Citation: Camacho, C., P. Saez-Gomez, J. Potti, and J. M. Fedriani. 2017. Nightjars, rabbits, and foxes interact on unpaved roads: spatial use of a secondary prey in a shared-predator system. Ecosphere 8(1):e01611. 10.1002/ecs2.1611Abstract. Linear developments, such as roads and firebreaks, can increase encounter rates between predator and prey, which could affect predator-prey interactions and community dynamics. However, the extent to which prey responses at the interface between natural and anthropogenic habitats may be compared to those at the interface between natural habitats is unclear. Here, we used a shared-predator system to investigate the spatial response of red-necked nightjars (Caprimulgus ruficollis) to changing predation risk on roads, measured as the abundance of red foxes (Vulpes vulpes), and their primary prey (rabbits, Oryctolagus cuniculus). Because all three species coexist closely on unpaved roads in Doñana National Park (Spain), we predicted that nightjars would experience increased predation risk during periods of high fox and low rabbit abundances. Birds could then modify their space use at a broad scale by moving away from risky unpaved roads or, at a finer scale, by seeking foraging microsites facilitating escape from attacks. Between 2011 and 2012, mean rabbit abundance on roads increased by 50%, and fox abundance decreased by 80%, indicating a substantial decrease in predation risk for nightjars. Unexpectedly, nightjar occurrence on roads did not increase as a consequence of the decrease in fox predation risk. However, nightjars foraging on roads became less apprehensive in their use of linear strips of roadside cover, which is known to function as a physical barrier against fox attacks. Specifically, under high predation risk, most nightjars perched on the ground nearby (<15 cm) tall (>150 cm) vegetation, whereas when predation risk decreased, they shifted to more exposed microsites near shorter (<1 m) stands, but rarely close to cover (>45 cm). Nightjars' preference for areas of high predator abundance strongly suggests that flexible microhabitat selection allows them to manage the overall predation risk independently of predator abundance. Our results highlight the importance of linear developments in determining risk exposure and prey use of apparently dangerous habitats and thus may contribute to a better understanding of risky behaviors of prey.