Spatial configuration of severely burned patches within megafires explains ecosystem resilience

It is becoming increasingly important to understand how ecosystems will recover from wildfires, which are increasing in frequency, severity and size, especially in coniferous forests. Megafires—defined as wildfires burning exceptionally large areas—are thought to have more negative effects on ecosystems than smaller fires. However, the effects of megafires vary substantially, and one hypothesis is that intra-fire heterogeneity of burn patches can dictate the recovery of ecosystems. We evaluated the role of spatial configuration of burn patches within megafires using remote sensing data of fires and vegetation at 30x30 m resolution across 36 years and field-survey data of forest recovery in the western USA. Megafires contributed 62% of total burned area, with their frequency explaining 83% of the variation in the inter-annual burned area from 1984-2020. However, megafire size alone did not inherently result in severe ecosystem transitions, with megafires that experienced large contiguous patches of severely burned forest taking longer to recover. Field surveys illustrated delayed recovery resulted from a tree dispersal-limitation threshold of ca. 150 m, such that increasing distance from intact coniferous forest significantly delayed recovery. Machine learning image classification revealed that the rate of recovery in the severely burned areas has declined by ca. 50% from 1984-2020, with distance from seed source being more important than all climate variables analysed. Consequently, spatial configuration of high-severity burn patches within fires—which have become both larger and more compact through time—are key for assessing the effect of megafires on forest resilience.