Artificial Neuronal Networks : Application to Ecology and Evolution

In this book, an easily understandable account of modelling methods with artificial neuronal networks for practical applications in ecology and evolution is provided. Special features include examples of applications using both supervised and unsupervised training, comparative analysis of artificial neural networks and conventional statistical methods, and proposals to deal with poor datasets. Extensive references and a large range of topics make this book a useful guide for ecologists, evolutionary ecologists and population geneticists.

I Introduction.- 1 Neuronal Networks: Algorithms and Architectures for Ecologists and Evolutionary Ecologists.- 1.1 Introduction.- 1.2 Back Propagation Neuronal Network (BPN).- 1.2.1 Structure of BPN.- 1.2.2 BPN Algorithm.- 1.2.3 Training the Network.- 1.2.4 Testing the Network.- 1.2.5 Overtraining or Overfitting the Network.- 1.2.6 Use Aspects.- 1.2.7 BPN versus MLR.- 1.3 Kohonen Self-Organizing Mapping (SOM).- 1.3.1 Algorithm.- 1.3.2 Missing Data.- 1.3.3 Outliers.- 1.3.4 Use of Different Metrics.- 1.3.5 Aspects of Use.- 1.4 Conclusion.- Acknowledgements.- References.- II Artificial Neuronal Networks in Landscape Ecology and Remote Sensing.- 2 Predicting Ecologically Important Vegetation Variables from Remotely Sensed Optical/Radar Data Using Neuronal Networks.- 2.1 Introduction.- 2.2 Traditional Extraction Techniques.- 2.3 Neuronal Networks.- 2.4 Uses of Neuronal Networks and Remote Sensing Data.- 2.4.1 Neuronal Networks as Initial Models.- 2.4.2 Neuronal Networks as Baseline Control.- 2.4.3 Neuronal Networks for Inverting Physically-Based Models.- 2.4.4 Neuronal Networks for Defining Relevant Variables.- 2.4.5 Neuronal Networks as Adaptable Systems.- 2.5 Disadvantages of Using Neuronal Networks with Remote Sensing Data.- 2.6 Conclusions and Implications.- References.- 3 Soft Mapping of Coastal Vegetation from Remotely Sensed Imagery with a Feed-Forward Neuronal Network.- 3.1 Introduction.- 3.2 Test Site and Data.- 3.3 Methods.- 3.4 Results and Discussion.- 3.5 Summary and Conclusions.- Acknowledgements.- References.- 4 Ultrafast Estimation of Neotropical Forest DBH Distributions from Ground Based Photographs Using a Neuronal Network.- 4.1 Introduction.- 4.2 Methods.- 4.2.1 Study Site.- 4.2.2 Tree Inventory.- 4.2.3 Photograph Sampling.- 4.2.4 Image Processing.- 4.2.5 Extraction of the Input Vector.- 4.2.6 Neuronal Network Design.- 4.3 Results.- 4.4 Discussion.- 4.5 Conclusion.- Acknowledgements.- References.- 5 Normalized Difference Vegetation Index Estimation in Grasslands of Patagonia by ANN Analysis of Satellite and Climatic Data.- 5.1 Introduction.- 5.2 Methodology.- 5.2.1 Artificial Neuronal Networks.- 5.2.2 The Data Set.- 5.3 Results and Discussion.- Acknowledgements.- References.- 6 On the Probabilistic Interpretation of Area Based Fuzzy Land Cover Mixing Proportions.- 6.1 Introduction.- 6.2 Conceptual Classification.- 6.2.1 The Probabilistic Interpretation of Sub-Pixel Area Proportions.- 6.2.2 Implications of the Probabilistic Interpretation.- 6.2.3 Summary.- 6.3 Sub-Pixel Area Proportion Estimation on the FLIERS Project.- 6.3.1 The Data.- 6.3.2 The Neuronal Networks.- 6.3.3 The Experiments.- 6.3.4 Results.- 6.4 Conclusion.- References.- III Artificial Neuronal Networks in Population, Community and Ecosystem Ecology.- 7 Patterning of Community Changes in Benthic Macroinvertebrates Collected from Urbanized Streams for the Short Time Prediction by Temporal Artificial Neuronal Networks.- 7.1 Introduction.- 7.2 Methods.- 7.2.1 Multilayer Perceptron with Time Delay.- 7.2.2 Recurrent Neuronal Network.- 7.2.3 Field Data.- 7.3 Training and Recognition.- 7.3.1 Multilayer Perceptron with Time Delay.- 7.3.2 Recurrent Neuronal Network.