Environmental DNA (eDNA) is used for biodiversity assessments in a variety of ecosystems across the globe, whereby different eDNA concentration, preservation and extraction methods can outperform others depending on the sampling conditions and environment. Tropical and subtropical ecosystems in Africa are among the less studied systems concerning eDNA-based monitoring. Waterholes in arid parts of southern Africa represent important agglomeration points for terrestrial mammals, and the eDNA shed into such waterbodies provides a powerful source of information for monitoring mammalian biodiversity in the surrounding area. However, the applied methods for eDNA sampling, preservation and filtering in different freshwater systems vary greatly, and rigorous protocol testing in African freshwater systems is still lacking. This study represents the first attempt to examine variations in eDNA concentration, preservation and extraction methods under remote field conditions using waterborne eDNA in a savanna system. Collected samples were heavily affected by microalgal and bacterial growth, impeding eDNA capture and PCR success. We demonstrate clear effects of the methodological choices, which also depend on the state of eDNA. A preliminary metabarcoding run showed little taxonomic overlap in mammal species detection between two metabarcoding primers tested. We recommend water filtering (using filters with pore sizes >1 μm) over centrifugation for eDNA concentration, Longmire's solution for ambient temperature sample preservation and Qiagen's DNeasy PowerSoil Pro Kit for DNA extraction of these inhibitor-prone samples. Furthermore, at least two independent metabarcoding markers should be utilized in order to maximize species detections in metabarcoding studies.