Propagation of NCEs in communities
Predation risk is typically defined as the probability of an individual becoming prey within a given place and time (Lima & Dill 1990). However, predation risk could just as easily be conceptualized as the probability of an individual becoming prey at a given place and time assuming no, or some set amount of, anti-predator investment (Lank & Ydenberg 2003). Predation risk under the former definition is more intuitive, given its direct link to observable patterns of mortality, and therefore lends itself to estimation via the combination of spatiotemporal probabilities of encountering and being killed by predators (Lima & Dill 1990; Lima 1992). Estimates of risk based on this definition reflect inherent properties of the location and time of interest and dynamic properties of the predator (e.g., decisions about whether to attack in response to prey behavior) and the prey (e.g., defensive investment). Accordingly, we view them as measures ofrealized predation risk . By contrast, predation risk under the latter definition, termed intrinsic predation risk ordanger , is harder to measure because it is an abstract construct (Lank & Ydenberg 2003). Nevertheless, this latter conceptualization importantly decomposes the process by which prey individuals experience and respond to the threat of predation into a series of steps beginning with exposure to risk stimuli and ending with changes to prey numbers and traits (e.g., behavior) that may affect additional species within the community (Fig. 1 ). Consequently, it provides a clearer mechanistic basis for understanding when and how various sources of contingency might direct the propagation of NCEs through communities than does the former definition. Hence, while acknowledging the validity of both approaches to defining predation risk, we focus on intrinsic risk for the remainder of our review.
Propagation of NCEs consists of three phases (Fig. 1 ) within a context of intrinsic risk. Every point in space and time is characterized by some value of intrinsic predation risk, which includes spatial properties of the situation that influence the likelihood of predator-induced mortality but that prey cannot easily modify through behavioral changes. These properties include availability of refuges, presence of escape impediments, dilution of risk by conspecifics and by other species, and the abundance of predators and species that might inhibit predator effectiveness (Lank & Ydenberg 2003). Collectively, they are often viewed as determinants of the background pattern of risk for a given location. Areas with elevated background risk are sometimes called risky places (Creel et al . 2008). Intrinsic risk is also influenced temporally by whether predators, and other species or environmental conditions (e.g., moonlight) that might influence the predator’s efficacy, are currently present at a location. Periods when the presence of predators or conditions heighten prey vulnerability are considered to be risky times (Creel et al . 2008).
Within the setting of intrinsic risk, phase one concerns whether the forager perceives any cues related to the current level of intrinsic risk. Prey may either detect spatiotemporal cues that reflect intrinsic predation risk (including an attack itself), setting up the possibility of NCEs, or fail to detect appropriate risk stimuli, in which case no NCEs will result (from the cue in question) and mortality from the predator will be more likely. Thus, factors influencing prey detection of intrinsic risk cues may operate as key sources of context dependencies in NCEs.
Foragers that perceive intrinsic cues can then respond to them in phase two. Perceived danger may or may not elicit a prey response of sufficient magnitude to precipitate a NCE. In response to background risk and risky times, prey individuals may manage this risk proactively. In response to immediate threats (including attacks), prey may respond reactively through behavioral countermeasures (Creel 2018). The energetic, reproductive, and opportunity costs that ensue from these adjustments determine the magnitude of any associated risk effects (Creel & Christianson 2008). Thus, the type of anti-predator behavior exhibited by a prey individual in any situation is crucial to whether and to what extent it will experience fitness penalties. Prey individuals that perceive danger may also experience stress, which may affect fitness (Clinchy et al . 2013) and thereby precipitate risk effects either alone or in concert with other (e.g., lost opportunity) costs of anti-predator behaviors. Accordingly, during phase two, factors that influence the strength of responses to perceived risk, the form of anti-predator behaviors, and the amount of associated stress could act as important drivers of contingency in associated predator risk effects experienced directly by prey and ensuing propagation of NCEs.
In phase three, the responses of the forager to intrinsic risk can give rise to indirect effects on other species. Risk effects from predator-induced risk management and stress can reduce prey population size (Creel & Christianson 2008) and thereby trigger indirect interactions if prey abundance drops enough to affect other community members. Moreover, the nature of prey risk management can determine whether and how other species in the community are affected indirectly. Some behavioral adjustments may only affect the prey species that responds to perceived risk, potentially leading to direct risk effects, whereas others may further (or exclusively) influence third parties and thereby propagate through ecological communities as indirect interactions. Therefore, any factor that modulates the impacts of perceived risk on prey population size and anti-predator behaviors also has the potential to shape indirect NCEs.