Why do some species become extinct while others diversify? What is the role of intrinsic and extrinsic factors in rendering this imbalance? How do the role of species in their ecosystems influence their evolution? These are fundamental questions in evolutionary biology that remains poorly understood. Answering this questions will shed light on the connections between ecological function, extinction, and the persistence of functional systems, which is key to inform current and future conservation policies. The fossil record allows us to track the evolution of species and their ecosystems during million of years. Recently developed methods based on a Bayesian framework give us the fundamental tool to study the speciation and extinction rates of fossil species in relation to abiotic factors. Moreover, by incorporating functional traits, we are able to study the effects of species ecology on extinction risk and diversification potential. The aim of this project is to apply Bayesian models, through the software PyRate, over a large dataset of worldwide Cenozoic large herbivores (more than 22,000 occurrences of 3,000 species). Besides, the dataset includes information of 14 functional traits to study the effect of ecology on their diversification dynamics (speciation and extinction). Thanks to multi-faceted approach, we will be able to study the effect of individual traits and their combination in species persistence and also in the evolution of ancient ecosystems. Also, we will work on the development of new approaches that will combine a Bayesian framework with neuronal networks. Neural networks will allow the creation of trained systems able to predict the presence of non-fossilized taxa or functional types in ancient ecosystems, bringing a more complete depiction of extant communities than the literal reading of the fossil record.