Predicted soil bacterial diversity across Sub-Saharan Africa: spatially continuous raster predictions of Hill diversity indices and evenness at 0.05° resolution

This dataset contains spatially continuous raster predictions of soil bacterial diversity across Sub-Saharan Africa, derived from Bayesian hierarchical models fitted to 16S rRNA amplicon sequencing data from 810 soil samples spanning 9 countries (Benin, Botswana, Côte d'Ivoire, Kenya, Mozambique, Namibia, South Africa, Zambia, and Zimbabwe), reprocessed from BioProject PRJNA807934. Diversity is quantified using Hill numbers (q = 0, q = 1, q = 2) and log-evenness (logE). Each index is provided as two layers: the posterior predictive mean and the 90% posterior predictive interval (PPI), reflecting full model uncertainty. Files: pred_mean_D0.tif: Mean predicted species richness (q = 0, ASVs) pred_mean_D1.tif: Mean predicted Shannon diversity (q = 1) pred_mean_D2.tif: Mean predicted Simpson diversity (q = 2) pred_mean_logE.tif: Mean predicted log-evenness pred_mean_D0_even.tif: Mean predicted richness-based evenness pred_ppi90_D0.tif: 90% PPI half-width, species richness pred_ppi90_D1.tif: 90% PPI half-width, Shannon diversity pred_ppi90_D2.tif: 90% PPI half-width, Simpson diversity pred_ppi90_logE.tif: 90% PPI half-width, log-evenness pred_ppi90_D0_even.tif: 90% PPI half-width, richness-based evenness Raster specifications:  CRS: EPSG:4326 (WGS 84)  Resolution : 0.05° (~5 km at the equator)  Format: GeoTiff  Coverage: Sub-Saharan Africa The underlying statistical model uses a Student-t likelihood Bayesian hierarchical model with country-level random effects, fitted in Stan via CmdStanR. Predictors include clay content, aridity, soil organic carbon, pH, and vegetation greenness (MODIS EVI2). These data support the Africa Soil Diversity Atlas (ediman-sda.share.connect.po…) and the associated manuscript currently under review. Source sequencing data: NCBI BioProject PRJNA807934.