284
ECOSIM, a multiresource forest growth-and-yield
model (1.1. Rogers et aI., 1984). Model output provided an estimate of changes with time in tree density, fuel loading, herbage production, streamflow,
and scenic beauty index under postsettlement conditions.
2. Baker (1989) implemented transition matrix
simulation models to determine whether a stable mosaic of forest patches existed during the period
1727-1868 in the Boundary Waters Canoe Area,
Minnesota, based on fire history data reconstructed
by Heinselman (1973). No stable patch mosaic was
found at any of the spatial scales examined (units
25,000- to 400,000-ha in size). Such a result implies
that even a very large conservation reserve would
not experience steady-state environments, leading to
potential conflict between the conservation goals of
perpetuating fluctuations in landscape structure versus maintaining populations of species that depend
on the presence of particular landscape structures.
3. CRBSUM is a vegetation succession model
that was developed for use in the interior Columbia River basin ecological assessment (Keane et al.,
1995). The model simulates changes in vegetation
structural stages and cover types over time in response to stochastic disturbance. Coarse-scale vegetation changes were predicted for a set of management strategies as part of the assessment.
4. Individual-based plant models (gap models;
Shugart, 1998) contain features that make them
suitable for reconstructing the range of variability
in species composition and structure under presettlement conditions, as well as predicting consequences of postsettlement activities. These features
include incorporation of stochastically implemented disturbance processes (such as fire, flooding, and storms), climate and other environmental
effects, and simulations over long time periods (Urban and Shugart, 1992). Gap models have been
tested for their ability to reproduce reconstructed
vegetation from pollen analysis over the postglacial
period in the eastern United States, producing results that are consistent with long-term variations
in forests, including temporal sequences and spatial patterns (Solomon and Shugart, 1984; Solomon
and Webb, 1985; Shugart, 1998).
5. LANDIS is a simulation model that is related
to gap models, but it is able to simulate larger areas than gap models by means of aggregated tree
age-class distributions and longer time steps, which
reduce processing constraints (Mladenoff et aI.,
1996; He and Mladenoff, 1999). The model is suitable for representing spatially explicit forest landscape dynamics in response to fire and wind throw
disturbance.
Methods for Determining Historical Range of Variability
19.10 Characterization of
Biophysical Environments
In the absence of direct sources of information about
the range of pattern or process variability or as a supplement to such sources (e.g., to extrapolate from
sampled to unsampled areas), aspects of the biophysical environments of landscapes can be characterized to describe their expected biotic components
and associated disturbance regimes (see Chapters 3
and 22). Biophysical attributes (e.g., topographic,
climatic, hydrologic, or geologic variables) are selected based on their hypothesized control of biotic
distributions and low temporal variability at a given
scale of interest (Bourgeron et aI., submitted). The
resulting biophysical environments are land units in
which each biophysical environment type is expected to contain a particular suite of biotic and disturbance responses (Bailey et aI., 1994).
Testing of the use of biophysical environments
for extrapolating HRV between spatial or temporal
points has been conducted informally for some
time. Stratification by categorical variables (e.g.,
geologic substrate) or biophysical environments
has been performed for a variety of analyses, such
as construction of species-environment models
(e.g., Austin et al. 1990); development of fire models, both conceptual (e.g., Fischer and Clayton,
1983; Fischer and Bradley, 1987) and quantitative
(e.g., Kessell, 1976; Kessell and Fischer, 1981);
and characterization of forest insect and disease
regimes (e.g., Harvey, 1994; Hessburg et aI., 1994;
Lehmkuhl et ai., 1994; Filip et ai., 1996). Formal
assessment of biophysical environment characterization as a guide to stratifying sampling effort and
extrapolating results is discussed in Chapter 7.
EXAMPLES
In areas with topographic complexity, changes to
the disturbance regime may be strongly correlated
with topography, enabling disturbance characteristics to be mapped based on topographic features.
1. Romme and Knight (1981) developed a
graphic model of the relationships among topographic position, fire-free interval, and development of mature spruce-fir forests. The model accounts for vegetation composition differences in
upland areas compared to sheltered ravines and valley bottoms in the subalpine zone of the Medicine
Bow Mountains, Wyoming.
