30
N+l
FIGURE 2.1. The hierarchical approach to ecosystem
analysis suggests that the phenomenon of interest, level
N, is created and maintained by mechanisms that are subsystems, level N - 1 of the system (N). The level N +
1 is the context of the level of interest and provides
higher-order control of the system.
has important substructure. Clearly, a management
unit we choose to study may be composed of subunits of soil or vegetation type, while the unit itself
may be only part of a watershed.
Any ecosystem is a hierarchy of differently
scaled structures and functions. The structure and
function at a particular level (Figure 2.1) that we
choose to observe, say a forest stand, has a particular spatiotemporal scale. This is the N level. The
lower level (N - 1) is composed of subsystems, the
components that together comprise the forest stand.
Depending on the process under investigation, the
relevant components might be patches or gaps or
individual trees, all of which have smaller and
quicker spatiotemporal dynamics and provide a
mechanism for creation and maintenance of the
stand. The higher level (N + 1) is larger in space
than the forest stand, but changes more slowly. This
larger, slower level might be the region, which exerts control on the forest stand through topography
and mesoscale climate.
Hierarchy theory (O'Neill et al., 1986, 1989) applied to systems implies such upper-level control
and lower-level mechanisms and provides an epistemology for measurement based on the scale of
observation and level of analysis. Of course, the
level of interest (N) and the N - 1, N + 1 levels
are set by the problem being investigated and may
not be the same hierarchy for all problems. Depending on the criterion employed, many different
subsystems (N - 1) can be defined for any particular level (N). Observer perception determines the
A Theoretical Framework for Ecological Assessment
relevant levels (MacMahon et al., 1978). Characterizing the hierarchical organization of an ecosystem includes determining the spatial and temporal
bounds of each pattern or process and the order in
which patterns or processes are nested within the
hierarchy (Urban et al., 1989). At any scale, we can
observe many different types of things. What is included for analysis depends on what is important
for the study.
The description of the system as a hierarchy of
mechanisms (N - 1), phenomena (N), and controls
(N + 1) helps in sorting though the complexity by
isolating the dynamics of interest and structuring
the research around important interactions (Johnson, 1996). This is the art. Leave out everything
that is unimportant in sorting through the complexity. Find the correct criterion to address the
problem. The hard part is to determine the correct
level in the system, that which is truly important to
the observer. It may be a single species or a biotic
community or an ecosystem type. Choose the important level N. Then carefully, remaining within
the same hierarchy, identify the N + 1 level of control and the N - 1 level of mechanisms. These
should drive our investigations whether they are
field tests or modeling, applied or theoretical
(Turner, 1995).
2.2.2 Grain and Extent
Imagine casting a net into a sea populated with fish
of widely varying sizes. The net will capture only
a certain range of body sizes. Fish smaller than the
mesh size can slip through, and fish larger than the
net can hold will simply swim around the edge
(O'Neill et aI., 1986; Wiens, 1989; Ah1 and Allen,
1996). The net represents our observational protocol, the mesh size is analogous to grain (the smallest features that can be retrieved from the data), and
the size of the net determines the extent (the largest
features captured by the observations). The concepts of grain and extent (Allen and Hoekstra,
1992) are particularly important because they set
the primary limits on our perception of the system.
We partially determine the grain and extent when
we choose a system to study, and further specify
grain and extent when we choose the protocols that
we use to study the system. Will we choose to study
a forest patch or watershed within a forest? How
often will we measure a dynamic process? How
will we analyze the data? These considerations
again must be tied directly to the question addressed by the assessment and the levels in the system appropriate to the question. Each choice we
make limits the information that we can expect to
N+l
FIGURE 2.1. The hierarchical approach to ecosystem
analysis suggests that the phenomenon of interest, level
N, is created and maintained by mechanisms that are subsystems, level N - 1 of the system (N). The level N +
1 is the context of the level of interest and provides
higher-order control of the system.
has important substructure. Clearly, a management
unit we choose to study may be composed of subunits of soil or vegetation type, while the unit itself
may be only part of a watershed.
Any ecosystem is a hierarchy of differently
scaled structures and functions. The structure and
function at a particular level (Figure 2.1) that we
choose to observe, say a forest stand, has a particular spatiotemporal scale. This is the N level. The
lower level (N - 1) is composed of subsystems, the
components that together comprise the forest stand.
Depending on the process under investigation, the
relevant components might be patches or gaps or
individual trees, all of which have smaller and
quicker spatiotemporal dynamics and provide a
mechanism for creation and maintenance of the
stand. The higher level (N + 1) is larger in space
than the forest stand, but changes more slowly. This
larger, slower level might be the region, which exerts control on the forest stand through topography
and mesoscale climate.
Hierarchy theory (O'Neill et al., 1986, 1989) applied to systems implies such upper-level control
and lower-level mechanisms and provides an epistemology for measurement based on the scale of
observation and level of analysis. Of course, the
level of interest (N) and the N - 1, N + 1 levels
are set by the problem being investigated and may
not be the same hierarchy for all problems. Depending on the criterion employed, many different
subsystems (N - 1) can be defined for any particular level (N). Observer perception determines the
A Theoretical Framework for Ecological Assessment
relevant levels (MacMahon et al., 1978). Characterizing the hierarchical organization of an ecosystem includes determining the spatial and temporal
bounds of each pattern or process and the order in
which patterns or processes are nested within the
hierarchy (Urban et al., 1989). At any scale, we can
observe many different types of things. What is included for analysis depends on what is important
for the study.
The description of the system as a hierarchy of
mechanisms (N - 1), phenomena (N), and controls
(N + 1) helps in sorting though the complexity by
isolating the dynamics of interest and structuring
the research around important interactions (Johnson, 1996). This is the art. Leave out everything
that is unimportant in sorting through the complexity. Find the correct criterion to address the
problem. The hard part is to determine the correct
level in the system, that which is truly important to
the observer. It may be a single species or a biotic
community or an ecosystem type. Choose the important level N. Then carefully, remaining within
the same hierarchy, identify the N + 1 level of control and the N - 1 level of mechanisms. These
should drive our investigations whether they are
field tests or modeling, applied or theoretical
(Turner, 1995).
2.2.2 Grain and Extent
Imagine casting a net into a sea populated with fish
of widely varying sizes. The net will capture only
a certain range of body sizes. Fish smaller than the
mesh size can slip through, and fish larger than the
net can hold will simply swim around the edge
(O'Neill et aI., 1986; Wiens, 1989; Ah1 and Allen,
1996). The net represents our observational protocol, the mesh size is analogous to grain (the smallest features that can be retrieved from the data), and
the size of the net determines the extent (the largest
features captured by the observations). The concepts of grain and extent (Allen and Hoekstra,
1992) are particularly important because they set
the primary limits on our perception of the system.
We partially determine the grain and extent when
we choose a system to study, and further specify
grain and extent when we choose the protocols that
we use to study the system. Will we choose to study
a forest patch or watershed within a forest? How
often will we measure a dynamic process? How
will we analyze the data? These considerations
again must be tied directly to the question addressed by the assessment and the levels in the system appropriate to the question. Each choice we
make limits the information that we can expect to