Investigating the patterns of recurrent fires in highly urbanized Mediterranean landscapes

Fire recurrence plays a key role in shaping landscapes in Mediterranean ecosystems. Shortterm recurrent fires are increasingly affecting the landscapes, especially the highly urbanized landscape. Few published studies have addressed fire recurrence by analyzing environmental, climatic and human factors. Current models use fire recurrence polygons as the dependent variable; yet no published study has focused its analysis on fire recurrence considering recurrent ignition points as the dependent variable. To fill this gap, remote sensing and local available data were combined to examine the role of human and biophysical variables in predicting both the likelihood and frequency of recurrent fire ignition points over a nine-year period (2004-2012) in Southern Italy. We used a Negative Binomial Hurdle (NBH) model, owing to the stochastic nature of the phenomenon (fire recurrence), and the available fire dataset characterized by a high number of non-occurrences. Our results suggest that the likelihood and frequency of recurrent fire ignition points (dependent variables) had a negative relationship with population and road density and positive relationship with land-cover variables. Road density was the strongest predictor of recurrent fire ignitions, followed by the presence of shrublands and grasslands. Vegetation indices (NDVI and NDWI) were also good predictors of fire recurrence. More broadly, this study is intended to be a further experimental step in fire-management analysis where constant alterations of the human and natural systems associated with population growth, natural fuels, and global change can create conditions for short-time interval recurrent fires.