Uncertainty in climate-vegetation feedbacks on fire regimes challenges reliable long-term projections of burn area from correlative models

Recent studies have applied simple correlative models to project an increase in future burnt area (BA) for the Mediterranean region. In one of these studies led by Marco Turco and co-workers in the journal Nature Communications (doi:10.1038/s41467-018-06358-z), the authors relate BA to regional estimates of cumulative drought surrogates derived from evapotranspiration indices (SPEI) and later, they use this relationship to infer changes derived from future climate data. However, estimates of future fire risk suffer from the critical shortcoming that negative feedbacks of climate changes on vegetation (i.e. climate may actually reduce vegetation growth and eventually decrease fire risk) are not included. To overcome this problem, these authors proposed a way around by using regional variability in the BA drought relationship (what they call nonstationary models) to account for future changes on fire regimes derived from climate effects on vegetation. Their analyses showed that sensitivity of fire activity to dry periods is stronger in cooler/productive sites and therefore, they propose to use this finding as a short cut in their BA projections using climate change scenarios. The main assumption behind this approach is that the BA-SPEI relationships under a given productivity gradient can be used to infer new BA-SPEI relationships arising in the future. While representing a step forward in acknowledging the pitfalls of current projections of BA, this short-cut falls short in allowing to account for the key process behind climate-vegetation-fire feedbacks. We argue that there are a series of mechanisms by which current correlations are not likely to be maintained in the future with major, overall still unknown, consequences on BA projections.