Attractor landscapes for characterising ecological resilience in real-world systems

Advanced warning of catastrophic changes in ecosystem function and composition are fundamental to protect and restore nature, particularly from novel climate threats. Critical slowing down (CSD) has been proposed as an early warning system for ecosystem critical transition and, alongside other state measurements, for defining a system’s ecological resilience. These concepts are based on the idea of stable states, which are difficult to define in ecological systems, and have not been demonstrated empirically because reference states are inherently static in field data. Using remote sensing data we show the dynamic regime of real-world ecological systems by mapping trajectories as a surface to quantify the geometry governing transitions between locally stable states. We find different attractor landscapes between ecosystems with extremes in biodiversity. Sites with high diversity (such as nature reserves) show a single local minima representing a single stable state - a resilient system, while ecosystems with low diversity exhibit multiple local stable states and trajectories between states after perturbation (drought), showing a critical transition. Our results evidence the theory of resilience and stability of ecological systems. We anticipate the use of our approach to better understand and visualize ecosystem resilience and as a tool for identifying ecosystems in critical transition that can be targets for intervention, such as ecological restoration.