Development of predictive ecological models for river restoration
D'heygere, T.; Mouton, A.; Dedecker, A.; Adriaenssens, V.; Goethals, P.L.M.; De Pauw, N.
Communications in Agricultural and Applied Biological Sciences 69(2): 15-18
2004
ISSN/ISBN: 1379-1176 PMID: 15560177 Document Number: 567965
Models based on different artificial intelligence techniques were developed and applied to predict the macroinvertebrate communities in the Zwalm river basin located in Flanders, Belgium. Artificial neural networks (ANN) and classification trees (CTs) were developed and applied to predict the impact of remeandering on the habitat suitability of eight macroinvertebrates species in the brook Traveinsbeek, a tributary of the Zwalm river. The eight macroinvertebrate species include Erpobdella, Glossiphonia, Baetis, Asellidae, Cloeon, Haliplidae, Sialis and Simuliidae. Based on the changes of habitat characteristics after remeandering. For six taxa, ANN (most precise predictions) models could be developed. For the latter, amore meaningful CT model could be developed. According to the CT model, the remeandering works did not alter the presence of Asellidae. For Glossiphonia, the development of an expert knowledge based model will be necessary (e.g. based on fuzzy logic). For the other six taxa, the following conclusions could be made: remeandering had no significant effect on the probability of presence of Cloeon and Sialis, while on the contrary, an increase of habitat suitability was detected for Baetis and Simuliidae, whereas a decrease was predicted for Erpobdella and Haliplidae. Cloeon and Baetis are indicators for good water and habitat quality, while Erpobdella occur in more impacted streams.