General introduction I
Using bioindicators to assess rivers in Europe: An overview 7
1.1. ILntroduction 7
1.2. Stream typology 7
1.3 Diatom ecology and use for river quality assessment 9
1.4. Typologies, assessment systems and prediction techniques based on 12
macroinvertebrates
.5. Advantages of using fish as an indicator taxon 16
1.6. Conclusions 18
Review of modelling techniques 21
2.1. Iroduction 21
2.2. Conventional statistical models 21
23. Artificial neural networks (ANNs 26
2.4. Bayesian and Mixture models 35
5 Support vector machines (SVMs, 37
2.6. Genetic algorithms (GAs) 38
2.7. Mutual information and regression naximisation (MIR-max) 39
2.8. Structural dynamic models 39
3 Fish community assemblages 41
3.1, Introduction 41
3.2. Pattemrning riverine fish assemblages using an unsupervised neural 43
network
3.3. Predicting fish assemblages in France and evaluating the influence 54
of their environmentai variables
3.4. Fish diversity conservation and river restoration in southwest 64
France: a review
3.5. Modelling of freshwater fish and macro-crustacean assemblages for 76
biological assessment in New Zealand
3.6. A Comparison ofvarious fitting techniques for predicting fish yield 90
in Ubolratana reservoir (Thailand) from a time series data
3.7. Patterning spatial variations in fish assemblage structures and diver- 100
sity in the Pilica River system
3.8. Optimisation of artificial neural networks for predicting fish assem- 114
blages in rivers
4. Macroinvertebrate community assemblages 131
4.. Introduction 131
4.2. Sensitivity and robustness of a stream model based on artificial neu- 133
ral networks for the simulation of different management scenarios
4.3. A neural network approach to the prediction of benthic macroinver- i47
tebrate fauna composition in rivers
4.4. Predicting Dutch macroinvertebrate species richness and functional 158
feeding groups using five modelling techniques
45. Coimparison of ciustering and ordination methods implemented to i67
the full and partial data of benthic macroinvertebrate communities
in streams and channels
4.6, Prediction of macroinvertebrate diversity of freshwater bodies by 189
adaptive learning algorithms
4.7. Hierarchical patterning of benthic macroinvertebrate communities 206
using unsupervised artificial neural networks
.8 Species spatial distriution and richness of stream insects in south- 22
w ester France using artificial neural networks with potential use
for biosurveillance
.9. Pattenming comnmunt changes in benthic macroinverLebrates in a 239
polluted steam by using alifiteial neural netw oris
4.1 Pattening, prdicting stream macro ivertebrate assemblages in Vic- 2
toria (Austraia usng artificial neural networks and genetic algo
rithms
5 Diatom and other algal assemblages 261
5.. troduction 26
52. Applying case-based reasoning to explore Ureshwater phytoplankton 263
dynamnics
5.3. Modelling community changes of cyanobacteria in a flow regulated 273
river (he lowerNado S. Korea) by means of a Self-
Organizing Map (SOM)
.. Use of artficial itelligence (MiR-max) and chemical index to de 288
fine type diatom assemblages in Rhone basin and Mediterranean re-
gino
55. Classification of stream diatom communities using a selforganizing 304
man
5.6 Diatom typology of owimpac ted conditions at a ulti-regional 3 7
sale: combined reults of mutivariate analyses and SOM
5.7 Prediction with artifial neura networks of diatom assemblages in 343
headwater streams of Luxemborg
5. Use of neural new ork models to predict diatom assemblages in the 355
Loire-Bretagne asin (France
. Development of community assessment techniques 367
6.1. introduction 36
.2. Evuation of relevant species in communties: development of 369
structring indices for the classifcation of communities using a sel
organizing map
63. Projec ion pursuit with robust indices for the analysis of ecological 381
data
6.4 A framework for computer-based data analysis and visualisation by 390
pattern recognition
6:5. A rule-based vs. a set-covering inplementation of the knowledge 401
system LIMPACT and its significance for maintenance and discov-
ery of ecological knowledge
1 User interface tool 435
7. Introduction 435
72 Software aims 436
7, System requirements 436
74. Installig nonstalling 436
7.5 Models implemented in the tol 436
7 How to use the software 439
. Organisms used in the PAEQAN software 447
8 General conclusions and perspectives 451
References 455