86
F. Recknagel
nutrient loadings. Each single agent is determined by EDE in order to maximise
(adapt) their performance (abundance) in relation to current environmental
conditions (nutrient loadings, light, temperature and abundance of competitors,
predators or preys). EDE utilise evolutionary algorithms in order to steadily
optimise parameter values and functions of the state variable-based agents by
means of differential equations as used by Park et al. (1974) and Recknagel and
Benndorf (1982). As a result, each agent adapts simultaneously to current
environmental conditions by producing "offspring" agents based on its evaluation
and selection of mates, recombination strategy and mutation strategy (see Figure
5.6b). Successful case studies on EDE have been conducted by Whigham and
Recknagel (2001a, b) and Recknagel et al. (2002).
5.5
Conclusions
1. Adaptive agents (AA) provide a realistic framework for ecosystem simulation,
evolving ecosystem structures and behaviours by emerging, submerging,
interacting and evolving ecological entities.
2. Individual-based AA prove applicable to a spatiaUy explicit simulation of
highly simplified terrestrial food webs.
3. State variable-based AA where evolutionary computation is embodied appear
to be relevant for simulations of aquatic food webs dynamics and plankton
species interactions.
4. Embodiment of evolutionary computation in adaptive agents for aquatic species
or functional groups can be achieved by evolving predictive rules (ER),
differential equations (EDE) or artificial neural networks (ANN) from a diverse
lake database.
5. Ecosystem simulation by state variable-based adaptive agents gains resilience
to environmental change from an agent bank providing alternative agents for
same species or functional groups evolved from a diverse lake database.
6. The presented concepts are currently tested by means of a multivariate timeseries database for nine lakes different in climate, eutrophication and
morphology.
Acknowledgements
I am very grateful to Michio Kumagai, Lake Biwa Research Institute, Japan,
Myriam Bormans, CSIRO Land and Water, Australia, Noriko Takamura, National
Institute for Environmental Studies, Japan, Mike Burch, Australian Water Quality
Centre, Australia, Wolfgang Horn, Dresden University of Technology, Germany,
Bomchul Kim, Kangwon National University, South Korea, Olli Varis, Helsinki
University of Technology, Finland, and Diederik van der Molen, Dutch Institute
for Inland Water management, The Netherlands, for making invaluable data
F. Recknagel
nutrient loadings. Each single agent is determined by EDE in order to maximise
(adapt) their performance (abundance) in relation to current environmental
conditions (nutrient loadings, light, temperature and abundance of competitors,
predators or preys). EDE utilise evolutionary algorithms in order to steadily
optimise parameter values and functions of the state variable-based agents by
means of differential equations as used by Park et al. (1974) and Recknagel and
Benndorf (1982). As a result, each agent adapts simultaneously to current
environmental conditions by producing "offspring" agents based on its evaluation
and selection of mates, recombination strategy and mutation strategy (see Figure
5.6b). Successful case studies on EDE have been conducted by Whigham and
Recknagel (2001a, b) and Recknagel et al. (2002).
5.5
Conclusions
1. Adaptive agents (AA) provide a realistic framework for ecosystem simulation,
evolving ecosystem structures and behaviours by emerging, submerging,
interacting and evolving ecological entities.
2. Individual-based AA prove applicable to a spatiaUy explicit simulation of
highly simplified terrestrial food webs.
3. State variable-based AA where evolutionary computation is embodied appear
to be relevant for simulations of aquatic food webs dynamics and plankton
species interactions.
4. Embodiment of evolutionary computation in adaptive agents for aquatic species
or functional groups can be achieved by evolving predictive rules (ER),
differential equations (EDE) or artificial neural networks (ANN) from a diverse
lake database.
5. Ecosystem simulation by state variable-based adaptive agents gains resilience
to environmental change from an agent bank providing alternative agents for
same species or functional groups evolved from a diverse lake database.
6. The presented concepts are currently tested by means of a multivariate timeseries database for nine lakes different in climate, eutrophication and
morphology.
Acknowledgements
I am very grateful to Michio Kumagai, Lake Biwa Research Institute, Japan,
Myriam Bormans, CSIRO Land and Water, Australia, Noriko Takamura, National
Institute for Environmental Studies, Japan, Mike Burch, Australian Water Quality
Centre, Australia, Wolfgang Horn, Dresden University of Technology, Germany,
Bomchul Kim, Kangwon National University, South Korea, Olli Varis, Helsinki
University of Technology, Finland, and Diederik van der Molen, Dutch Institute
for Inland Water management, The Netherlands, for making invaluable data