Data for: Forecasting forest fragmentation across Nigeria's major biomes using Sentinel-2, Random Forest, and CA-Markov

Classified maps (high, low and no vegetation) and Random Forest class probabilities for eight Nigerian forest reserves in the 2019 and 2024 dry seasons; change maps; calibrated stochastic CA-Markov ensembles (2024 hindcast, 2028 and 2033 projections) with run-agreement layers; photo-interpreted reference points and consensus training samples; all article and supplementary tables, landscape metrics for every map, realisation and run; and figures. GeoTIFFs are 10 m, local UTM, and nodata 255. See README.md.