In recent decades the ‘landscape of fear’ has grown in popularity to become a central consideration in wildlife management, and has even been reconceptualized as the ‘landscape of coexistence’ for understanding human- wildlife conflicts such as predator attacks on livestock. Yet fear effects are not always the predominant driver of predator-prey interactions. Thus, guiding ecological principles have not been assembled to explain the broader food web interactions that shape the context dependency of carnivore-livestock conflict. We address this gap by developing a conceptual framework as a way to think about the contingencies under which inducing non- consumptive ‘fear effects’ on predators would be effective to mitigate carnivore-livestock conflict. The frame- work specifically considers interactions among wildlife (carnivore predators, wild ungulate prey) and humans (people and livestock) in terms of spatial predator-prey assemblages in which the nature of wildlife-human interactions - as either a carnivore-livestock conflict or a coexistence food web - is contingent on the nature of spatial movement and overlap of humans and wildlife across landscapes. Considering human-wildlife interac- tions within such a spatial food web context can assist in enabling people and wildlife, especially imperiled carnivores, to coexist in human-modified landscapes. The framework offers predictions that should be tested via adaptive management experiments that evaluate whether conflict mitigation solutions aligned with particular spatial human-livestock-carnivore contexts do indeed resolve conflict