The cold black abyss of network prediction in space

Predicting species interactions is a difficult enough challenge, but predicting an entire species interaction network based on species traits, latent features of the network, or interactions in other locations is on another level. Some of these approaches would tacitly assume that species interactions could be present anytime two previously interacting species are present, ignoring the role of species densities, host switching, or environmental effects. Let's make a bunch of assumptions and explore how we do at predicting host-parasite networks across space, eh?