2. Camp et ai. (1997) produced statistical models to predict historical fire refugia (forest patches
minimally affected by recurrent fires) based on
topographic variables in the Wenatchee Mountains,
ECOSIM, a multiresource forest growth-and-yield
model (1.1. Rogers et aI., 1984). Model output provided an estimate of changes with time in tree density, fuel loading, herbage production, streamflow,
and scenic beauty index under postsettlement conditions.
2. Baker (1989) implemented transition matrix
simulation models to determine whether a stable mosaic of forest patches existed during the period
1727-1868 in the Boundary Waters Canoe Area,
Minnesota, based on fire history data reconstructed
by Heinselman (1973). No stable patch mosaic was
found at any of the spatial scales examined (units
25,000- to 400,000-ha in size). Such a result implies
that even a very large conservation reserve would
not experience steady-state environments, leading to
potential conflict between the conservation goals of
perpetuating fluctuations in landscape structure versus maintaining populations of species that depend
on the presence of particular landscape structures.
3. CRBSUM is a vegetation succession model
that was developed for use in the interior Columbia River basin ecological assessment (Keane et al.,
1995). The model simulates changes in vegetation
structural stages and cover types over time in response to stochastic disturbance. Coarse-scale vegetation changes were predicted for a set of management strategies as part of the assessment.
4. Individual-based plant models (gap models;
Shugart, 1998) contain features that make them
suitable for reconstructing the range of variability
in species composition and structure under presettlement conditions, as well as predicting consequences of postsettlement activities. These features
include incorporation of stochastically implemented disturbance processes (such as fire, flooding, and storms), climate and other environmental
effects, and simulations over long time periods (Urban and Shugart, 1992). Gap models have been
tested for their ability to reproduce reconstructed
vegetation from pollen analysis over the postglacial
period in the eastern United States, producing results that are consistent with long-term variations
in forests, including temporal sequences and spatial patterns (Solomon and Shugart, 1984; Solomon
and Webb, 1985; Shugart, 1998).
5. LANDIS is a simulation model that is related
to gap models, but it is able to simulate larger areas than gap models by means of aggregated tree
age-class distributions and longer time steps, which
reduce processing constraints (Mladenoff et aI.,
1996; He and Mladenoff, 1999). The model is suitable for representing spatially explicit forest landscape dynamics in response to fire and wind throw
disturbance.
Methods for Determining Historical Range of Variability
19.10 Characterization of
Biophysical Environments
In the absence of direct sources of information about
the range of pattern or process variability or as a supplement to such sources (e.g., to extrapolate from
sampled to unsampled areas), aspects of the biophysical environments of landscapes can be characterized to describe their expected biotic components
and associated disturbance regimes (see Chapters 3
and 22). Biophysical attributes (e.g., topographic,
climatic, hydrologic, or geologic variables) are selected based on their hypothesized control of biotic
distributions and low temporal variability at a given
scale of interest (Bourgeron et aI., submitted). The
resulting biophysical environments are land units in
which each biophysical environment type is expected to contain a particular suite of biotic and disturbance responses (Bailey et aI., 1994).
Testing of the use of biophysical environments
for extrapolating HRV between spatial or temporal
points has been conducted informally for some
time. Stratification by categorical variables (e.g.,
geologic substrate) or biophysical environments
has been performed for a variety of analyses, such
as construction of species-environment models
(e.g., Austin et al. 1990); development of fire models, both conceptual (e.g., Fischer and Clayton,
1983; Fischer and Bradley, 1987) and quantitative
(e.g., Kessell, 1976; Kessell and Fischer, 1981);
and characterization of forest insect and disease
regimes (e.g., Harvey, 1994; Hessburg et aI., 1994;
Lehmkuhl et ai., 1994; Filip et ai., 1996). Formal
assessment of biophysical environment characterization as a guide to stratifying sampling effort and
extrapolating results is discussed in Chapter 7.
EXAMPLES
In areas with topographic complexity, changes to
the disturbance regime may be strongly correlated
with topography, enabling disturbance characteristics to be mapped based on topographic features.
1. Romme and Knight (1981) developed a
graphic model of the relationships among topographic position, fire-free interval, and development of mature spruce-fir forests. The model accounts for vegetation composition differences in
upland areas compared to sheltered ravines and valley bottoms in the subalpine zone of the Medicine
Bow Mountains, Wyoming.
2. Camp et ai. (1997) produced statistical models to predict historical fire refugia (forest patches
minimally affected by recurrent fires) based on
topographic variables in the Wenatchee Mountains,