Abstract
Food webs represent an important nexus between biodiversity, ecological communities and ecosystem functioning, and they could be radically changed by anthropogenic climate change and habitat loss. Considering changes in food webs around the world has been limited by data availability, we systematically amassed information about species’ diet, traits, distributions, habitat use, and phylogenetics and used machine learning to predict changes in global and local meta-food webs of terrestrial vertebrates. Terrestrial vertebrate food webs are expected to be greatly changed by the end of this century. Future food webs will be smaller and more densely connected, with systematic declines in predator generality and diversity across trophic groups. Decreases in food-web modularity and connectance will make terrestrial vertebrate communities vulnerable to the propagation of extinctions. Increased species' dispersal could ameliorate these trends. Overall, terrestrial food webs will face increased risks of collapse from climate and land-use change, posing significant challenges to biodiversity conservation.