Most species in the bacterial family of Pasteurellaceae colonize one specific host species. Vertebrates of very different evolutionary descent including fish, turtles, marsupials, eutherians and birds are colonized by different members of Pasteurellaceae. This one-to-one microbial-host species partnership makes Pasteurellaceae species valuable candidates to study biodiversity, bacterial-host co-evolution and host adaptation, and their widespread distribution across vertebrates provide the possibility to collect a wide array of data, where wildlife species are essential. However, obtaining samples from wild animals comes with logistic, technical and ethical challenges, and previous microbiota studies have led to the presumption that captive animals are poor models for microbial studies in wildlife. Here, we show that colonization of polar bears by Ursidibacter maritimus is unaffected by factors related to captivity, reflecting a deep symbiotic bond to the host. We argue that the study of ecological and evolutionary principles in captive wildlife is possible for host-adapted taxa such as those in the Pasteurellaceae family. Moreover, studying captive, often trained animals protects wild populations from the stress associated with obtaining samples.