interacting subcomponents described by a dynamical system (typically ordinary differential equations) and to connect these compartments by flows of
material among them (e.g., biomass and nutrients). However, this method
forces the modeler to use only one mathematical approach to structure the
system. New methods are developing to allow linkages among system
components that take into account differing levels of detail to describe the
interactions between them. Advocates of this multimodeling methodology
argue that the use of a single modeling approach is inappropriate for problems spanning a wide variety of temporal, spatial, and organismal scales.
Multimodeling does not refer to multiple models representing the same
components of a system to determine the importance of additional detail.
Rather, it refers to using different modeling approaches for different
components of the system and linking these different models to study the
interactions among the components.
One example of such a multimodel is the ATLSS (Across Trophic Level
System Simulation) project, constructed to aid analysis of the ecological
impacts of planning for the hydrologic restoration of the Everglades of
South Florida (DeAngelis et al. 1998). The ATLSS uses a mixture of
approaches based upon the inherent temporal and spatial resolution
and extent of various trophic components, linked together by spatially
explicit information on the underlying environmental (e.g., water and
soil-structure), biotic (e.g., vegetation), and anthropogenic (e.g., land-use)
factors. The linked components include spatially explicit indices (Curnutt et
al. 2000), compartment models, differential equations for structured populations and communities (Gaff et al. 2000), and individual-based models
(DeAngelis et al. 2000). Linking models that operate at very different spatial
and temporal extents is a major challenge, requiring a variety of spatial
interpolation methods (Luh et al. 1997) and careful design of model interfaces (Duke-Sylvester and Gross 2002). The multimodeling approach can
readily be expanded to include economic, land-use, and human-population
impacts, although this will require careful error-propagation analysis.
8.6 Lessons Learned from Earlier
Modeling Approaches
The application of ecological models by managers has sometimes fallen
short of expectations. An analysis of two examples may be instructive for
ecological modelers in general.
8.6.1 FORPLAN
The FORPLAN linear-programming (LP) model was the primary analytical tool used by the U.S. Department of Agriculture (USDA) Forest Service
for natural resource analysis and forest planning in the 1980s (Iverson
and Alston 1986). However, FORPLAN fell from favor by the mid-1990s
158
Eric Gustafson et al.
material among them (e.g., biomass and nutrients). However, this method
forces the modeler to use only one mathematical approach to structure the
system. New methods are developing to allow linkages among system
components that take into account differing levels of detail to describe the
interactions between them. Advocates of this multimodeling methodology
argue that the use of a single modeling approach is inappropriate for problems spanning a wide variety of temporal, spatial, and organismal scales.
Multimodeling does not refer to multiple models representing the same
components of a system to determine the importance of additional detail.
Rather, it refers to using different modeling approaches for different
components of the system and linking these different models to study the
interactions among the components.
One example of such a multimodel is the ATLSS (Across Trophic Level
System Simulation) project, constructed to aid analysis of the ecological
impacts of planning for the hydrologic restoration of the Everglades of
South Florida (DeAngelis et al. 1998). The ATLSS uses a mixture of
approaches based upon the inherent temporal and spatial resolution
and extent of various trophic components, linked together by spatially
explicit information on the underlying environmental (e.g., water and
soil-structure), biotic (e.g., vegetation), and anthropogenic (e.g., land-use)
factors. The linked components include spatially explicit indices (Curnutt et
al. 2000), compartment models, differential equations for structured populations and communities (Gaff et al. 2000), and individual-based models
(DeAngelis et al. 2000). Linking models that operate at very different spatial
and temporal extents is a major challenge, requiring a variety of spatial
interpolation methods (Luh et al. 1997) and careful design of model interfaces (Duke-Sylvester and Gross 2002). The multimodeling approach can
readily be expanded to include economic, land-use, and human-population
impacts, although this will require careful error-propagation analysis.
8.6 Lessons Learned from Earlier
Modeling Approaches
The application of ecological models by managers has sometimes fallen
short of expectations. An analysis of two examples may be instructive for
ecological modelers in general.
8.6.1 FORPLAN
The FORPLAN linear-programming (LP) model was the primary analytical tool used by the U.S. Department of Agriculture (USDA) Forest Service
for natural resource analysis and forest planning in the 1980s (Iverson
and Alston 1986). However, FORPLAN fell from favor by the mid-1990s
158
Eric Gustafson et al.