18.2 Models of Ecosystem Structure
est structure and composition, bird species densities, and the economic value of wood products in
the western Cascades, Oregon (Hansen et aI.,
1995).
18.2.3 Individual-based Animal Models
Individual-based animal simulation models that incorporate habitat complexity in a spatially explicit
framework can be used to evaluate species responses to changes in heterogeneous landscapes,
including changes in management strategies (Dunning et aI., 1995). Such models require habitatspecific information about demography, dispersal
behavior, and habitat selection for the animal
species modeled, as well as a map describing the
spatial arrangement of habitat patches in the landscape. Individuals are tracked by location and acquire fitness characteristics based on the landscape
cells and hence the habitat they occupy. The models can explore the consequences for animal populations of habitat fragmentation, isolation, shape,
and patch size resulting from various management
alternatives, including species recovery plans
(Dunning et aI., 1995, Turner et aI., 1995). Questions can be addressed regarding animal responses
to resource dynamics, effect of landscape composition on extinction probability, and efficiency of
reserve design (Dunning et aI., 1995).
The models assume that habitat can be adequately described by a set of mapped categories
such as vegetation types or suitable-unsuitable
habitat. Detailed demographic and behavioral data
derived from field studies are required to parameterize the models (Dunning et aI., 1995). Lack of
such data for an area or at a particular scale may
limit implementation of the models (Turner et aI.,
1995). In addition, highly accurate quantitative
predictions of population responses to specific environmental changes are generally not possible at
present (Dunning et aI., 1995). Most spatially explicit, individual-based animal models consider
only one or a few animal species at a time, which
may further limit their utility for ecological assessments.
Plant simulation models can be linked to animal
models to provide a more detailed characterization
of the vegetation structure and composition comprising habitat than is generally included in most
animal models (Holt et aI., 1995). Individual-based
plant models such as those considered previously
can be utilized to predict landscape dynamics over
long time periods and under conditions of climate
change. For example, output from a spatially explicit plant model could be used to generate tran261
sition rules for changes in landscape cells from one
habitat type to another (Holt et al., 1995).
The model BACHMAP simulates Bachman's
sparrow populations in pine forests of the southeast United States (Pulliam et aI., 1992). Individuals are tracked in grid cells scaled to the size of
sparrow territories. Each cell is classified into one
of 22 vegetation types. Sparrow birth, death, and
movement rates are functions of vegetation type,
which may change annually in response to agerelated management practices. Another Bachman's
sparrow model, ECOLECON, incorporates the economics of timber production so that managers can
balance sparrow population size with timber outputs (Liu, 1992, 1993; Turner et aI., 1995).
MOSAIC is an individual-based model designed
to project owl numbers and distribution across large
landscapes (McKelvey et aI., 1992; Bart, 1995).
Habitat in grid cells is classified as either suitable
or unsuitable. Each cell is scored according to the
amount of suitable habitat, number of owls using
the cell, and the amount of habitat in surrounding
cells. Territories and home ranges have benefits to
owls based on the sum of cell scores and costs
based on the size of the territory or home range. A
spatially explicit routine is invoked to allow dispersing owls to search for suitable territories.
NOYELP, a model used to explore mortality of
wintering elk and bison in response to the scale and
pattern of fires in northern Yellowstone National
Park, includes simulation offorage availability, snow
conditions, ungulate movement and foraging, and ungulate energetics (Turner et al., 1993, 1994, 1995).
The model is initialized with forage distributed
among six vegetation types according to their actual
spatial distribution. Resource levels in occupied
patches are depleted by foraging animals such that
the distribution of resources changes with time. The
model provides managers with a tool to evaluate how
species are affected by management activities.
18.2.4 Gradient Models
Gradient models, developed by Kessell and colleagues, predict the effects of fire on vegetation and
fuels based on biotic--environmental gradient relationships (Kessell, 1976, 1977, 1979, 1990; Kessell
and Cattelino, 1978; Kessell and Good, 1982;
Kessell et aI., 1984). The models are implemented
on a grid of cells, with cell sizes of 0.01 to 1 ha.
For each cell, values of environmental attributes
such as elevation, topography, moisture, time since
last bum, primary succession, drainage system, and
nonfire disturbance categories determine species
composition and fuel levels using gradient sub-
est structure and composition, bird species densities, and the economic value of wood products in
the western Cascades, Oregon (Hansen et aI.,
1995).
