Contents
XVII
River ................................................................................................... 201
10.4.2 Configuring the Neural Network Architecture for
Predictability ....................................................................................... 203
10.4.3Elucidation ofEcological Hypothesis .............................................. 205
10.4.3.1 Microcystis aeruginosa ................................................................. 207
10.4.3.2Stephanodiscus hantzschii ............................................................ 208
10.5Implications of Ecological Informatics for Limnology ...................... 208
10.6Conclusions ........................................................................................ 210
Acknowledgements ............................................................................ 210
References .......................................................................................... 210
11. An Evaluation ot Methods tor the Selection ot Inputs tor an
Artitieial Neural Network Based River Model ....................... 215
11.1 Introduction ....................................................................................... 215
11.2 Methods ............................................................................................. 217
11.2.1 Unsupervised Input Preprocessing ....................................... 217
11.2.2 Supervised Input Determination ........................................... 220
11.3 Case Study ......................................................................................... 222
11.4 Model Development .......................................................................... 222
11.4.1 Performance Measures and Model Validation ...................... 223
11.4.2 Data Division ........................................................................ 223
11.4.3 Determination of Model Inputs ............................................ 224
11.5 Results and Discussion ...................................................................... 224
11.6 Conclusions ....................................................................................... 230
Acknowledgements ........................................................................... 230
References ......................................................................................... 231
12. Utllity ot Sensltlvity Analysis by Artiticial Neural Network
Models to Study Patterns 01 Endemie Fish Species .. ........... 233
12.1 Introduction ......................................................................................... 233
12.2Contribution of Environmental Variables ........................................... 234
12.3Application to Ecological Data ........................................................... 235
12.4Results ................................................................................................. 236
12.4.1 Predictive Power ...................................................................... 236
12.4.2 Sensitivity Analysis .................................................................. 238
12.5Discussion ........................................................................................... 242
12.6Conclusions ......................................................................................... 244
References ........................................................................................... 244