Directional centroid trajectories reveal shifting fire activity across Brazilian biomes

Abstract Understanding how fire activity reorganizes across space is key to interpreting long-term interactions among climate variability, land-use change, and integrated fire management in Brazil. Yet this reorganization is difficult to read directly: annual maps of millions of burn scars saturate the national territory, and year-to-year spatial shifts are visually indistinguishable at that scale. We examine this as a problem of legibility, using annual burned-area centroids as a compact, macro-scale summary of the burned-area distribution and its displacement over time. Analyzing MapBiomas Fire Collection 4 from 1985 to 2024, we distinguish fires in Natural Vegetation from those in Anthropogenic Land Use, quantify the spatial separation between these contexts, and decompose national centroid trajectories into biome contributions and two interannual mechanisms: biome reweighting (changes in each biome’s share of burned area) and within-biome migration (shifts in biome centroids). Across four decades, the two contexts form distinct centroid clouds separated by a mean of 308.2 km (95% CI 304.6–312.9; range 73–499 km); a within-year permutation test rejects spatial coincidence far beyond chance, and the contexts are strongly separable along the Cerrado–Amazon axis (AUC = 0.904, 95% CI 0.845–0.958). Biome decomposition uncovers a consistent structure dominated by the Cerrado–Amazon gradient, with the Cerrado anchoring the national centroid under Natural Vegetation and the Amazon dominating under Anthropogenic Land Use. Interannual shifts show contributions from reweighting and migration, which frequently partially cancel, and context separation projects largely, though not exclusively, onto the Cerrado–Amazon axis. By reducing millions of scars to interpretable trajectories with quantified uncertainty, this framework makes a persistent biome-scale organization, invisible in the raw distribution, legible, and links national fire patterns to conservation-relevant contrasts between fire-dependent and fire-sensitive systems.