Fine-scale mangrove species monitoring and ecological threshold assessment for coastal management in Quanzhou Bay, China.

Reliable species-level monitoring of mangrove ecosystems is increasingly required for environmental assessment and regulatory decision-making, yet a methodological gap persists between remote sensing outputs and ecological risk evaluation. This study proposes an integrated assessment framework that links multi-sensor satellite imagery with ecological niche modeling to support precision monitoring and threshold-based management. Using Sentinel-2 and SPOT-6 data combined with an object-based U-Net model, we achieved high species-level classification accuracy (OA = 89.02%, Kappa = 0.82) for approximately 483.6 ha of mangroves in Quanzhou Bay, China. Generalized Additive Models (GAMs) quantified key environmental thresholds, revealing a structured zonation scheme: Kandelia obovata prevails in highly productive, hydrologically connected zones (NDVI > 0.43; NDTI > 0.37), Aegiceras corniculatum persists in moderately stressed transitional habitats, while Avicennia marina exhibits critical niche compression, with occurrence probability sharply declining beyond approximately 200 m from tidal creeks. These quantified thresholds provide measurable indicators for ecological monitoring, enabling the early identification of vulnerable species and habitat degradation risks. By integrating species-level distribution mapping with quantified environmental thresholds, the proposed approach operationalizes remote sensing products into a tiered set of management-relevant indicators, including mangrove habitat extent, fine-scale species distribution patterns, and key environmental drivers of species zonation. These evaluation-ready metrics support targeted hydrological restoration, species-specific rehabilitation planning, and invasive species control. This framework demonstrates strong transferability for environmental condition assessment and adaptive monitoring in coastal wetlands under increasing anthropogenic pressure.