18.2.3 Individual-based Animal Models
Individual-based animal simulation models that incorporate habitat complexity in a spatially explicit
framework can be used to evaluate species responses to changes in heterogeneous landscapes,
including changes in management strategies (Dunning et aI., 1995). Such models require habitatspecific information about demography, dispersal
behavior, and habitat selection for the animal
species modeled, as well as a map describing the
spatial arrangement of habitat patches in the landscape. Individuals are tracked by location and acquire fitness characteristics based on the landscape
cells and hence the habitat they occupy. The models can explore the consequences for animal populations of habitat fragmentation, isolation, shape,
and patch size resulting from various management
alternatives, including species recovery plans
(Dunning et aI., 1995, Turner et aI., 1995). Questions can be addressed regarding animal responses
to resource dynamics, effect of landscape composition on extinction probability, and efficiency of
reserve design (Dunning et aI., 1995).
The models assume that habitat can be adequately described by a set of mapped categories
such as vegetation types or suitable-unsuitable
habitat. Detailed demographic and behavioral data
derived from field studies are required to parameterize the models (Dunning et aI., 1995). Lack of
such data for an area or at a particular scale may
limit implementation of the models (Turner et aI.,
1995). In addition, highly accurate quantitative
predictions of population responses to specific environmental changes are generally not possible at
present (Dunning et aI., 1995). Most spatially explicit, individual-based animal models consider
only one or a few animal species at a time, which
may further limit their utility for ecological assessments.
Plant simulation models can be linked to animal
models to provide a more detailed characterization
of the vegetation structure and composition comprising habitat than is generally included in most
animal models (Holt et aI., 1995). Individual-based
plant models such as those considered previously
can be utilized to predict landscape dynamics over
long time periods and under conditions of climate
change. For example, output from a spatially explicit plant model could be used to generate tran261
sition rules for changes in landscape cells from one
habitat type to another (Holt et al., 1995).
The model BACHMAP simulates Bachman's
sparrow populations in pine forests of the southeast United States (Pulliam et aI., 1992). Individuals are tracked in grid cells scaled to the size of
sparrow territories. Each cell is classified into one
of 22 vegetation types. Sparrow birth, death, and
movement rates are functions of vegetation type,
which may change annually in response to agerelated management practices. Another Bachman's
sparrow model, ECOLECON, incorporates the economics of timber production so that managers can
balance sparrow population size with timber outputs (Liu, 1992, 1993; Turner et aI., 1995).
MOSAIC is an individual-based model designed
to project owl numbers and distribution across large
landscapes (McKelvey et aI., 1992; Bart, 1995).
Habitat in grid cells is classified as either suitable
or unsuitable. Each cell is scored according to the
amount of suitable habitat, number of owls using
the cell, and the amount of habitat in surrounding
cells. Territories and home ranges have benefits to
owls based on the sum of cell scores and costs
based on the size of the territory or home range. A
spatially explicit routine is invoked to allow dispersing owls to search for suitable territories.
NOYELP, a model used to explore mortality of
wintering elk and bison in response to the scale and
pattern of fires in northern Yellowstone National
Park, includes simulation offorage availability, snow
conditions, ungulate movement and foraging, and ungulate energetics (Turner et al., 1993, 1994, 1995).
The model is initialized with forage distributed
among six vegetation types according to their actual
spatial distribution. Resource levels in occupied
patches are depleted by foraging animals such that
the distribution of resources changes with time. The
model provides managers with a tool to evaluate how
species are affected by management activities.
18.2.4 Gradient Models
Gradient models, developed by Kessell and colleagues, predict the effects of fire on vegetation and
fuels based on biotic--environmental gradient relationships (Kessell, 1976, 1977, 1979, 1990; Kessell
and Cattelino, 1978; Kessell and Good, 1982;
Kessell et aI., 1984). The models are implemented
on a grid of cells, with cell sizes of 0.01 to 1 ha.
For each cell, values of environmental attributes
such as elevation, topography, moisture, time since
last bum, primary succession, drainage system, and
nonfire disturbance categories determine species
composition and fuel levels using gradient sub-