Fire is an important process that shapes the structure and functioning of African savanna ecosystems, and frequently occurs as either prescribed burns or unintentional wildfires in protected areas. Though the level of understanding of the ecological effects of fires has grown substantially over the past century, comprehensive information on the practical application of fire is still restricted, and management information is scattered. Similarly, an improved understanding of how fire affects African mammals is important for the management of both fire regimes and mammal populations. This is also the case in Majete Wildlife Reserve (MWR), Malawi, where a lack of understanding of the past occurrence, determinants, features and effects of prevailing fire regimes prevents the development of appropriate fire management policies. Two separate reviews were conducted to describe the approaches to, and goals of, fire management in African savanna protected areas, as well as the response of large (>5 kg) mammals to fire. For MWR, combinations of remote-sensing and on-the-ground surveys were used to develop a spatially-explicit dataset of the recent fire regime (2001-2019), and to classify, describe and map the woody plant communities present. Additionally, the effect of long-term fire frequencies on vegetation composition, woody plant structure, and large mammal assemblages were assessed, as well as the immediate post-fire habitat selection of large herbivores in a comparative burnt and unburnt landscape. For protected areas, fifteen distinct fire management practices, used to achieve 10 broad ecological (e.g. reversing woody encroachment) and nonecological (e.g. protecting infrastructure) goals, were identified. Additionally, the responses of 51 mammal species to fire were identified, showing that body size was strongly correlated with fire response, with smaller grazers more likely to respond positively to fire than larger browsers. In MWR, it was found that frequent fires dominate the landscape, with ~57% of MWR burning at intervals of two years or less, and an additional ~30% burning at intervals of 3-5 years. A current mismatch between intended fire management goals and actual trends was also highlighted. Five distinct woody plant communities, two of which were subdivided into three sub-communities each, were recognised, along with 118 woody species identified. Fire frequency had little effect on woody plant community composition, but did affect grass species composition. Mammal species clearly selected for either frequently-burnt or infrequently-burnt areas. Clear selection for either burnt (e.g. impala and warthog) or unburnt (e.g. elephant and bushbuck) habitats, that were unrelated to the availability of above-ground herbaceous biomass, were also shown post-fire. This information is intended to provide a basis for improved fire management planning and policy development, as well as providing a baseline against which to monitor change. Managers should re-evaluate fire policies based on these findings, setting clearly defined targets for the different vegetation types, and introducing flexibility in fire regimes to accommodate natural variation. Establishing a mosaic of patches exposed to different fire frequencies, intensities, seasons and sizes will likely be needed to create a range of habitat types that would best allow for the persistence of all facets of biodiversity in MWR.
