Identifying Core Indicators to Monitor Deforestation and Restoration Trends in South Korea Island Ecosystems

  Island regions serve as critical ecological functions, maintaining unique biodiversity and endemic species. Due to their geographical isolation, island ecosystems are exceptionally vulnerable to external anthropogenic and natural disturbances, including overgrazing by wildlife and feral livestock, land-use changes, and the impacts of climate change. By this susceptibility, establishing precise quantification and configuration of deforestation and restoration are essential for the sustainable management of island regions. This study aims to identify and validate core remote sensing indicators capable of detecting long-term deforestation and restoration efforts across diverse island landscapes in South Korea.

  A comprehensive island library of deforestation and recovery was constructed by synthesizing historical literature and field reports. This library contains a list of deforestation and restoration islands, disturbance types, and disturbed periods. To evaluate spatio-temporal dynamics in island landscape, we utilized multi-temporal satellite imagery and estimated three indicators to represent the vital and biophysical conditions of island ecosystems: Normalized Difference Vegetation Index (NDVI), Soil-Adjusted Vegetation Index (SAVI), Inverted Tasseled Cap Wetness-Greenness Difference (TCWGD_inv).

  The analytical framework employed Sen’s slope to determine the magnitude of monotonic trends in each indicators, to provide a robust rate of degradation or recovery over time. To ensure the statistical significance, the Mann-Kendall trend test was conducted, allowing for a rigorous spatial assessment of areas undergoing significant ecological dynamics. The study focused on four representative island sites: Two island sites, Guleop-do island and Anma-do island, damaged from the intense browsing by the ungulate, one island site, Geoje-do island, damaged by pests, and one active restoration site, Wonsan-do island.

  All three indicators consistently detected significant decreasing trends in deforestation areas, effectively quantifying the reduction of vital of tree species. In the restoration site, NDVI and SAVI showed increased trends over time, efficiently detecting successful restoration. However, TCWGD_inv demonstrated inverse trend, likely due to the high ambient soil moisture characteristics inherent to island regions.

  This study demonstrates that while the integration of these three indicators provides a useful tool for monitoring forest degradation and quantifying damage, the application of moisture-based indices like TCWGD_inv for assessing restoration requires careful calibration according to site-specific environmental variables. These results provide a scientific foundation for developing optimized, data-driven strategies for the long-term conservation and ecological restoration of vulnerable island forest ecosystems.