Spatial prioritization of conservation efforts in the Magaliesberg biosphere based on land use and land cover change dynamics

Abstract

This study assessed land use and land cover (LULC) change in the Magaliesberg Biosphere (MB) from 2007 to 2016 and from 2016 to 2024 to identify where conservation action should be prioritised and to evaluate how the zoning system is performing under long-term pressure. It is the first multi-temporal LULC assessment of the MB since its designation as a United Nations Educational, Scientific and Cultural Organization (UNESCO) site. The analysis used mediumresolution (resampled to 20 m) Landsat imagery for all years, specifically Landsat 7 Enhanced Thematic Mapper Plus (ETM+) for 2007, Landsat 8 Operational Land Imager (OLI) for 2016 and Landsat 9 Operational Land Imager-2 (OLI2) for 2024. These datasets provided consistent long-term coverage suitable for biosphere-scale monitoring. Supervised classification and index-based Change Vector Analysis (CVA) were combined to detect both visible class transitions and continuous spectral change. This mixed-method approach captured clear conversions, subtle ecological shifts and repeated disturbance that may not yet be reflected in LULC classes. A field survey was conducted to create ground truth points that assessed the accuracy of the 2024 LULC dataset. The LULC maps achieved overall accuracies above 90% with balanced user’s and producer’s accuracies across major classes. The CVA change-no change mask reached 92.5% accuracy and a Kappa value of 0.85, allowing a reliable hotspot map to be produced using the 90th-percentile magnitude threshold. Results show that change is spatially uneven and strongly shaped by biosphere reserve zoning. The transition zone recorded the highest concentration of change in both intervals (2007 to 2016 and 2016 to 2024), followed by the buffer zone, while the core zone remained stable. Intensity analysis, which compared observed change rates with a hypothetical uniform-change scenario, confirmed this gradient by showing that conversions from natural to human-modified classes were more concentrated and progressed faster in the transition and buffer zones than would be expected if change were evenly distributed across the MB. Grassland classes formed the main sources of transitions into agriculture, built-up areas, bushveld and mining disturbance. Persistent hotspots occurred along the Rustenburg mining belt, the northern agricultural plains and the eastern peri-urban fringe. CVA identified emerging pressure south of the N4 highway, where small-scale mining and chrome washing are moving towards the Magaliesberg Protected Natural Environment (MPNE) boundary. Several hotspots appeared in locations that remained categorically unchanged, indicating early vegetation degradation that classification alone could not detect. Overall, the categorical and continuous results provide a spatial evidence base that shows where pressure is intensifying and where protections remain effective. These insights support more targeted zoning management and identify priority areas for monitoring, enforcement and conservation action. The study demonstrates the value of combining supervised classification with continuous spectral analysis in biosphere landscapes for long-term monitoring of the MB.