A new tropical savanna PFT, variable root growth and fire improve Cerrado vegetation dynamics simulations in a Dynamic Global Vegetation Model

Abstract. The Cerrado, South America's second largest biome, has been historically underrepresented in Dynamic Global Vegetation Models (DGVMs). Therefore, this study introduces a novel Plant Functional Type (PFT) tailored to the Cerrado biome into the DGVM LPJmL-VR-SPITFIRE. The parametrization of the new PFT, called a Tropical Broadleaved Savanna tree (TrBS), integrates key ecological traits of Cerrado trees, including specific allometric relationships, wood density, specific leaf area (SLA), deep-rooting strategies, and fire-adaptive characteristics. The inclusion of TrBS in LPJmL-VR-SPITFIRE led to notable improvements in simulated vegetation distribution. TrBS became dominant across Brazil's savanna regions, particularly in the Cerrado and Pantanal. The model also better reproduced the above- and belowground biomass patterns, accurately reflecting the “inverted forest” structure of the Cerrado, characterized by a substantial investment in root systems. Moreover, the presence of TrBS improved the simulation of fire dynamics, increasing estimates of burned area and yielding seasonal fire patterns more consistent with observational data. Model validation confirmed the enhanced performance of the model with the new PFT in capturing vegetation structure and fire regimes in Brazil. Additionally, a global-scale test demonstrated reasonable alignment between the simulated and observed global distribution of savannas. In summary, the integration of the TrBS PFT marks a critical advancement for LPJmL-VR-SPITFIRE, offering a more robust framework for investigating the interaction of above- with belowground ecological processes, fire disturbance and the impacts of climate change across the Cerrado and other tropical savanna ecosystems that together account for approximately 30 % of the primary production of all terrestrial vegetation.