Predicting natural hyperdense regeneration after wildfires in Pinus halepensis (Mill.) forests using prefire site factors, forest structure and fire severity

Postfire Pinus halepensis (Aleppo pine) regeneration is often hyperdense. The overstocked stands created by this hyperdense regeneration considerably increase the risk of biotic and abiotic disturbances, especially fires, by increasing the potential for widespread forest losses. Our aim was to understand the relation between prefire site factors (climate, geographical position, topography, soil), prefire forest structure variables and fire severity with regeneration density after fire. We specifically wondered: (1) what are the general drivers of natural regeneration in these forests after fire?; (2) what are the necessary prefire conditions for establishing Aleppo pine hyperdense regenerations (>4,000 plants/ha)? To answer these questions, we sampled 147 plots in 15 wildfires located in the Comunitat Valenciana, which were representative of Aleppo pine Mediterranean forests. We used full and partial redundancy analyses (RDAs) for variance partitioning, and a decision tree analysis to look for the key site factors that drive regeneration density after fire. We found that all the site factors measured in the study explained 34.4 % of total variation in regeneration density. Prefire site factors and fire severity together explained 28.4 % of total variability, while the measured postfire factors explained only 7.5 %. Forest structure and climate explained 8.3 % and 6.7 % of variation, respectively. Five specific site factors drove regeneration density after fire: average minimum temperature, tree density before fire, resprouting shrubs coverage before fire, soil depth and bedrock type. The conclusions of this study were: (i) the average minimum temperature was the main significant variable that classified regeneration density and split data into three significant groups of Aleppo pine burned sites; (ii) the prefire forest structure (overstorey density and understorey coverage) controls regeneration density at colder burned sites, but soil depth and bedrock can be more important at warmer sites; (iii) fire severity relates positively to pine regeneration density, but negatively to resprouting vegetation coverage after fire; (iv) overstocked stands are not expected if prefire stand density is below 100 trees/ha at colder burned sites. These results may facilitate the planning of forest management and restoration actions because it may be used to identify those areas more likely to regenerate overstocked stands when faced with a changing fire regime.This article forms part of Project PROMETEO/2019/110