An open-source method for spatially and temporally explicit herbivory monitoring in semi-arid savannas

Effective management of protected areas is crucial for addressing the global biodiversity crisis. In water-limited savannas, altered herbivory regimes contribute to ecosystem degradation, creating a need for tools to track herbivore impact on vegetation. Here, we present the Spatially and Temporally Explicit Herbivory Monitoring (STEHM) tool, a novel methodology for monitoring herbivory pressure in near real-time. This approach combines herbivore abundance estimates, derived using a detection algorithm (YOLO v10) performed on imagery collected with camera traps placed at waterpoints across protected areas, with satellite imagery (Sentinel-2) to classify vegetation cover. STEHM enables weekly herbivory assessments, facilitating adaptive herbivore management at scales down to a few square kilometers. By linking herbivore dynamics to surface water availability-a primary factor influencing large herbivore distributions-STEHM provides a framework to disentangle ecological drivers of plant-herbivore interactions. Over one year, we collected and applied STEHM to a total of 2,275,309 individual camera trap images across two case study sites in northern Namibia, leading to the detection of 100,826 waterpoint visits by ten focal species. These observations revealed consistent differences in herbivory pressure between waterpoints, with some areas experiencing concentrated pressure, and a seasonal decline in less water-dependent species during the rainy season, while water-dependent species remained present. Findings indicate that water availability manipulation can alleviate pressure in high-impact areas as non-selective grazers shift to other waterpoints. This refined monitoring capability supports adaptive conservation strategies, providing spatially explicit, near real-time data on herbivore densities to enable targeted management and promote savanna restoration.