Predicting regime shifts under non-uniform harvesting in multilayer ecological networks

Human harvesting in real-world ecosystems is typically characterized by pronounced spatial selectivity, whereas most studies on regime shifts are based on the assumption of uniform external pressure. This paper constructs a three-layer ecological network dynamical model incorporating plant–pollinator mutualistic interactions and plant–parasite antagonistic interactions. A non-uniform harvesting mechanism is introduced at the node level, and a harvesting-based grouping method for low-dimensional representation is proposed, mapping the high-dimensional network system into effective group-level dynamics that preserve structural heterogeneity. Through simulation-based parameterization and numerical validation, the regime boundaries jointly determined by harvesting intensity and coverage are systematically identified. The results show that, under the same total harvesting magnitude, concentrating pressure on a single functional layer significantly compresses the system’s safe operating space and shifts the critical harvesting intensity toward lower values. Structural analysis further reveals that system vulnerability is highly concentrated in a small number of cross-layer hub species, whose perturbation effects are primarily amplified through parasitic negative feedback channels. The study demonstrates that regime shifts in multilayer ecological systems are determined not only by disturbance intensity but also by the spatial positioning of disturbances and their matching relationship with cross-layer coupling structures. This paper establishes, at the mechanistic level, a quantitative mapping between microscopic structural roles and macroscopic regime boundaries, providing a theoretical basis for structure-oriented ecological risk assessment and conservation strategy design.