Emerging infectious diseases (EIDs) threaten biodiversity, yet identifying key host species in complex ecological communities remains a major challenge. Here, we develop a quantitative framework combining field data, epidemiological modelling, simulations, and Bayesian inference to pinpoint key viral hosts in multispecies bee communities. Using flower–visitor interaction data and molecular virus screening, we estimate species-specific basic reproduction numbers (R₀) and assess the role of key hosts in virus transmission and persistence by testing two competing hypotheses: viral spread driven by a key host versus community-wide traits (e.g. network connectance). We show that, while honeybees often act as primary reservoirs, others, such as the red-tailed bumblebee, can drive the spread of specific viruses. Viral dynamics are primarily explained by exposure to key hosts, while the community effects are highly variable. Identification of non-honeybee key hosts challenges existing assumptions and highlights drivers of transmission and pathogen persistence in complex host–pathogen networks.