Because some native ungulates have lived without top predators for generations, it has been uncertain whether runaway predation would occur when predators are newly restored to these systems. We show that landscape features and vegetation, which influence predator detection and capture of prey, shape large-scale patterns of predation in a newly restored predator-prey system. We analysed the spatial distribution of wolf (Canis lupus) predation on elk (Cervus elaphus) on the Northern Range of Yellowstone National Park over 10 consecutive winters. The influence of wolf distribution on kill sites diminished over the course of this study, a result that was likely caused by territorial constraints on wolf distribution. In contrast, landscape factors strongly influenced kill sites, creating distinct hunting grounds and prey refugia. Elk in this newly restored predator-prey system should be able to mediate their risk of predation by movement and habitat selection across a heterogeneous risk landscape.

Environmental change has been observed to generate simultaneous responses in population dynamics, life history, gene frequencies, and morphology in a number of species. But how common are such eco-evolutionary responses to environmental change likely to be? Are they inevitable, or do they require a specific type of change? Can we accurately predict eco-evolutionary responses? We address these questions using theory and data from the study of Yellowstone wolves. We show that environmental change is expected to generate eco-evolutionary change, that changes in the average environment will affect wolves to a greater extent than changes in how variable it is, and that accurate prediction of the consequences of environmental change will probably prove elusive.

A “landscape of fear” (LOF) is a map that describes continuous spatial variation in an animal's perception of predation risk. The relief on this map reflects, for example, places that an animal avoids to minimize risk. Although the LOF concept is a potentially unifying theme in ecology that is often invoked to explain the ecological and conservation significance of fear, little is known about the daily dynamics of an LOF. Despite theory and data to the contrary, investigators often assume, implicitly or explicitly, that an LOF is a static consequence of a predator's mere presence within an ecosystem. We tested the prediction that an LOF in a large‐scale, free‐living system is a highly dynamic map with “peaks” and “valleys” that alternate across the diel (24‐h) cycle in response to daily lulls in predator activity. We did so with extensive data from the case study of Yellowstone elk (Cervus elaphus) and wolves (Canis lupus) that was the original basis for the LOF concept. We quantified the elk LOF, defined here as spatial allocation of time away from risky places and times, across nearly 1,000‐km2 of northern Yellowstone National Park and found that it fluctuated with the crepuscular activity pattern of wolves, enabling elk to use risky places during wolf downtimes. This may help explain evidence that wolf predation risk has no effect on elk stress levels, body condition, pregnancy, or herbivory. The ability of free‐living animals to adaptively allocate habitat use across periods of high and low predator activity within the diel cycle is an underappreciated aspect of animal behavior that helps explain why strong antipredator responses may trigger weak ecological effects, and why an LOF may have less conceptual and practical importance than direct killing.

Knowledge of characteristics that promote group success during intraspecific encounters is key to understanding the adaptive advantages of sociality for many group-living species. In addition, some individuals in a group may be more likely than others to influence intergroup conflicts, a relatively neglected idea in research on social animals. Here we use observations of aggressive interactions between wolf (Canis lupus) packs over an extended period and use pack characteristics to determine which groups had an advantage over their opponents. During 16 years of observation in Yellowstone National Park from 1995 to 2010, we documented 121 interpack aggressive interactions. We recorded pack sizes, compositions, and spatial orientation related to residency to determine their effects on the outcomes of interactions between packs. Relative pack size (RPS) improved the odds of a pack displacing its opponent. However, pack composition moderated the effect of RPS as packs with relatively more old members (>6.0 years old) or adult males had higher odds of winning despite a numerical disadvantage. The location of the interaction with respect to pack territories had no effect on the outcome of interpack interactions. Although the importance of RPS in successful territorial defense suggests the evolution and maintenance of group living may be at least partly due to larger packs' success during interpack interactions, group composition is also an important factor, highlighting that some individuals are more valuable than others during interpack conflicts.

Summary1. Understanding the population dynamics of top-predators is essential to assess their impact on ecosystems and to guide their management. Key to this understanding is identifying the mechanisms regulating vital rates. 2. Determining the influence of density on survival is necessary to understand the extent to which human-caused mortality is compensatory or additive. In wolves (Canis lupus), empirical evidence for density-dependent survival is lacking. Dispersal is considered the principal way in which wolves adjust their numbers to prey supply or compensate for human exploitation. However, studies to date have primarily focused on exploited wolf populations, in which density-dependent mechanisms are likely weak due to artificially low wolf densities. 3. Using 13 years of data on 280 collared wolves in Yellowstone National Park, we assessed the effect of wolf density, prey abundance and population structure, as well as winter severity, on age-specific survival in two areas (prey-rich vs. prey-poor) of the national park. We further analysed cause-specific mortality and explored the factors driving intraspecific aggression in the prey-rich northern area of the park. 4. Overall, survival rates decreased during the study. In northern Yellowstone, density dependence regulated adult survival through an increase in intraspecific aggression, independent of prey availability. In the interior of the park, adult survival was less variable and density-independent, despite reduced prey availability. There was no effect of prey population structure in northern Yellowstone, or of winter severity in either area. Survival was similar among yearlings and adults, but lower for adults older than 6 years. 5. Our results indicate that density-dependent intraspecific aggression is a major driver of adult wolf survival in northern Yellowstone, suggesting intrinsic density-dependent mechanisms have the potential to regulate wolf populations at high ungulate densities. When low prey availability or high removal rates maintain wolves at lower densities, limited inter-pack interactions may prevent density-dependent survival, consistent with our findings in the interior of the park.