Spatio‐temporal variations of wildfire in Hwange National Park, Zimbabwe 2002–2022

Abstract

Knowledge of fire frequency and burnt areas at site‐explicit scales is essential for establishing long‐term fire management strategies in protected areas (PAs). To the best of our knowledge, this research serves as a first attempt to quantify the spatio‐temporal wildfire variability among management sections in Hwange National Park (HNP, with a total area of 14,651 km 2 ). The study aimed to assess fire frequency variability and burnt area (in km 2 ) trends across HNP management sections (Main Camp, Sinamatella, Robins and Makona). MODIS Version 6‐derived fire data were used to assess wildfire variations in HNP from 2002 to 2022. Kruskal–Wallis and Chi‐square tests were used to analyse wildfire frequency and burnt area variations between the four HNP management sections. In addition, the Pearson's correlation analysis was employed to establish the relationship of MODIS land cover, as estimated by the Normalised Difference Vegetation Index (NDVI), Global Precipitation Measurements, topography and land surface temperature (LST) with wildfire‐burnt areas. Mann–Kendall and Sen's slope tests were used to assess the temporal dynamics of per‐year burnt areas, land cover/use types, annual precipitation and LST. There was a significant difference in fire frequency across different management sections ( χ 2  = 111.8, df = 3, p  < 0.001), with Robins experiencing the highest fire frequency, followed by Sinamatella, while Makona had the lowest. Across the study period, approximately 37% of HNP experienced at least a single fire, with Robins having the largest proportional burnt area (73%) and Makona having the smallest (18%). Although there was no significant trend in total burnt area in HNP, areas close to the park boundary were observed to be associated with more recurrent fires. Over the study period, grasslands expanded by 0.31%/year while woodlands and shrublands have decreased by 0.03%/year and 0.29%/year, respectively. Findings showed a significant positive dependency of wildfire with precipitation and LST ( p  < 0.05). Statistically significant negative correlation was also observed between LST and NDVI. Synthesis and applications . The relationship between LST, precipitation and wildfire incidents suggests the role of climate in shaping wildfire dynamics. Spatio‐temporal trends from remotely sensed data promote spatial resource prioritisation for management interventions. Large savannah PAs should not be viewed as homogeneous landscapes for effective wildfire mitigation.