Modeling ecosystem-wide responses to environmental stressors: A multi-trophic hierarchical Bayesian network approach

Area of Science:

• Ecology
• Environmental Science
• Computational Biology

Background:

• Effective aquatic ecosystem management necessitates models capturing complex trophic interactions and environmental stressors.
• Existing data-driven ecological models often lack scope, focusing on limited taxa or simplified structures, hindering ecosystem-wide assessments.
• Hierarchical Bayesian networks (HBNs) offer a solution by integrating latent variables for correlated ecological relationships, reducing complexity and enhancing interpretability.

Purpose of the Study:

• To develop and apply a Hierarchical Bayesian Network (HBN) for predicting aquatic community responses to diverse environmental drivers.
• To assess the HBN's predictive performance against conventional Bayesian networks for multi-trophic dynamics in riverine ecosystems.
• To identify key environmental factors influencing aquatic community structure in South Korean river basins.

Main Methods:

• Developed an HBN incorporating phytoplankton, zooplankton, benthic macroinvertebrates, and fish responses to meteorological, water quality, hydrological, and riverbed variables.
• Utilized KF-METAWEB, a trophic interaction database for Korean freshwater ecosystems, to inform the HBN structure.
• Compared HBN performance (accuracy, AUC) against knowledge-based and data-driven conventional Bayesian networks.

Main Results:

• The HBN demonstrated superior predictive performance (mean accuracy = 0.787, AUC = 0.705) compared to conventional Bayesian networks.
• Sensitivity and scenario analyses revealed water quality and substrate composition as critical drivers for benthic macroinvertebrate and fish communities.
• The hierarchical structure effectively captured cascading effects across trophic levels and environmental gradients.

Conclusions:

• This study presents the first application of HBNs for predicting multi-trophic dynamics in riverine ecosystems.
• HBNs provide a transparent, data-informed tool for ecological assessment and adaptive river basin management.
• The findings highlight the importance of water quality and substrate for maintaining aquatic community structure in river systems.