44
D. Morrall
must be sustained in the face of a continuously changing environment. Natural
change is accelerated through human alteration of habitats (e.g., channelization of
streams), addition of pesticides and fertilizers, and the introduction of non-native
species. For organisms to survive they must be able to adapt to both natural and
human-induced change. Ecological models must also incorporate evolution and
co-evolution into their frameworks if they are to predict the sustainability of
various ecologies (e.g., Janssen 1998). Evolution can be incorporated into
ecological models through adaptation of species currently in the system and by
forecasting changes in species composition (e.g. Maier et al. 1998). Co-evolution
in ecological systems is described by the Gaia theory (Lovelock and Margulis
1974; Downing and Zvirnsky 1999) and which incorporates the feedback
mechanisms common in natural systems. Gaia refers to the circular pathway
whereby organisms respond to their environment, modify the environment, and
are, in-turn, modified by the environment. Representation of this phenomenon is
critical for predicting the future of ecological systems.
As Jorgensen (1999) noted, most traditional models are limited because they
use a static representation of ecological systems that is developed and
parameterized based on the system characteristics at a certain point in time. They
do not take into account the flexibility and adaptive capabilities of natural systems
and therefore may over-predict or erroneously predict the effect of stressors. A
combination of genetic algorithms with traditional engineering based models and
other artificial intelligence techniques (e.g., cellular automata and neural
networks) can provide a dynamic representation of how adaptive responses to
environmental change govern species change. These dynamic approach es
facilitate exploration of various possible trajectories of adaptation that might result
from changes in the environment.
Because of the broad scale at wh ich the environment is being changed, tools are
needed that can accurately predict the sustainability of populations, communities,
and ecosystems. This need will become increasingly important in the future.
Dynamic simulation of organism adaptation and interdependencies marks the
beginning of a newage in ecological modelling and offers the possibility for
development of superior predictive models. GAs based on the fundamentals of
theoretical ecology can help us find the mathematical foundations for the concepts
on which predictive ecology is based.
3.6
The Next Generation: Hybrid Genetic Aigorithms
The examples noted above point out the capabilities of GAs to optimize
parameters, disco ver equations, search for patterns, and develop classifier and
control strategies. Yet GAs are not the best technique for all problems. Neural
nets, for example, are often superior at pattern recognition. Many readily
available statistical techniques perform as weil as or better than GAs at developing
regression and classification systems. These techniques, however, do not have the
D. Morrall
must be sustained in the face of a continuously changing environment. Natural
change is accelerated through human alteration of habitats (e.g., channelization of
streams), addition of pesticides and fertilizers, and the introduction of non-native
species. For organisms to survive they must be able to adapt to both natural and
human-induced change. Ecological models must also incorporate evolution and
co-evolution into their frameworks if they are to predict the sustainability of
various ecologies (e.g., Janssen 1998). Evolution can be incorporated into
ecological models through adaptation of species currently in the system and by
forecasting changes in species composition (e.g. Maier et al. 1998). Co-evolution
in ecological systems is described by the Gaia theory (Lovelock and Margulis
1974; Downing and Zvirnsky 1999) and which incorporates the feedback
mechanisms common in natural systems. Gaia refers to the circular pathway
whereby organisms respond to their environment, modify the environment, and
are, in-turn, modified by the environment. Representation of this phenomenon is
critical for predicting the future of ecological systems.
As Jorgensen (1999) noted, most traditional models are limited because they
use a static representation of ecological systems that is developed and
parameterized based on the system characteristics at a certain point in time. They
do not take into account the flexibility and adaptive capabilities of natural systems
and therefore may over-predict or erroneously predict the effect of stressors. A
combination of genetic algorithms with traditional engineering based models and
other artificial intelligence techniques (e.g., cellular automata and neural
networks) can provide a dynamic representation of how adaptive responses to
environmental change govern species change. These dynamic approach es
facilitate exploration of various possible trajectories of adaptation that might result
from changes in the environment.
Because of the broad scale at wh ich the environment is being changed, tools are
needed that can accurately predict the sustainability of populations, communities,
and ecosystems. This need will become increasingly important in the future.
Dynamic simulation of organism adaptation and interdependencies marks the
beginning of a newage in ecological modelling and offers the possibility for
development of superior predictive models. GAs based on the fundamentals of
theoretical ecology can help us find the mathematical foundations for the concepts
on which predictive ecology is based.
3.6
The Next Generation: Hybrid Genetic Aigorithms
The examples noted above point out the capabilities of GAs to optimize
parameters, disco ver equations, search for patterns, and develop classifier and
control strategies. Yet GAs are not the best technique for all problems. Neural
nets, for example, are often superior at pattern recognition. Many readily
available statistical techniques perform as weil as or better than GAs at developing
regression and classification systems. These techniques, however, do not have the