Author(s): Zobia Khatoon; Suiliang Huang
Linked Author(s): Suiliang Huang
Keywords: Allelopathic optimization; Harmful algal blooms; Machine learning; Threshold dynamics
Abstract: Allelochemicals are eco-friendly agents for controlling harmful algal blooms (HABs), but their effectiveness and the accuracy of machine learning (ML) models in predicting chlorophyll removal are not well studied. This research uses ML models to predict chlorophyll removal rates, finding Random Forest Regressor (RFR) to be the most accurate. Most experiments used low allelochemical concentrations and lasted briefly, achieving an average chlorophyll removal of 53.34%. Maximum removal occurred between 5–20 days with concentrations of 0–2 g/L and initial chlorophyll a below 6 mg/L. Influential factors included allelochemical concentration, contact time, and starting chlorophyll-a levels. Key allelochemicals identified were Kaempferol, Nuciferine, Tea polyphenols, L-Lysine, Vallisneria extract, 4-tert-butylpyrocatechol, linoleic acid microspheres, and extracts from Ranunculus aquatilis. The study supports using allelochemicals for HAB control and emphasizes the need to validate these findings in real environments.
Year: 2026