Metabolism operates as a multiscale process that induces interactions between microbial species and leads to division of labor and cross-feeding in complex microbial communities. By combining dynamic flux balance modeling with diffusion equations, we model the spatio-temporal dynamics of communities in complex environments, starting from the genomes of individual species. We envisage that these approaches will be increasingly valuable for understanding natural communities (e.g., ocean and soil consortia), and for guiding targeted synthetic ecology experiments with artificial communities that can be used both as benchmark for our models and as avenues for environmental remediation and metabolic engineering.