Machine learning reveals ecological thresholds and predicts future bacterial community shifts in a chronically...

Area of Science:

• Environmental microbiology
• Estuarine ecology
• Machine learning applications

Background:

• Eutrophic estuaries face climate change impacts and complex microbial responses.
• Predicting microbial community shifts under multiple stressors is challenging due to non-linear interactions.

Purpose of the Study:

• Develop a machine learning framework to model bacterial community dynamics in the Pearl River Estuary.
• Identify key environmental drivers and ecological thresholds influencing microbial composition.
• Forecast estuarine bacterial community shifts under future climate scenarios.

Main Methods:

• Applied machine learning to model seasonal bacterial community data from a chronically eutrophic estuary.
• Analyzed the influence of environmental variables like temperature, silicate, and nutrient levels.
• Identified specific thresholds for key bacterial phyla (Firmicutes, Actinobacteriota, Proteobacteria).

Main Results:

• Temperature, silicate, and nitrite/dissolved oxygen explained 66-82% of community variance.
• Specific environmental thresholds were identified for significant changes in dominant bacterial phyla.
• Future high-emission scenarios project increased beta-diversity and a potential decline in key biomarker taxa.

Conclusions:

• Machine learning provides a predictive tool for estuarine microbial management.
• Identified thresholds offer data-driven targets for pollution control and ecosystem restoration.
• The framework is transferable to other complex estuarine systems globally.