Dynamic Modeling of Agroecosystem Resilience: Entropy-Weighted Stability Index and XGBoost-Driven Policy Optimization for Organic Agriculture

Abstract

This study proposes a multidimensional framework integrating dynamic food web modeling, entropy weighting, and XGBoost optimization to evaluate the ecological and economic impacts of transitioning to organic agriculture. The study has developed a Composite Stability Index (CSI) that synthesizes five indicators—the Plant Health Index (PHI), the Insect Population Stability Index (IPSI), the Bird/Bat Population Stability Indices (BPSI/BtPSI), and the Species Diversity Index (SDI)—to quantify agroecosystem resilience. The dynamic food web model, governed by nonlinear differential equations, simulates interactions among crops, weeds, pests, birds, and bats under scenarios including herbicide reduction, species reintroduction, and bat population augmentation. The entropy weighting method objectively assigns indicator weights, validated by an R² value of 0.89. Sensitivity analysis demonstrates that low-intensity pesticide use enhances CSI by 23.7%, while bat introduction boosts pest control efficiency by 41.2%. Long-term simulations demonstrate an 18.5% improvement in soil fertility recovery over a decade, aligning with findings from long-term organic trials. The XGBoost algorithm has been developed to optimize trade-offs between economic profit and sustainability, providing actionable insights for policymakers. This work bridges ecological dynamics and agricultural economics, offering a decision-support tool for balancing productivity, biodiversity conservation, and long-term soil health in organic transition strategies.