Title: Factors affecting cyanobacterial blooms and their prediction in a tropical water supply reservoir
Authors: Aline Almeida De Jesus Magalhães; Lafayette Dantas Da Luz; Terencio Rebello De Aguiar Junior
Addresses: Coordination of Environmental Engineering, Federal Institute of Bahia – IFBA, Vitória da Conquista, Bahia State, Brazil ' Department of Environmental Engineering, Federal University of Bahia – UFBA, Salvador, Bahia State, Brazil ' Department of Environmental Engineering, Federal University of Bahia – UFBA, Salvador, Bahia State, Brazil
Abstract: In this study, our objectives were to determine the factors affecting cyanobacterial bloom and predict the abundance and probability of bloom occurrence in the tropical reservoir using generalised linear models (GLMs) with negative and positive binomial distributions. The water quality, weather, and water level data in the reservoir during 2006-2015 were used to develop the models. The results pointed out that: 1) predictive modelling using binomial distribution was the strongest and most parsimonious GLM; 2) total P and electric conductivity were positive factors, whereas 30-d mean water level and dissolved Al were negative factors; 3) water level had a strong significant effect on the probability of bloom events, while total phosphorus was not statistically significant and thus was not the major factor. The set of GLMs was not only an effective tool for identifying key contributing factors affecting cyanobacterial blooms, but also to improve the management of eutrophic tropical reservoirs.
Keywords: cyanobacterial blooms; environmental factors; bloom prediction; bloom occurrence probability; tropical reservoir; human supply reservoir; eutrophic reservoir; modelling; generalised linear model; GLM; negative binomial distribution; binomial distribution; water level; total phosphorus; South America; Brazil.
DOI: 10.1504/IJGENVI.2019.110406
International Journal of Global Environmental Issues, 2019 Vol.18 No.4, pp.314 - 334
Received: 16 Jan 2019
Accepted: 09 Mar 2020
Published online: 19 Oct 2020 *