Combining stand-level and remote sensing data to model post-fire recovery of Mediterranean tree-forest communities – A case study in Spain.

Mediterranean forests are recurrently affected by wildfires. Fire activity is expected to accelerate in the future due to landscape homogenization, fuel accumulation, and climate warming. A key aspect to prevent and mitigate the negative impacts of wildfires on ecosystems is to understand the factors that govern the recovery of forest communities. This study analyzes the post-fire recovery potential of four representative Mediterranean tree-communities (Pinus halepensis, Pinus nigra, Pinus pinaster, and Quercus ilex) affected by large wildfires (> 500 ha) during the summer of 1994 in Spain. For this purpose, information collected in the field 25 years after the fires in 203 forest plots (131 burned and 72 unburned control plots) was coupled with remote sensing, geospatial, and forest inventory data, to build an empirical model capable of assessing recovery. Remote sensing data provided a proxy for burn severity, through the Composite Burn Index, and allowed modelling the local topography (slope and aspect) of the terrain. The geospatial data included climatic information on temperature and precipitation trends. These data were entered into the model, calibrated using Random Forest, to provide information on the degree of recovery, inferred from the similarity (in terms of vegetation height, aboveground biomass, species diversity) between the burned and unburned control plots. Results showed that only the 25% of the burned plots can be considered as recovered. The burn severity had a significant effect on the recovery albeit strongly modulated by local topography. Overall, the key features of the recovered plots were a low-to-moderate burn severity and a favorable topographical setting, especially the shading effect of steep northwestern slopes. Furthermore, a warmer and more humid climate improved the capacity of recovery. These results constitute a valuable tool for improving forest management and preserving ecosystem services.