Climatic factors influencing vegetation net primary productivity in semi-arid savanna reserves of Limpopo Province, South Africa

Abstract

Savanna ecosystems are important components of the terrestrial carbon cycle, but their vegetation productivity is highly sensitive to climatic variability, especially in semi-arid environments. This study assessed the spatiotemporal dynamics and climatic drivers of vegetation net primary productivity (NPP) across semi-arid savanna reserves in Limpopo Province, South Africa, from 2001 to 2025. Landsat-derived Enhanced Vegetation Index (EVI) was analyzed to characterize long-term vegetation greenness, while MODIS products were used to assess annual NPP and seasonal productivity. Climatic variables, including precipitation, soil moisture, potential evapotranspiration, solar radiation, and temperature, were obtained from CHIRPS, South African Weather Service observations, and ERA5-Land datasets. Geospatial processing was performed using Google Earth Engine and ArcGIS Pro, while Theil–Sen, Mann–Kendall, coefficient of variation, correlation analysis, and Geo-detector were implemented in Google Colab. Results showed a positive annual mean NPP trend of 4.94 g C m -2 yr -1 , despite strong interannual variability. NPP trends were mostly directionally positive but largely statistically insignificant, while 86.72% of the landscape showed high or relatively high stability. Precipitation showed strong interannual variability with a weak upward trend, while temperature increased gradually over the study period. Correlation analysis based on ecological-season climate summaries showed that NPP was positively associated with precipitation and soil moisture and negatively associated with solar radiation, PET, and temperature, highlighting the dominant role of growing-season moisture availability. Temporal-window analysis further indicated stronger recent climate–NPP sensitivity, especially to soil moisture, precipitation, PET, and solar radiation. Geo-detector results revealed strong interactions among climatic and topographic variables, notably slope–precipitation, slope–PET, and slope–temperature. These findings suggest that vegetation productivity in protected semi-arid savanna reserves is influenced by both moisture availability and spatial environmental gradients, offering valuable insights for climate-adaptation planning and savanna ecosystem management.