Mangrove ecosystems in Sinjai Regency, South Sulawesi, Indonesia, are increasingly exposed to coastal land-use conversion, aquaculture expansion, and biophysical pressures, yet explicit spatial evidence to guide restoration priorities remains limited. This study assessed the spatial and temporal dynamics of mangroves using Landsat SR imagery from 2005 to 2025. Six land-cover classes were mapped (mangrove, non‑mangrove vegetation, rice fields, aquaculture ponds, built‑up areas, and water bodies) through supervised classification, comparing the performance of three machine learning algorithms: Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost). Reliable classification accuracy was achieved with Kappa values exceeding 0.80. Mangrove extent declined from 267 ha in 2005 to 231 ha in 2025, representing a net loss of 36.09 ha. Over the same period, aquaculture ponds expanded by 74.79 ha and built‑up areas increased by 13.59 ha, while water bodies decreased by 49.05 ha. Linear trend analysis indicated a consistent decline in mangrove extent, with an estimated loss rate of approximately 1.80 ha yr⁻¹ and a strong temporal fit (R² = 0.72). Landscape metrics revealed a decreasing number of mangrove patches and edge density, suggesting reduced ecosystem connectivity. Hotspot analysis provided a spatial basis for identifying management zones with high carbon stock potential and areas requiring restoration priority. By integrating remote sensing, land‑cover transition analysis, and landscape metrics, this study offers a reproducible framework to support sustainable mangrove ecosystem management.