Quantifying microbial and harmful algal bloom interactions dynamics through machine learning predictive frameworks
Year: 2025
Type: article
Abstract: Harmful algal bloom (HAB) mitigation using microorganisms faces challenges, like inconsistent algicidal efficacy, scalability issues, limited microbial diversity, and traditional modeling constraints. To address these challenges, for the first time, this study investigates the quantitative impacts of microorganisms on HABs using machine learning (ML). We conducted an analysis of over 100 microbial species, laying the foundation to optimize treatments for more consistent outcomes. Our findings showed that these microorganisms exhibited an average algicidal ratio of 71 % against HABs. Random Forest, Bagging, and Extreme Gradient Boosting Regressors ML models demonstrated robust predictive performance for algicidal ratios. Nonlinear relationships with time, microbial concentrations, and algal biomass were observed. Optimal culture time for maximizing algicidal ratios ranged from 7 to 10 days; extending time did not enhance the algicidal process. The optimum concentration of microorganisms was identified as 0.5 × 10 10 cells/mL, with higher concentrations showing diminishing returns. Additionally, algal biomass influenced the algicidal process, with a sharp decline in algicidal ratio at low biomass levels (0–1 × 10 7 cells/mL), followed by stabilization at higher densities (1.5 × 10 7 cells/mL). Algal biomass, culture time, Brevibacillus , microbial concentration, Enterobacter , Microcystis aeruginosa , and Alexandrium tamarense emerged as crucial factors driving microbial algicidal processes against HABs. Overall, ML analysis of microbial impacts on HABs guides effective treatment optimization with future work emphasizing real-world field applications. • Inconsistent data undermine microbes as biocontrol agents for algal blooms. • Sophisticated ML methods provide accurate predictions for algicidal ratios. • The experimental parameters and their interactive effects were thoroughly quantified and validated. • Established thresholds offer improved protection of aquatic ecosystems.
Source: Water Biology and Security
Authors: Zobia Khatoon, Suiliang Huang
Institution: Nankai University
Language: English
Subfield: Environmental Chemistry
Field: Environmental Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4414823144