Host-associated microorganisms are critical for eukaryotic host functioning. Most studies to date have focused on observational approaches or have used model systems to understand host-microbiome interactions. To progress our understanding of host-microbiome interactions, causal relationships must be determined within natural ecosystems. By using a dominant seaweed host, we demonstrate an experimental framework where host-microbiome interactions can be empirically tested through a combination of microbial manipulations. Four antibiotic treatments were employed to disrupt seaweed-associated microbial communities and test the hypothesis that resulting disruption to the host-microbiome would negatively affect host-performance. Disruption to the microbial communities resulted in significant differences in both alpha and beta diversity which was directly linked with poor host condition. To establish a causal link between specific microbial taxa and host condition, two bacterial taxa correlating with poor host-performance were isolated and used in subsequent inoculation experiments, independently or in combination with microbiome disruptions. When disruption was combined with inoculations the effect on host condition was strongest and lasted the entire experimental period whereas within all other treatments host condition recovered. This experimental framework allows for causal relationships to be determined within natural systems, allowing researchers to better predict how these ecosystems will respond to future environmental disruption.