Animals play important roles as seed dispersers in tropical forest ecosystems and so the pervasive decline of animal populations (“defaunation”) will lead to cascading effects on tree populations, ecosystem structure and carbon storage. However, predicting the impacts of defaunation on seed dispersal failure remains a challenge due to the paucity of ecological interaction information globally. This project aims to fill these gaps through the development of a global database of seed dispersal information and machine learning interaction models.