics should be properties of the entire system
and not of the component parts (see Section 3.6),
should be functional, and should correspond to the
phenomenon of interest.
For example, the process of landform genesis is
a logical guiding principle for geomorphological
classifications, and the degree and form of slopes,
sedimentation, and erosion pattern constitute properties. Slope degree, shape of slope, and horizontal configuration of slope types (pattern) are often
used as diagnostic characteristics of geomorphological classifications. For vegetation classification, the development of stable successional stages
can be distinguished as a guiding principle, and the
vertical structure of the vegetation and floristic
species composition are therefore good examples
of measurable properties that can be used as diagnostic characteristics. This approach has also been
used for soil, geological, and landscape classifications (see discussion in Zonneveld, 1994).
The degree of human influence in the landscape
is important in selecting guiding principles and diagnostic characteristics. In landscapes that can be
described as ranging from natural to seminatural,
geomorphologic and climatic factors, in combination with pedogenic processes, may have the greatest constraining effect on ecosystem patterns. In
contrast, land-use history may be the most important constraining factor in cultural landscapes
(Haber, 1994; Zonneveld, 1994).
Mapping is a visual representation of the classification units. Much had been said about the merits of a priori versus a posteriori classifications during mapping efforts (e.g., see the classic work of
Kuchler, 1973; also see Kuchler and Zonneveld,
1988). The mapping process forces a surveyor to
confront all aspects of the landscape (Rowe, 1996).
Although confusing samples (e.g., outliers) can be
ignored in a classification framework, mapping necessitates the delineation of units over the entire
surveyed area. The method for recognizing units,
that is, the delineation of mapping unit boundaries,
constitutes a set of de facto hypotheses about the
ecosystems (see Chapter 23) include fire (Christensen et al., 1989; Baker, 1992a, b; Turner et aI.,
1994) and grazing (McNaughton, 1985), and for
aquatic systems (see Chapter 26) include flooding
(e.g., Poff and Ward, 1989; Sparks, 1995) and fluvial geomorphologic processes (Ligon et aI., 1995).
As a consequence of the dynamic nature and hierarchical structure of ecosystems, there are likely to
be limits to the predictability of natural patterns and
processes at different scales (Costanza et aI., 1993).
There are three general steps in conducting
ecosystem characterization: (1) development of a
hierarchical land classification of ecological units,
including both a geographical or regional definition and a biological definition of the classes
(Austin and Margules, 1986; Mackey et aI., 1988;
Belbin, 1993); (2) definition of the relevant ecological properties (e.g., processes) of these units;
and (3) a method of allocating areas to the ecological units (mapping). Data resulting from ecosystem characterization can be used to assess the ecological conditions of land areas and to provide the
framework for making regulatory and land management decisions (see Chapter 1; also see an example for regional conservation planning in Chapter 20).
Although much is known about ecosystem function and landscape dynamics, many questions need
to be answered, including how to integrate population and landscape ecology, how to interpret the
ecological effects of spatial patterns and changes
in these patterns, and how to identify controls on
ecological processes at different spatial scales
(Turner et al., 1995). Ecosystem characterization
provides a basis for answering many of these questions and for overcoming two major challenges to
broad-scale ecological studies (Turner et aI., 1995):
assembling spatial databases over large areas (see
Chapters 5 through 7) and offering alternative approaches to traditional experiments.
3.3 Delineating Ecosystem Units
Two main aspects of ecosystem characterization
are classification and regionalization-mapping.
Characterization shares with classification the fact
that it is an abstraction (Zonneveld, 1994; Rowe,
1996). During the characterization process, we distinguish guiding principles, properties, and diagnostic characteristics (Zonneveld, 1994). Guiding
principles are rules for choosing and calibrating
properties and assessing the hierarchy of an ecological classification. These rules are derived based
on intrinsic characteristics of the system (e.g.,
Ecosystem Characterization and Ecological Assessments
ecosystem function), relationships with the surroundings, or the genesis of the system, depending
on the purpose of the characterization. The general
guiding principle of characterization is the convergence of evidence (Zonneveld, 1994), which means
that properties are chosen that result in convergence
on an object (e.g., an ecological unit) or class of
objects and lead to contrast of this object or class
of objects with others. Properties are all the attributes of the object to be characterized; diagnostic
characteristics for a class (or ecosystem) are selected from the suite of properties. Diagnostic characteristics should be properties of the entire system
and not of the component parts (see Section 3.6),
should be functional, and should correspond to the
phenomenon of interest.
For example, the process of landform genesis is
a logical guiding principle for geomorphological
classifications, and the degree and form of slopes,
sedimentation, and erosion pattern constitute properties. Slope degree, shape of slope, and horizontal configuration of slope types (pattern) are often
used as diagnostic characteristics of geomorphological classifications. For vegetation classification, the development of stable successional stages
can be distinguished as a guiding principle, and the
vertical structure of the vegetation and floristic
species composition are therefore good examples
of measurable properties that can be used as diagnostic characteristics. This approach has also been
used for soil, geological, and landscape classifications (see discussion in Zonneveld, 1994).
The degree of human influence in the landscape
is important in selecting guiding principles and diagnostic characteristics. In landscapes that can be
described as ranging from natural to seminatural,
geomorphologic and climatic factors, in combination with pedogenic processes, may have the greatest constraining effect on ecosystem patterns. In
contrast, land-use history may be the most important constraining factor in cultural landscapes
(Haber, 1994; Zonneveld, 1994).
Mapping is a visual representation of the classification units. Much had been said about the merits of a priori versus a posteriori classifications during mapping efforts (e.g., see the classic work of
Kuchler, 1973; also see Kuchler and Zonneveld,
1988). The mapping process forces a surveyor to
confront all aspects of the landscape (Rowe, 1996).
Although confusing samples (e.g., outliers) can be
ignored in a classification framework, mapping necessitates the delineation of units over the entire
surveyed area. The method for recognizing units,
that is, the delineation of mapping unit boundaries,
constitutes a set of de facto hypotheses about the