(moving, feeding, mating, avoiding predators, respiring, etc.), they interact
with their surroundings and with other agents. Such models allow study of
the relationship between individual actions and complex ecological systems
(DeAngelis and Gross 1992). The models can be linked to geographic information systems (GIS) and to models simulating other species. There are few
limits to the sophistication of these models. The state of individuals may
include location, sex, size, social status, and fat content, and the behavioral
rules may be related to environmental factors (e.g., temperature, water,
nutrient availability, and habitat condition), other agents, physiological
stress, environmental cues, or random actions. Model behavior can often be
compared directly with empirical observations.
As an example, an individual-based, landscape-scale model was constructed to simulate the interaction of dispersing American martens with
the spatial variability of energy (i.e., acquisition of prey) and mortality
risk (by predation) associated with different habitat types (Gardner and
Gustafson in press). Movement decision rules vary with the physiological
state of the individual, such that martens tend to select habitats that minimize predation risk, except when energy reserves are low, in which case they
select habitats that provide increased energy intake. Marten movements are
simulated on heterogeneous, grid-cell landscapes, and the movement paths,
percentage of dispersing martens killed or starved, and proportion of
martens successfully dispersing to a new home range are measured. The
agent-based approach is well suited to modeling the dispersal process
because it formalizes the behavior of an individual and allows the study
of how that behavior interacts with the landscape structure produced by
management, disturbance, and development.
8.3.1.3 New Approaches for Dealing with Scale
A number of studies in theoretical ecology point to the importance of scale
in ecological modeling (Kolasa 1989; Rahel 1990; Levin 1992; Holling 1992).
Levin (1992) argues that “the problem of pattern and scale is the central
problem in ecology.” Kolasa (1989), Rahel (1990), and Holling (1992)
acknowledge that spatial scale and temporal scale are paramount to understanding community dynamics.
Two scale considerations constrain realistic ecosystem simulation. First,
ecological systems are comprised of processes that occur across a wide
range of spatial and temporal scales. At one extreme lie small-scale, shorttime-period processes, such as the collision of molecules. At the other
extreme lie large-scale processes, such as global population dynamics (and
associated movement patterns), that may span thousands of kilometers
in space and decades in time. Studying one extreme or the other cannot
provide a comprehensive view of ecological systems. Second, ecological
systems have emergent properties that can only be described across multiple hierarchical levels (O’Neill et al. 1989). Hybrid modeling frameworks
have been developed to explicitly resolve mismatches of scale.
8. Evolving Approaches and Technologies
141
with their surroundings and with other agents. Such models allow study of
the relationship between individual actions and complex ecological systems
(DeAngelis and Gross 1992). The models can be linked to geographic information systems (GIS) and to models simulating other species. There are few
limits to the sophistication of these models. The state of individuals may
include location, sex, size, social status, and fat content, and the behavioral
rules may be related to environmental factors (e.g., temperature, water,
nutrient availability, and habitat condition), other agents, physiological
stress, environmental cues, or random actions. Model behavior can often be
compared directly with empirical observations.
As an example, an individual-based, landscape-scale model was constructed to simulate the interaction of dispersing American martens with
the spatial variability of energy (i.e., acquisition of prey) and mortality
risk (by predation) associated with different habitat types (Gardner and
Gustafson in press). Movement decision rules vary with the physiological
state of the individual, such that martens tend to select habitats that minimize predation risk, except when energy reserves are low, in which case they
select habitats that provide increased energy intake. Marten movements are
simulated on heterogeneous, grid-cell landscapes, and the movement paths,
percentage of dispersing martens killed or starved, and proportion of
martens successfully dispersing to a new home range are measured. The
agent-based approach is well suited to modeling the dispersal process
because it formalizes the behavior of an individual and allows the study
of how that behavior interacts with the landscape structure produced by
management, disturbance, and development.
8.3.1.3 New Approaches for Dealing with Scale
A number of studies in theoretical ecology point to the importance of scale
in ecological modeling (Kolasa 1989; Rahel 1990; Levin 1992; Holling 1992).
Levin (1992) argues that “the problem of pattern and scale is the central
problem in ecology.” Kolasa (1989), Rahel (1990), and Holling (1992)
acknowledge that spatial scale and temporal scale are paramount to understanding community dynamics.
Two scale considerations constrain realistic ecosystem simulation. First,
ecological systems are comprised of processes that occur across a wide
range of spatial and temporal scales. At one extreme lie small-scale, shorttime-period processes, such as the collision of molecules. At the other
extreme lie large-scale processes, such as global population dynamics (and
associated movement patterns), that may span thousands of kilometers
in space and decades in time. Studying one extreme or the other cannot
provide a comprehensive view of ecological systems. Second, ecological
systems have emergent properties that can only be described across multiple hierarchical levels (O’Neill et al. 1989). Hybrid modeling frameworks
have been developed to explicitly resolve mismatches of scale.
8. Evolving Approaches and Technologies
141