Koutsias N., Panourgiα K., Nakas G., Petanidou T. (2024). The importance of landscape and fire-history as factors explaining short- and medium-term post-fire vegetation recovery in a Mediterranean island using Sentinel-2 satellite data. Science of the Total Environment 957: 177443 https://doi.org/10.1016/j.scitotenv.2024.177443

Fire is a natural and human-induced process which under specific conditions can be considered a critical disturbance in Mediterranean ecosystems, that shapes their development and ecology. This study aims to monitor post-fire vegetation recovery using remote sensing methodologies, more specifically, spectral indices of Sentinel-2 satellite data as proxies to burn severity and post-fire recovery. Additionally, we considered other underlying factors such as times burned, topographic variables (elevation, slope, aspect), and prior land cover types. Spanning from 2017 to 2021, the research focuses on the southern region of Chios Island, Greece, which has experienced two fires in close succession, viz. in 2012 and 2016. Negative binomial generalized linear models (GLM) and generalized additive models (GAM) were employed to assess the factors impacting vegetation recovery. Interestingly, times burned was found to have no significant effect on vegetation regeneration, with areas burned once (in 2016) showing no significant differences in regeneration rate vs. those burned twice (in 2012 and 2016). The analysis revealed a distinct pattern of regeneration, with the initial post-fire year displaying a different pattern of recovery compared to subsequent years. Environmental factors such as elevation, slope, and aspect, as well as prior land cover types, emerged as significant influencers of regeneration rates only for the first post-fire year of recovery. These factors were replaced by elevation and burn severity, together with prior land cover types as underlying explanatory variables in the fifth year of post-fire recovery, suggesting varying effects over time. Elevation and burn severity showed non-linear relationships, as justified by the choice of a GAM smoother for both variables. High severity areas demonstrated greater recovery, possibly due to increased space for regeneration.