Revealing Physiochemical Factors and Zooplankton Influencing Microcystis Bloom Toxicity in a Large-Shallow Lake Using...

Area of Science:

• Environmental Science
• Ecotoxicology
• Limnology

Background:

• Toxic cyanobacterial blooms pose a global health and environmental risk.
• Research often links cyanobacterial composition to toxin production, but environmental drivers of toxin hazard are less understood.

Purpose of the Study:

• To investigate environmental conditions influencing cyanotoxin hazard in a large, shallow lake.
• To quantify the response of *Microcystis* bloom toxicity to physicochemical and zooplankton factors using machine learning.

Main Methods:

• Analysis of a 22-site dataset with monthly water quality, cyanobacteria, zooplankton, and microcystin (MC) data.
• Imputation of missing MC values using non-negative latent factor (NLF) analysis.
• Application of Bayesian additive regression trees (BART) to model *Microcystis* toxicity and biomass.

Main Results:

• BART models outperformed comparative methods in predicting *Microcystis* biomass and MC concentrations.
• Shade index was the primary predictor of MC concentrations, indicating light limitation effects.
• pH, dissolved inorganic nitrogen, total phosphorus, protozoa, and copepods were significant factors influencing bloom biomass and toxicity.

Conclusions:

• Light limitation, pH, and nutrient availability are key drivers of *Microcystis* toxicity.
• Zooplankton, particularly protozoa and copepods, play a role in regulating bloom dynamics and toxin levels.
• Machine learning approaches are valuable for understanding complex, nonlinear relationships in harmful algal bloom research.