"Any study
which throws light upon the nature of 'order' or
'pattern' in the universe is surely nontrivial"
-- Gregory Bateson in 'Steps to an Ecology of Mind'
Virtual models enable the user to interact with the
ecosystem under examination and to experiment by
interactively changing ecological parameters such as
habitat and weather. Because the subject of ecological
modeling is a highly complex ecosystem composed of many
interrelating groups and processes, a model can cover
many scales of resolution (levels of detail).
The model must consider the system or subsystem under
study in the context of its own spatial and temporal
resolution as well as its relationship to other
subsystems. A key feature of ecological modeling is that
every entity is treated as a part of a complex system.
And, every entity is described as being a conglomerate
of smaller components. By viewing the ecosystem as a
complex system, there is a framework to both study
entities at different levels and examine their
relationships to other components.
This seemingly simple way in how a model is viewed has
had a large influence on the way that models have helped
explain ecological phenomena. Colonization, flocking,
and population distributions are three areas that have
benefited from ecosystem modeling.
The main advantage of virtual ecosystem models is that
they provide useful visual illustrations of the general
nature of ecosystem dynamics. They show mechanisms that
give rise to unexpected events. The limitations of
ecosystem models are that they cannot be used to
precisely forecast events and they are difficult to
validate.
The types and characteristics of ecosystem models are:
Conventional Models (not necessarily virtual)
Characteristics of the entire population are
averaged together.
Model attempts to simulate changes in these
averaged characteristics for the whole population.
These models are usually not visual - employing
statistics or differential equations instead.
Individual Models
Discrete objects are modeled using local rules.
Flocking/schooling models are examples.
The model focuses on specific individuals
distributed in the space.
The geographic position of an individual is the
primary visualization.
They might occupy only a few grid cells and more
than one distinct type of individual might live in
the same grid.
They portray the global dynamics resulting from
local interactions of members of a population.
Agent Models
The objects have the ability to learn about their
environment and modify their behavior accordingly.
The objects adapt, learn, and evolve.
Usually some genetic algorithm is used.
A widespread conclusion from agent based model
simulations is that an organism's environment has a
substantial influence on its behavior and
subsequently on the overall dynamics of the
population from which the organism is a part.
Cellular automata (CA) models
A checkerboard of square cells whose states are
updated in discrete time steps. A simple grid
usually represents the spatial domain. Objects and
their states are represented by colored squares in
the grid.
The state of each cell is defined by a set of
deterministic rules that define a cell's state based
on the states of neighboring cells.
A CA simulation will always end up in one of four
configurations:
Spatially homogenous state (point attractor or
no active cells).
Sequence of simple stable or periodic
structures (pertiodic attractor).
Chaotic aperiodic dynamics (strange
attractor).
Complicated localized structures - some
propagating (the edge of chaos)
Used to model spatial phenomena such as seed
dispersal, animal migration, and forest growth.
Combined models
Represents an entire ecosystem and can combine
model types-- for example.
A CA model for the terrain and environment.
Individual or agent model to depict biological
components.
