The firewall between Cerrado and Amazonia: Interaction of temperature and fire govern seed recruitment in a Neotropical savanna

Questions Models predicting the distribution of savannas worldwide have identified rainfall and fire as their primary determinants. However, most of them have relied upon adult traits, while juvenile traits, at the bottleneck of the plant's life cycle, have been largely overlooked. We developed a novel mechanistic model based on the effects of temperature and fire on germination traits to predict the distribution of Cerrado, i.e., the largest neotropical savanna. Location Cerrado and neighboring biomes. Methods We compiled data on the germination of seeds subjected to temperature and heat shock treatments and used generalized additive mixed models to predict germination potential as a function of temperature, species, physiognomy (forest/savanna), habits (herbs/shrubs/trees), and fire frequency. Results The best model showed that seasonal temperatures set the germination limits for seeds of both savanna and forest physiognomies. Forest seeds presented a higher germinability in the optimum temperature range, but savanna seeds had higher survival rates after heat shocks. The model revealed that the southern limit of Cerrado is determined by low winter temperatures, while the western and eastern boundaries are set by high summer temperatures. The model also predicted an area of high germination potential that coincides with high biodiversity and climate stability in the Cerrado. Conclusions Germination traits are highly valuable to predict vegetation responses to climate. Seasonal temperatures are primary determinants of the Cerrado’s extent, while fire favors the recruitment of savanna species over the Cerrado–Amazonia ecotone. Global warming may significantly impact the germination potential of native species. The use of juvenile traits to predict plant distribution has been overlooked in the literature. We developed a model based on germination responses to temperature and heat shock and predicted the extent of the Cerrado biome, the largest neotropical savanna. Our model also revealed that seasonal temperatures and fire frequency highly determine the distribution of vegetations by their effects on recruitment.