19.9 Simulation Modeling
tat, and insect and disease hazard were described
by comparing maps derived from the aerial photographs. Mapped vegetation patches were also assigned potential fire behavior attributes to evaluate
the effects of vegetation change on fire behavior
and smoke production (Huff et al., 1995).
9. Bahre (1991) compared Soil Conservation
Service aerial photographs taken between 1935 and
1937 with National High Altitude Photography
photographs from 1983 and 1984 in conjunction
with other sources of information to evaluate
changes in vegetation distribution in southeastern
Arizona. He concluded that anthropogenic impacts
have played a substantial role in these changes.
10. Cole and Taylor (1995) examined aerial
photographs covering the period 1959 to 1987 to
document encroachment of forest into formerly
open prairie in Indiana.
11. Miller et al. (1995) examined changes in a riparian landscape in southeastern Wyoming using
aerial photographs taken in 1973 and 1990. Quantification of changes included a 75% decline in wetted area and indicated a shift from young dense cottonwood stands to older, more open stands following
changes in the frequency and intensity of flooding
in the North Platte River.
12. Allen and Breshears (1998) analyzed a sequence of aerial photographs taken between 1935
and 1975 to quantify an extremely rapid shift in an
ecotone between ponderosa pine forest and
pinyon-juniper woodland in northern New Mexico. The ecotonal shift, occurring over 2 km or
more in less than five years, resulted from mortality of ponderosa pine in response to a severe
drought in the 1950s. Other factors, such as fire
suppression, increased pinyon-juniper density, and
drought-triggered bark beetle infestation, amplified
the effect of the drought on ponderosa pine. The
shift has persisted for more than 40 years, resulting in greater fragmentation of forest patches and
changes in ecosystem properties due to soil erosion
(Allen and Breshears, 1998).
19.8 Maps from Historical Data
Some sources of historical data on vegetation pattern are either provided in the form of maps or contain sufficient detail that maps can be prepared from
them.
EXAMPLES
1. Goldberg and Turner (1986) studied a set of
permanent vegetation plots in the Sonoran Desert
near Tucson, Arizona, in which locations of indi283
vidual woody and succulent plants had been
mapped as early as 1906 for some plots and subsequently remapped at irregular intervals over the
next 72 years. Time series analyses revealed no
consistent directional changes in vegetation composition during this period.
2. White and Mladenoff (1994) examined forest
landscape transitions in north-central Wisconsin
across three points in time. A pre settlement forest
cover map (1860) was constructed from General
Land Office surveys using associations of tree
species with major forest types to contour witness
tree samples by forest cover type. The presettlement map was compared with forest cover maps
produced in 1930 (based on inventory data and aerial photography) and 1989 (based on color infrared
photography).
3. Foster (1992) constructed a vegetation map
for a county in central Massachusetts based on a
historical description of the landscape and vegetation within each township in the county in 1793.
This map of forest vegetation prior to extensive impact by settlement was compared with data and
maps describing postsettlement forest clearing followed by subsequent reforestation.
4. Chou et al. (1993) used digitized maps of major fires occurring between 1911 and 1984 in the
San Jacinto Mountains, California, as a basis for
constructing statistical models of probability of fire
occurrence.
19.9 Simulation Modeling
Simulation models provide information about ranges
of variability in patterns and processes over time.
They can be used to explore the implications and
limitations of reconstructed presettlement biotic distributions and disturbance regimes derived from the
other methods described in this chapter. Chapter 18
describes a number of simulation models, including
some of the models discussed here (transition matrix models, CRBSUM, individual-based models,
and LANDIS). When selecting a simulation model,
consideration should be given to operation at appropriate spatial and temporal scales, parameter and
input data needs, and robustness of output as assessed by verification and validation procedures.
