Abstract:
The Niassa National Reserve (RNN) has been registering an increase in the incidence rates of anthropogenic fires, which has led to the implementation of a fire management plan. This work aimed to study the burning regime of the RNN miombo ecosystem as a basis for preparing the burning management plan. MODIS images (2000-2012) of active fires and burnt areas were used from which pixels were extracted with confidence levels equal to or greater than 80% and 1-366 Julian days, respectively, using Erdas 9.0 and ArcGis 9.3. With the data, the frequency, return interval (IRQ), seasonality, intensity, density and extent of the burned area were calculated. Subsequently, logistic regression analyzes were performed in SAS 9.3 to assess the relationship between the type of land use and cover, precipitation, temperature, altitude, elephant density, human population density, distance in relation to access roads and the frequency of fires to determine the factors that influence the occurrence of fires in the RNN. Field data was collected in 108 circular plots of 15m radius to reinforce the accuracy of remote sensing analyzes. The results showed that on average the fire returns to a particular place every 3.29 years at a frequency of 0.36 times / year with a peak in the dry season. Approximately 43% of the total area of the RNN was burned every 1-2 years and the Central-North and East regions recorded more frequent fires, being more intense in the areas of low frequency and density of fires. The occurrence of fires is favored by the type of land cover, density of elephants and temperature and disadvantaged by the increase in population density. Thus, priority areas for the control of burnings in the RNN are those whose IRQ is equal to or less than 2 years, dominated by deciduous forests, with more than 1 elephant / Km 2 and low population density, including inselbergs and dambos (TRADUÇÃO NOSSA)
