Preface
XI
neural networks for the bioassessment of the Zwalm river system in Belgium.
Schleiter, Obach, Wagner, Werner, Schmidt and Borchardt carried out a
comprehensive study of the Breitenbach stream (Germany) based on a variety of
unsupervised and supervised learning algorithms for artificial neural networks.
They draw interesting conclusions regarding suitability of different algorithms for
bioindication of stream habitats and input sensitivity of streams. Chon, Park,
K wak and Cha provide a summary of achievements in the structural classification
and dynamic prediction of macroinvertebrate communities in Korean streams by
artificial neural networks. They also discuss patterning of organizational aspects
of macroinvertebrate communities. Huong, Recknagel, MarshalI and Choy study
relationships between environmental factors, stream habitat characteristics and the
occurrence of macroinvertebrate taxa in the Queensland stream system (Australia)
by means of a neural network based sensitivity analysis.
Chapters 10 to 12 contain examples of time series analysis of river water
quality by artificial neural networks. Jeong, Recknagel and Joo apply recurrent
neural networks to explain and predict the seasonal abundance and succession of
different algae species in the River Nakdong (Korea). Validation results reveal a
reasonable correspondence between seven days ahead forecasts and observations
of algal abundance. Information on favouring conditions and processes for certain
algal species discovered by a comprehensive sensitivity analysis comply weIl with
domain knowledge. Bowden, Maier and Dandy combine super- and unsupervised
artificial neural networks as weIl as genetic algorithms for automated input
determination of neural networks in order to forecast the abundance of an algae
species in the River Murray (Australia). Gevrey, Lek and Oberdorff apply two
approaches of sensitivity analysis for the study of riverine fish species by means
of artificial neural networks.
Chapters 14 to 17 provide case studies for the application of fuzzy logic,
artificial neural networks and evolutionary algorithms to freshwater lakes and
marine fishery systems. Karul and Soyupak compare results for the chlorophyIl-a
estimation in three Turkish lakes achieved by multiple regression and artificial
neural networks. Wilson and Recknagel design a generic neural network model
for forecasting algal blooms that is validated by means of six lake databases. It
considers bootstrapping, bagging and time-Iagged training as crucial techniques
for minimising prediction errors. Bobbin and Recknagel apply evolutionary
algorithms to discover rules for the abundance and succession of blue green algae
species in the hypereutrophic Lake Kasumigaura (Japan). Resulting rules
correspond with literature findings, reveal hypothetical relationships and are able
to predict timing and magnitudes of algal dynamics.
Reick, Gruenewald and Page address the issue of data quality in the context of
ecological time-series analysis and prediction. They describe cross-validation and
automated training termination of neural networks applied for multivariate timeseries predictions of marine zooplankton in the German Northern Sea. Chen
combines fuzzy logic and artificial neural networks in order to classify fish stockrecruitment relationships in different environmental regimes near the West Coast
Vancouver Island (Canada) and southeast Alaska (USA).
Chapters 18 to 20 provide examples for the classification of ecological images
at micro and macro scale by artificial neural networks. Wilkins, Boddy and
XI
neural networks for the bioassessment of the Zwalm river system in Belgium.
Schleiter, Obach, Wagner, Werner, Schmidt and Borchardt carried out a
comprehensive study of the Breitenbach stream (Germany) based on a variety of
unsupervised and supervised learning algorithms for artificial neural networks.
They draw interesting conclusions regarding suitability of different algorithms for
bioindication of stream habitats and input sensitivity of streams. Chon, Park,
K wak and Cha provide a summary of achievements in the structural classification
and dynamic prediction of macroinvertebrate communities in Korean streams by
artificial neural networks. They also discuss patterning of organizational aspects
of macroinvertebrate communities. Huong, Recknagel, MarshalI and Choy study
relationships between environmental factors, stream habitat characteristics and the
occurrence of macroinvertebrate taxa in the Queensland stream system (Australia)
by means of a neural network based sensitivity analysis.
Chapters 10 to 12 contain examples of time series analysis of river water
quality by artificial neural networks. Jeong, Recknagel and Joo apply recurrent
neural networks to explain and predict the seasonal abundance and succession of
different algae species in the River Nakdong (Korea). Validation results reveal a
reasonable correspondence between seven days ahead forecasts and observations
of algal abundance. Information on favouring conditions and processes for certain
algal species discovered by a comprehensive sensitivity analysis comply weIl with
domain knowledge. Bowden, Maier and Dandy combine super- and unsupervised
artificial neural networks as weIl as genetic algorithms for automated input
determination of neural networks in order to forecast the abundance of an algae
species in the River Murray (Australia). Gevrey, Lek and Oberdorff apply two
approaches of sensitivity analysis for the study of riverine fish species by means
of artificial neural networks.
Chapters 14 to 17 provide case studies for the application of fuzzy logic,
artificial neural networks and evolutionary algorithms to freshwater lakes and
marine fishery systems. Karul and Soyupak compare results for the chlorophyIl-a
estimation in three Turkish lakes achieved by multiple regression and artificial
neural networks. Wilson and Recknagel design a generic neural network model
for forecasting algal blooms that is validated by means of six lake databases. It
considers bootstrapping, bagging and time-Iagged training as crucial techniques
for minimising prediction errors. Bobbin and Recknagel apply evolutionary
algorithms to discover rules for the abundance and succession of blue green algae
species in the hypereutrophic Lake Kasumigaura (Japan). Resulting rules
correspond with literature findings, reveal hypothetical relationships and are able
to predict timing and magnitudes of algal dynamics.
Reick, Gruenewald and Page address the issue of data quality in the context of
ecological time-series analysis and prediction. They describe cross-validation and
automated training termination of neural networks applied for multivariate timeseries predictions of marine zooplankton in the German Northern Sea. Chen
combines fuzzy logic and artificial neural networks in order to classify fish stockrecruitment relationships in different environmental regimes near the West Coast
Vancouver Island (Canada) and southeast Alaska (USA).
Chapters 18 to 20 provide examples for the classification of ecological images
at micro and macro scale by artificial neural networks. Wilkins, Boddy and