Tropical forests are an essential component of the earth system and play a critical role for land surface feedbacks to climate change. Uncertainty in the magnitude of the role of these forests on the global carbon cycle is driven by the high spatial heterogeneity of tropical forests, with varying carbon dynamics, forest structure and species composition. Within this project, we hypothesize that by failing to capture this spatial heterogeneity of tropical forests in dynamic global vegetation models, model projections on pantropical carbon dynamics will remain unreliable. We therefore aim to introduce a better model structure and parameterization for tropical forests in the vegetation model ED (Ecosystem Demography model). Specifically, we will introduce innovative continent-specific tropical forest plant functional type (PFT) parameterizations making use of recent datasets on plant traits, forest structure and allometry. Moreover, for several processes and plant traits (including allocation, biomass residence time, leaf nutrient stoichiometry) we will introduce dependency on environmental conditions. Furthermore, large differences in rainfall seasonality across continents drives seasonality in net primary productivity, and it will be investigated if this seasonality is reliably captured by the model. By incorporating tropical heterogeneity at these different levels, we aim to improve simulations of pantropical forest biomass stocks and carbon dynamics.