The landscape of fear (LOF) framework predicts that prey adapt their behavior to mitigate predation risk, yet the framework's expression in resource-limited, unfenced systems remains poorly understood. Across seven large herbivore species in an unfenced arid system, space use is governed by bottom-up constraints while predator risk is buffered in time, producing consistent low diel overlap with nocturnal carnivores and trait-dependent moderation of spatial responses. We used camera trap data from northwest Namibia to examine how five herbivore species (gemsbok Oryx gazella, southern giraffe Giraffa giraffa, greater kudu Tragelaphus strepsiceros, Hartmann's mountain zebra Equus zebra, springbok Antidorcas marsupialis) and two megaherbivores (black rhinoceros Diceros bicornis, African bush elephant Loxodonta africana) navigate bottom-up environmental constraints and top-down predation risk from lions (Panthera leo) and spotted hyenas (Crocuta crocuta). We tested four hypotheses: (1) that bottom-up environmental constraints would explain more variance in herbivore space use than predator covariates; (2) that herbivores would reduce diel overlap (Delta) with nocturnal predators rather than vacating resource patches; (3) that megaherbivores would show minimal spatiotemporal avoidance of predators, relative to smaller herbivores; and (4) that herbivores would respond reactively to recent predator presence, rather than proactively by avoiding areas of chronic predator activity. Our findings support a hierarchical LOF in which bottom-up constraints (dry-season progression, visibility for grazers, habitat structure for browsers) contribute to herbivore space use and top-down predator pressures guide temporal activity: most species exhibited strongly diurnal activity, resulting in low temporal overlap with nocturnal carnivores (Delta approximate to 0.11-0.21 for grazers; Delta approximate to 0.06-0.26 for browsers). By contrast, spatial responses to predator presence were modulated by environmental context, reinforcing the importance of spatiotemporal plasticity. These results advance understandings of predator-prey dynamics, particularly in dryland ecosystems. We suggest refinements to the LOF framework for multi-predator, resource-constrained landscapes.