Toward Cross-Scale and Intelligent Assessment of Aquatic Ecosystem Health: A Paradigm Shift from Description to Prediction

Freshwater ecosystems are undergoing unprecedented degradation driven by accelerating global change and intensive human activities, exposing fundamental limitations of conventional, indicator-based assessments that remain largely descriptive and retrospective. Here, we synthesize the conceptual evolution of aquatic ecosystem health assessment and summarize the frameworks of different countries and regions. Several key bottlenecks are identified, including baseline drift, weak mechanistic links between biological indicators and stressors, scale mismatches, and limited capacity to detect non-linear tipping points. To address these challenges, we advocate a paradigm shift toward cross-scale and intelligent assessment, transitioning ecosystem management from post-hoc diagnosis to predictive early warning. We propose a gene-to-landscape framework that integrates micro-scale multi-omics for early risk detection, mini-scale environmental DNA for biodiversity surveillance, ecosystem-scale observations, and macro-scale AI-enabled remote sensing. By coupling multi-source data with artificial intelligence, this approach enables causal inference, tipping-point detection, and anticipatory decision support. We argue that cross-scale intelligence provides a critical foundation for mechanistic understanding and adaptive management, redefining aquatic ecosystem health assessment in the era of global change and accelerating the transition from descriptive monitoring toward predictive, actionable freshwater governance.