Burned area trends in Niassa Special Reserve, Mozambique, are mostly driven by late dry season fires and weakly dependent on climate

Fire plays a critical ecological role in African savannas yet shifting fire regimes driven by climate change and anthropogenic pressures raise concerns over their ecological and socio-economic consequences. This study provides the first spatially explicit and seasonally disaggregated analysis of burned area (BA) trends in the Niassa Special Reserve (NSR), Mozambique, from 2000 to 2023. Burned-area maps derived from Landsat imagery were analysed on a 5-km grid using the Theil–Sen slope estimator and the Contextual Mann–Kendall (CMK) test to detect statistically significant trends during the full dry season (FDS), early dry season (EDS), and late dry season (LDS) periods. In clusters of cells with significant BA trends, we related BA interannual variability to Canadian Fire Weather Index (FWI) components—the Daily Severity Rating (DSR), Fine Fuel Moisture Content (FFMC), and Initial Spread Index (ISI)—as well as MODIS-derived net primary productivity (NPP) as a proxy for antecedent fuel availability. Most of the NSR exhibited no significant BA trends, but coherent spatial clusters were detected, particularly of LDS BA decrease. Relationships between BA and fire-weather indices and NPP were generally weak. Only one LDS cluster showed a significant BA–ISI relationship, and antecedent NPP was not significantly related to BA. Residual trend analysis indicated that many BA patterns evolved independently of climatic or fuel productivity drivers, underscoring the importance of other factors, most likely of anthropogenic origin. These findings highlight the need for fine-scale, seasonally resolved fire monitoring to inform adaptive fire governance in protected African savanna landscapes. • High spatial-resolution trends in burned area assessed across seasonal fire periods. • Largest fire declines occur in Late Dry Season (LDS) and mirror annual trend clusters. • Early Dry Season (EDS) fire increases are localized, possibly tied to early burning for management. • Burned-area trends show only sparse and weak associations with fire-weather indices (DSR, FFMC, ISI) and antecedent fuel productivity (MODIS NPP). • Residual trend analysis reveals dominant role of non-climatic fire drivers.