<p>Wetlands are rapidly changing due to agriculture, climate shifts, and human activities, threatening their ecological balance globally. This study provides a comprehensive analysis of land use and land cover (LULC) changes within the Gungiajuri Haor wetland ecosystem in northeastern Bangladesh from 2000 to 2024. Using remote sensing and GIS, this research supports the sustainable management of this ecologically important wetland. High-resolution Landsat-7 and Landsat-8 images were analyzed in Google Earth Engine to calculate the Normalized Difference Vegetation Index (NDVI), helping monitor vegetation cover changes over time. The study employed a Classification and Regression Trees (CART) machine learning algorithm for precise land cover classification. Ground-truthing through drone surveys and historical Google Earth imagery yielded overall classification accuracies of 87.5–92.5% with kappa coefficients of 0.803–0.887, exceeding those typically reported for wetland studies in Bangladesh (&lt; 85%) and indicating improved reliability from the integration of machine learning, cloud-based processing, and drone-validated reference data. The findings indicate a notable increase in cropland, growing from 49.5% in 2000 to 62.7% in 2024 during the dry season, and from 44.3% to 58.3% in the rainy season. Meanwhile, vegetation coverage has significantly declined, decreasing from 38% to 22.4% in the dry season and from 33.9% to 19.9% in the rainy season, pointing to concerning habitat loss. Water bodies in the ecosystem have remained relatively stable, with seasonal variations. This research highlights the negative effects of expanding agriculture on biodiversity and ecosystem services, emphasizing the need for sustainable land management strategies. The findings serve as an essential resource for policymakers and conservationists, guiding the development of balanced approaches that protect ecological health while supporting economic needs.</p> Graphical Abstract <p>This Graphical Abstract Provides a Detailed Spatiotemporal Analysis of Land Use and Land Cover (LULC) Dynamics in Gungiajuri Haor, an Important Wetland Ecosystem in Northeastern Bangladesh, Covering the Period from 2000 To 2024. The Left Panel Shows the Geographical Context of the Study area, Highlighting its Position Within the Habiganj District and across Four subdistricts. Using Landsat 7 and 8 Satellite Imagery Processed Through Google Earth Engine, the Research Employed NDVI Analysis and Supervised Classification with the CART Algorithm To Monitor Seasonal LULC Changes. The Center and Right Panels Present Key Findings Through Land Cover Maps for both Dry and Rainy Seasons at Four time Points: 2000, 2014, 2018, and 2024. The Analysis Identified Three Main LULC Classes: Cropland (yellow), Vegetation (green), and Water Bodies (blue). Over the 24 years, Cropland Consistently expanded, Especially during the Dry season, Increasing from 49.5% To 62.7%, while Vegetation Decreased from 38% To 22.4%. Water Bodies Showed Fluctuations but Remained Relatively Stable Throughout the Study. A Bar Graph below the Maps Illustrates these Seasonal and Temporal Trends quantitatively. The Study Reveals a Clear Trend of Agricultural Intensification that Reduces Natural Vegetation, Raising Serious Ecological concerns. By Combining Remote sensing, Machine learning, and Temporal Analysis, this Research Highlights the Importance of Sustainable Land Management Policies To Protect Haor Biodiversity and Ecosystem services. The Graphic Summary Supports Policy Discussions and Conservation efforts, Offering a Replicable Framework for Other Wetlands Experiencing Rapid human-driven Change</p>