Forecasting the cyanotoxins presence in fresh waters: A new model based on genetic algorithms combined with the MARS technique

• Published in: Ecological engineering Vol. 53; pp. 68 - 78
• Main Authors: Alonso Fernández, J.R., Díaz Muñiz, C., Garcia Nieto, P.J., de Cos Juez, F.J., Sánchez Lasheras, F., Roqueñí, M.N.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.04.2013; Elsevier
• Subjects: Air. Soil. Water. Waste. Feeding; algorithms; Animal, plant and microbial ecology; Biological; Biological and medical sciences; Cyanobacteria; cyanobacterial toxins; Cyanotoxins; drinking water; Ecology; Environment. Living conditions; Forecasting; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Genetic algorithms; Genetic algorithms (GAs); humans; Mathematical models; Medical sciences; Methods and techniques (sampling, tagging, trapping, modelling...); Multivariate adaptive regression splines (MARS); public health; Public health. Hygiene; Public health. Hygiene-occupational medicine; recreation; Reservoirs; Risk; Spain; Statistical machine learning techniques; water reservoirs; Spain; Europe; Multivariate adaptive regression splines (MARS); Genetic algorithms (GAs); Statistical machine learning techniques; Cyanobacteria; Cyanotoxins; Fresh water; Risk; Multivariate analysis; Forecasting; Multivariate adaptive regression splines; Statistical method; Prevention; Toxin; Genetic algorithm; MARS; Bacteria; Phytoplankton; Public health
• ISSN: 0925-8574; 1872-6992
• DOI: 10.1016/j.ecoleng.2012.12.015

► A hybrid GA–MARS model is built as a predictive model of cyanotoxins presence. ► Cyanobacterial HABs are dangerous for environment and people in fresh waters. ► Biological and physical–chemical variables in this process are studied in depth. ► The obtained regression accuracy of our method is 98%. ► The results show that GA–MARS model can assist in the diagnosis of cyanotoxins. Cyanobacteria are one of the major concerns to public health since some of them produce a range of potent toxins (cyanotoxins). This group of microorganism can be present in drinking and recreation waters representing a health risk for animals and human being. For this reason, as prevention, it is important to bring forward their presence. In this study, using physical–chemical and biological parameters, a hybrid approach based on genetic algorithms (GAs) combined with the multivariative adaptative regression splines (MARS) technique, was developed and applied for forecasting the presence of cyanobacteria in a water reservoir (Trasona reservoir, Northern Spain) and in consequence, the cyanotoxin risk. The significance of each biological and physical–chemical variables used for its determination was assessed and a predictive model useful for preventing the presence of cyanobacteria, and consequently of cyanotoxins, was defined.