Abstract
Coastal cetaceans confront intensifying anthropogenic pressures, yet quantifying historical habitat loss remains methodologically challenging in data-scarce regions where shifting baseline syndrome obscures conservation targets. Using the critically endangered Chinese white dolphin ( Sousa chinensis ) as a sentinel species, we synthesized occurrences from historical documents (n = 3944) and local ecological knowledge (LEK, n = 252) to reconstruct its historical distribution shifts in southeast China; and used a two-stage analytical framework to disentangle natural versus anthropogenic drivers of observed range contractions. Maxent models with seven natural variables identified baseline suitable habitats (AUC = 0.918) that congruent with the reconstructed historical range. Generalized linear model analyses demonstrated significant effects of all five anthropogenic disturbances to the recent range contraction of S. chinensis ( P < 0.05), with mariculture exerted the strongest negative effect (β = -1.358), followed by inshore fishing intensity (β = -1.231), terrestrial stressors (β = -0.754), coastal reclamation intensity (β = -0.522), and shipping activities (β = -0.257). We propose three risk-adaptive governance actions: (1) Artificial Intelligence-driven integration of multi-source ecological data with coordinated monitoring networks for coastal cetaceans to bridge data gaps and enable evidence-based governance at regional scales; (2) mitigate risks from ghost gear entanglement, coastal and estuarine maritime engineering, and vessel collisions through targeted technological interventions and adaptive marine spatial planning frameworks; (3) implement ecosystem-based management approaches to reconcile biodiversity conservation with coastal urbanization. This historical ecology-spatial planning nexus provides a transferable framework for conserving data-limited coastal megafauna amid cumulative anthropogenic impacts.