- 7.4 Discussion and Conclusion.- Acknowledgements.- References.- 8 Neuronal Network Models of Phytoplankton Primary Production.- 8.1 Introduction.- 8.2 Materials and Methods.- 8.3 Results.- 8.3.1 Basic Primary Production Modelling: Neuronal Networks vs. Linear Regressions.- 8.3.2 A Depth-Resolved Primary Production Model.- 8.3.3 Modelling Primary Production in the Oceans.- 8.3.4 Neuronal Network Models vs. Conventional Models.- 8.3.5 Sensitivity Analysis.- 8.4 Discussion.- Acknowledgements.- References.- 9 Predicting Presence of Fish Species in the Seine River Basin Using Artificial Neuronal Networks.- 9.1 Introduction.- 9.2 Description and Selection of Data.- 9.3 Methodology.- 9.3.1 Choice of Implementation and Tuning of Parameters.- 9.3.2 Architecture of the Network and Error Criterion.- 9.3.3 Weighting of Data.- 9.3.4 Prediction Error Assessment.- 9.4 Results.- 9.5 Discussion.- 9.5.1 Ecological Soundness.- 9.5.2 On Methodology.- References.- 10 Elucidation and Prediction of Aquatic Ecosystems by Artificial Neuronal Networks.- 10.1 Introduction.- 10.2 Phytoplankton Abundance in Lakes and Rivers.- 10.2.1 Prediction.- 10.2.2 Elucidation.- 10.3 Prediction of Density of Brown Trout Redds in Streams.- 10.4 Conclusions.- Acknowledgements.- References.- 11 Performance Comparison between Regression and Neuronal Network Models for Forecasting Pacific Sardine (Sardinops caeruleus) Biomass.- 11.1 Introduction.- 11.2 Materials.- 11.3 Methods.- 11.4 Results and Discussion.- Acknowledgements.- References.- 12 A Comparison of Artificial Neuronal Network and Conventional Statistical Techniques for Analyzing Environmental Data.- 12.1 Introduction.- 12.2 Methods.- 12.2.1 Database Development.- 12.2.2 Development and Analysis of ANN Models.- 12.2.3 Conventional Statistical Methods.- 12.2.4 Combination of PCA with Other Techniques.- 12.3 Results.- 12.3.1 ANN Analysis and Testing.- 12.3.2 Conventional Statistical Analysis.- 12.3.3 Combination of PCA with Least Squares Regression (LSR) and ANNs.- 12.3.4 Summary of Results.- 12.4 Discussion.- 12.4.1 Comparison of Modelling Performance.- 12.4.2 Determination of the Importance of Input Variables.- 12.5 Conclusion.- References.- IV Artificial Neuronal Networks in Genetics and Evolutionary Ecology.- 13 Application of the Self-Organizing Mapping and Fuzzy Clustering to Microsatellite Data: How to Detect Genetic Structure in Brown Trout (Salmo trutta) Populations.- 13.1 Introduction.- 13.2 Material and Methods.- 13.2.1 Biological Samples.- 13.2.2 Microsatellites.- 13.2.3 Artificial Neuronal Networks.- 13.3 Results and Discussion.- 13.3.1 The Self-Organizing Map.- 13.3.2 The Fuzzy Clustering.- 13.4 Conclusion.- Acknowledgements.- References.- 14 The Macroepidemiology of Parasitic and Infectious Diseases: A Comparative Study Using Artificial Neuronal Nets and Logistic Regressions.- 14.1 Introduction.- 14.2 Materials and Methods.- 14.2.1 Materials.- 14.2.2 Methods.- 14.3 Results.- 14.3.1 Logistic Regressions and Artificial Neuronal Networks Face to Face.- 14.3.2 Occurrence and Threshold Effects.- 14.4 Discussion.- 14.5 Conclusion.- Acknowledgements.- References.- 15 Evolutionarily Optimal Networks for Controlling Energy Allocation to Growth, Reproduction and Repair in Men and Women.- 15.1 Introduction.- 15.2 Optimal Control Model.- 15.3 Computing Optimal Strategies.- 15.4 Fitting the Optimal Control Model.- 15.5 Network Control Model.- 15.6 Optimizing the Network Model.- 15.7 Discussion.- Acknowledgements.- References.- V Perspectives.- 16 Can Neuronal Networks be Used in Data-Poor Situations?.- 16.1 Introduction.- 16.2 Neuronal Network Training and its Limitations.- 16.3 The Geochemical Data Problem.- 16.4 Preprocessing Data.- 16.5 Results.- 16.6 Discussion and Conclusion.- References.