Coastal ecosystems such as mangroves, salt marshes, and seagrass meadows are critical to global climate regulation and ecological sustainability. They serve as highly efficient long-term carbon sinks, regulate water quality, and provide essential ecosystem services including shoreline stabilization and biodiversity support. Yet, these systems face unprecedented pressures from anthropogenic disturbance, land-use change, and climate-driven stressors such as sea-level rise and ocean acidification. Conventional monitoring and modeling frameworks, while valuable for mechanistic understanding, often lack the capacity to fully capture the nonlinear interactions, spatial heterogeneity, and temporal variability that define coastal ecosystems. To address this gap, this study develops and evaluates an AI-integrated predictive framework designed to enhance the assessment of carbon sequestration capacity and water quality dynamics in coastal environments. The framework leverages advanced machine learning and deep learning models, remote sensing technologies, and in situ ecological indicators to deliver high-resolution spatiotemporal predictions that link carbon flux, nutrient cycling, and pollutant dispersion to ecosystem performance. Findings from the meta-analysis and empirical validation reveal robust evidence that coastal vegetated ecosystems sequester significantly higher quantities of carbon compared to degraded or non-vegetated controls. Mangroves exhibited the largest effect sizes, with soil carbon densities frequently exceeding 1,000 Mg C ha⁻¹, while salt marshes demonstrated strong sediment-trapping efficiency and seagrass meadows provided moderate but significant contributions, heavily influenced by water clarity and hydrodynamics. For water quality, pooled results confirmed consistent associations between nutrient enrichment, elevated chlorophyll-a, and declining dissolved oxygen levels, with hypoxia severity most pronounced in stratified estuaries. Comparative assessments demonstrated that AI-driven models, including random forests, gradient boosting, and recurrent neural networks, outperformed traditional statistical and process-based frameworks, achieving lower RMSE values, higher predictive power, and stronger capacity to capture nonlinear thresholds. Integrated analyses revealed that improvements in water quality, such as reduced nutrient loading and enhanced optical clarity, directly supported higher carbon burial rates in seagrass meadows and marsh soils, highlighting the reciprocal benefits of ecosystem management interventions. These findings reinforce the interconnected nature of blue carbon and water quality services, validating AI-enhanced approaches as critical assets for adaptive coastal governance, restoration planning, and international sustainability initiatives.
Md. Rashedul Islam (Sun,) studied this question.