EXAMPLES
1. Simulation modeling was incorporated with
dendroecological reconstruction of fire regimes and
forest structure in an Arizona ponderosa pine forest (Covington and Moore, 1994a; 1994b). Plot-derived data on pre settlement trees was entered into
tat, and insect and disease hazard were described
by comparing maps derived from the aerial photographs. Mapped vegetation patches were also assigned potential fire behavior attributes to evaluate
the effects of vegetation change on fire behavior
and smoke production (Huff et al., 1995).
9. Bahre (1991) compared Soil Conservation
Service aerial photographs taken between 1935 and
1937 with National High Altitude Photography
photographs from 1983 and 1984 in conjunction
with other sources of information to evaluate
changes in vegetation distribution in southeastern
Arizona. He concluded that anthropogenic impacts
have played a substantial role in these changes.
10. Cole and Taylor (1995) examined aerial
photographs covering the period 1959 to 1987 to
document encroachment of forest into formerly
open prairie in Indiana.
11. Miller et al. (1995) examined changes in a riparian landscape in southeastern Wyoming using
aerial photographs taken in 1973 and 1990. Quantification of changes included a 75% decline in wetted area and indicated a shift from young dense cottonwood stands to older, more open stands following
changes in the frequency and intensity of flooding
in the North Platte River.
12. Allen and Breshears (1998) analyzed a sequence of aerial photographs taken between 1935
and 1975 to quantify an extremely rapid shift in an
ecotone between ponderosa pine forest and
pinyon-juniper woodland in northern New Mexico. The ecotonal shift, occurring over 2 km or
more in less than five years, resulted from mortality of ponderosa pine in response to a severe
drought in the 1950s. Other factors, such as fire
suppression, increased pinyon-juniper density, and
drought-triggered bark beetle infestation, amplified
the effect of the drought on ponderosa pine. The
shift has persisted for more than 40 years, resulting in greater fragmentation of forest patches and
changes in ecosystem properties due to soil erosion
(Allen and Breshears, 1998).
19.8 Maps from Historical Data
Some sources of historical data on vegetation pattern are either provided in the form of maps or contain sufficient detail that maps can be prepared from
them.
EXAMPLES
1. Goldberg and Turner (1986) studied a set of
permanent vegetation plots in the Sonoran Desert
near Tucson, Arizona, in which locations of indi283
vidual woody and succulent plants had been
mapped as early as 1906 for some plots and subsequently remapped at irregular intervals over the
next 72 years. Time series analyses revealed no
consistent directional changes in vegetation composition during this period.
2. White and Mladenoff (1994) examined forest
landscape transitions in north-central Wisconsin
across three points in time. A pre settlement forest
cover map (1860) was constructed from General
Land Office surveys using associations of tree
species with major forest types to contour witness
tree samples by forest cover type. The presettlement map was compared with forest cover maps
produced in 1930 (based on inventory data and aerial photography) and 1989 (based on color infrared
photography).
3. Foster (1992) constructed a vegetation map
for a county in central Massachusetts based on a
historical description of the landscape and vegetation within each township in the county in 1793.
This map of forest vegetation prior to extensive impact by settlement was compared with data and
maps describing postsettlement forest clearing followed by subsequent reforestation.
4. Chou et al. (1993) used digitized maps of major fires occurring between 1911 and 1984 in the
San Jacinto Mountains, California, as a basis for
constructing statistical models of probability of fire
occurrence.
19.9 Simulation Modeling
Simulation models provide information about ranges
of variability in patterns and processes over time.
They can be used to explore the implications and
limitations of reconstructed presettlement biotic distributions and disturbance regimes derived from the
other methods described in this chapter. Chapter 18
describes a number of simulation models, including
some of the models discussed here (transition matrix models, CRBSUM, individual-based models,
and LANDIS). When selecting a simulation model,
consideration should be given to operation at appropriate spatial and temporal scales, parameter and
input data needs, and robustness of output as assessed by verification and validation procedures.
EXAMPLES
1. Simulation modeling was incorporated with
dendroecological reconstruction of fire regimes and
forest structure in an Arizona ponderosa pine forest (Covington and Moore, 1994a; 1994b). Plot-derived data on pre settlement trees was entered into