Abstract
Semi-arid savanna ecosystems are increasingly vulnerable to climate change and anthropogenic disturbances, leading to significant changes in woody cover and structure. Monitoring and managing these changes are critical, especially in protected areas where resources are limited and traditional methods such as in situ field data collection are labour-intensive, expensive and not sufficient to capture large-scale spatial and temporal shifts. This study aims to evaluate the reliability of remote sensing for detecting abrupt ecosystem shifts by comparing it with expert knowledge and field observations, identifying their ecological drivers, and assessing its potential to support landscape-scale conservation planning. The analysis was performed in the Kruger National Park, South Africa using several data sources from the time period 1982 to 2023, allowing the use of findings for prioritising management actions. Despite differences in methods, remote sensing and expert-delineated maps showed strong agreement, with a spatial overlap of 65.6% and an overall accuracy of 82%. Remote sensing proved particularly valuable in detecting changes in inaccessible areas, validating its role in broad-scale ecological monitoring. Furthermore, we observed significant differences in species diversity, dominant species and vegetation structure in areas where remote sensing detected shifts, confirming the validity of the method. Shrub encroachment, which may be associated with herbivory pressure and increased rainfall severity, emerged as a key factor in ecosystem transformation. The integration of remote sensing, expert knowledge and ground-level data provides a robust approach to understanding and managing savanna ecosystem dynamics. This research highlights the importance of integrating multiple data sources to enhance ecosystem monitoring and inform management strategies to ensure resilient savanna landscapes.
Funding Information
  • VLIRUOS (ZA2019TEA490A101)

This publication has 64 references indexed in Scilit: