Spatio-temporal Dynamics and Driving Mechanisms of Global Human-Transformed fishponds the “Blue Transformation” Paradigm

With the progressive advancement of the global “Blue Transition” strategy, extensive areas of natural wetlands have been converted into high-intensity fishponds to secure the global supply of aquatic protein. Aquaculture currently accounts for nearly 50% of the world’s edible fish production, with Asia contributing approximately 88% of the total output. Despite its critical role in food security, this large-scale transformation of land cover has generated substantial environmental risks, including wetland degradation and loss, declining water quality, and disruptions to regional hydrological cycles. Owing to the fragmented spatial configuration of aquaculture ponds and their pronounced spectral heterogeneity, existing global land-cover products remain inadequate for accurately detecting these subtle yet widespread human modifications, frequently misclassifying aquaculture ponds as natural water bodies or agricultural land. The absence of consistent, long-term benchmark datasets has therefore severely constrained rigorous assessments of the global environmental footprint of aquaculture expansion.

To address this critical data gap, this study leverages the Google Earth Engine (GEE) cloud-computing platform to integrate more than 200,000 multi-source satellite images from the Landsat and Sentinel missions spanning the period 2000–2025. Based on this extensive archive, a global dynamic monitoring framework for fishponds was developed at a spatial resolution of 30 m. A standardized validation dataset comprising 15,000 reference points was established across 12 representative geographic regions worldwide through systematic, expert-level visual interpretation. The proposed methodological framework combines a newly developed Aquaculture Pond Index (API) with a Support Vector Machine (SVM) classifier, explicitly targeting the persistent spectral confusion between aquaculture ponds and paddy fields in complex inland environments.

Using this framework, preliminary quantitative analyses yielded the following key findings. (1) Classification accuracy: The proposed approach achieved a global Overall Accuracy (OA) of 81.6% with a Kappa coefficient of 0.79. In complex inland landscapes, classification performance improved by approximately 12% relative to existing mainstream global land-cover products, demonstrating the robustness of the API in discriminating aquaculture ponds from spectrally similar land features. (2) Areal dynamics: Between 2000 and 2025, global fishponds exhibited a persistent expansion trend, with total area increasing by approximately 35%. Approximately 65% of this expansion was attributed to the conversion of natural wetlands and low-lying agricultural land. (3) Spatial patterns: A pronounced pattern of land-oriented clustering was identified, with inland aquaculture expansion rates surpassing those of traditional coastal regions. Emerging economies in Southeast Asia and East Africa have increasingly become new growth centers for global aquaculture development.

This study fills a critical gap in long-term, spatially explicit monitoring of inland aquaculture at the global scale. The findings provide a robust scientific basis for evaluating the long-term impacts of human-driven aquaculture expansion on surface water resources and offer essential spatial benchmark information to support the reconciliation of global food security objectives with wetland conservation priorities under the United Nations Sustainable Development Goals (SDGs).