Vegetation fires are known to profoundly impact ecosystem structure and composition, posing threats to ecosystem stability and human safety. In Zimbabwe, uncontrolled fires have been recurrent, yet a rigorous analysis of the key drivers is still lacking. Previous studies in Zimbabwe have predominantly focused on spatio-temporal dynamics of the occurrence of vegetation fire, leaving a gap in understanding the underlying drivers. Accurate prediction of fire occurrence and identification of the major drivers is imperative for effective fire management strategies. The study employs the Maxent model, a machine-learning approach, to analyze historical MODIS fire data alongside bioclimatic, topographic, anthropogenic, and vegetation variables, to assess the likelihood of fire occurrence in Zimbabwe. The research also aims to elucidate the major factors that influence fire occurrence within the region. The independent contributions of predictor variables to the model's goodness of fit are evaluated using a jackknife test, while model accuracy is assessed using the AUC (area under the receiver operating characteristic curve). Results indicate that elevation, precipitation seasonality, temperature annual range and human footprint emerge as the major factors influencing fire occurrence in Zimbabwe. The model demonstrates an acceptable accuracy, with an average AUC of 0.77. This study underscores the utility of the Maxent model in elucidating the contributions of various environmental factors to vegetation fire occurrence. Moreover, the ability of the model to predict the probability of fire occurrence offers valuable insights for fire managers, facilitating the assessment of the spatial vulnerability of vegetation to fire occurrence. Overall, this research contributes to an improved understanding of the drivers of vegetation fires in Zimbabwe and provides a practical tool for enhancing fire management efforts in the region and beyond.
Read full abstract