Our understanding of pathogen spillover is largely based on viral systems associated with bats. Bacterial infections of zoonotic origin also pose a significant public health burden, many of which are associated with small mammals like rodents that live abundantly among humans. When ecologically different species of small mammals are brought in contact with each other due to various factors, including habitat modification, there is potential for bacterial spillover within the host community and subsequently to humans. Small mammals and their ectoparasite communities present a unique system to investigate the eco-epidemiology of multi-host pathogens, which can inform specific bacterial spillover determinants and their interplay. In a rainforest-human-use mosaic, we applied ecological and evolutionary analysis to investigateBartonellaspp. across small mammal and ectoparasite communities.We observed substantial overlap among small mammal communities in different habitat types, predominantly driven by habitat generalists. Most ectoparasites were generalists, infecting multiple hosts, and several host species displayed polyparasitism, i.e., they were infected by multiple ectoparasites. We observed highBartonellaprevalence at both study sites – a forest-plantation mosaic (47.4%) and a protected area (28.8%). Seven of the ten ectoparasite morphotypes sampled were also positive forBartonella, following the prevalence trend in their hosts. A Generalised Linear Model (GLM) revealed an independent association between aggregated ectoparasite load in species andBartonellaprevalence, implicating ectoparasites in transmission.Bartonellalineages from small mammals were host-specific, while ectoparasites showed no host-specific associations. They carriedBartonellaassociated with other hosts, indicating the potential for cross-species transmission. Phylogenetic ancestral trait reconstruction ofBartonellahaplotypes suggests historic spillover events in the small mammal community, validating the potential for contemporary spillover. These results highlight infection risk from such underdiagnosed agents in the landscape, and the necessity to disentangle complex multi-host multi-vector systems to understand their epidemiological consequences.