Infection risk varies within urbanized landscapes: the case of coyotes and heartworm

Background: Urbanization can have profound effects on ecological interactions. For host–pathogen interactions, differences have been detected between urban and non-urban landscapes. However, host–pathogen interactions may also differ within highly heterogeneous, urbanized landscapes. Methods: We investigated differences in infection risk (i.e., probability of infection) within urbanized landscapes using the coyote (Canis latrans) and mosquito-borne nematode, Dirofilaria immitis (the causative agent for canine heartworm), as a case study. We focused on a coyote population in Chicago for which extensive behavioral and heartworm infection data has been collected between 2001 and 2016. Our objectives were to: (i) determine how onset and duration of the heartworm transmission season varied over the 16-year period and across the urban–suburban gradient; and (ii) investigate how heartworm infection risk in coyotes varied over the years, across the urban–suburban gradient, by coyote characteristics (e.g., age, sex, resident status), and coyote use of the urbanized landscape (e.g., use of urban areas, mosquito habitats). Results: While onset of the heartworm transmission season differed neither by year nor across the urban–suburban gradient, it was longer closer to the core of Chicago. Of the 315 coyotes sampled, 31.1% were infected with D. immitis. Older coyotes and coyotes sampled in later years (i.e., 2012–2016) were more likely to have heartworm. While coyote location in the urban–suburban gradient was not a significant predictor of infection, the proportion of urban land in coyote home ranges was. Importantly, the size and direction of this association varied by age class. For adults and pups, infection risk declined with urbanization, whereas for subadults it increased. Further, models had a higher predictive power when focusing on resident coyotes (and excluding transient coyotes). The proportion of mosquito habitat in coyote home ranges was not a significant predictor of infection. Conclusions: Our findings suggest that urbanization may affect host exposure to vectors of D. immitis, that risk of infection can vary within urbanized landscapes, and that urbanization–wildlife infection associations may only be detected for animals with certain characteristics (e.g., age class and resident status). Graphical abstract: [Figure not available: see fulltext.].