Seasonal spatial-temporal trends of vegetation recovery in burned areas across Africa.

Africa is entering a new fire paradigm, with climate change and increasing anthropogenic pressure shifting the patterns of frequency and severity. Thus, it is crucial to use available information and technologies to understand vegetation dynamics during the post-fire recovery processes. The main objective of this study was to evaluate the seasonal spatio-temporal trends of vegetation recovery in response to fires across Africa, from 2001 to 2020. Non-parametric tests were used to analyze MODIS Normalized Difference Vegetation Index (NDVI) products comparing the following three-month seasonal periods: December-February (DJF), March-May (MAM), June-August (JJA), and September-November (SON). We evaluated the seasonal spatial trends of NDVI in burned areas by hemisphere, territory, or country, and by land cover types, and fire recurrences, with a focus on forested areas. The relationships between the seasonal spatial trend and three climatic variables (i.e. maximum air temperature, precipitation, and vapor pressure deficit) were then analyzed. For the 8.7 million km2 burned in Africa over the past 22 years, we observed several seasonal spatial trends of NDVI. The highest proportions of areas with increasing trend (p < 0.05) was recorded in MAM for both hemispheres, with 22.0% in the Northern Hemisphere and 17.4% in the Southern Hemisphere. In contrast, areas with decreasing trends (p < 0.05), showed 4.8-5.5% of burned area in the Northern Hemisphere, peaking in JJA, while the Southern Hemisphere showed a range of 7.1 to 10.9% with the highest proportion also in JJA. Regarding land cover types, 48.0% of fires occurred in forests, 24.1% in shrublands, 16.6% in agricultural fields, and 8.9% in grasslands/savannas. Consistent with the overall trend, the area exhibiting an increasing trend in NDVI values (p < 0.05) within forested regions had the highest proportion in MAM, with 19.9% in the Northern Hemisphere and 20.6% in the Southern Hemisphere. Conversely, the largest decreasing trend (p < 0.05) was observed in DJF in the Northern Hemisphere (2.7-2.9%) and in JJA in the Southern Hemisphere (7.2-10.4%). Seasonally, we found a high variability of regeneration trends of forested areas based on fire recurrences. In addition, we found that of the three climatic variables, increasing vapor pressure deficit values were more related to decreasing NDVI levels. These results indicate a strong component of seasonality with respect to fires, trends of vegetation increase or decrease in the different vegetation covers of the African continent, and they contribute to the understanding of climatic conditions that contribute to vegetation recovery. This information is helpful for researchers and decision makers to act on specific sites during restoration processes.