Data and code from: Increasing rainfall reorganizes fire and herbivores in savannas

This dataset contains data supporting the study “Increasing rainfall reorganizes fire and herbivores in savannas.” The study examines how spatial and temporal variation in rainfall influences fire regimes, wildebeest space use, grass fuels, and the relative importance of fire and herbivory as consumers of herbaceous biomass across the Serengeti Ecosystem in Tanzania. The repository integrates remotely sensed rainfall, fire, and vegetation data with weather-station observations, GPS-collar–derived estimates of wildebeest space use, derived Snapshot Serengeti camera-trap summaries, and field measurements of grass fuel moisture. Analyses focus on three regions—the south-east, central, and north-west—that span the Serengeti rainfall gradient. These data are used to evaluate changes in rainfall amount and variability, burned area, fire size and frequency, wildebeest distribution, and interactions among rainfall, herbivory and fire. Deposited materials include annual and monthly fire and rainfall datasets; weather-station observations; derived wildebeest proportional-use and camera-trap datasets; grass biomass and fuel-moisture measurements; model outputs; and derived spatial raster products. Analyses were conducted in R version 4.5.1. Derived datasets are provided where processing of the original data is computationally intensive or where source data cannot be redistributed. Related Data Sources Climate Hazards Center InfraRed Precipitation with Station data (CHIRPS) – remotely sensed precipitation data used for rainfall analyses. MODIS – remotely sensed products used for vegetation and fire-related analyses. Snapshot Serengeti – camera-trap observations used to characterize herbivore occurrence and abundance. Available here: snapshotserengeti.org Movebank wildebeest GPS-collar data – animal movement data used to characterize wildebeest spatial use. Available here: movebank.org Archibald & Hempson (2016) – continental fire–herbivory–rainfall relationships used to derive prediction curves. Original source observations are available in Archibald and Hempson (2016).