Properties of the setting
Both prey and predator traits crucially predict the outcome of their non-consumptive interactions. Additionally, the propagation of NCEs depends also on the setting in which the interaction takes place. During phase one, changes to the environment may impair prey detection of predator cues by disrupting acoustic (e.g., owing to anthropogenic noise; Chan et al . 2010), chemosensory (e.g., because of pollution; Lürling & Scheffer 2007), or visual (e.g., via increased turbidity; Abrahams & Kattenfield 1997) systems. These environmental changes may reduce the likelihood of, or even preclude, anti-predator behavior. For example, predator avoidance by freshwater snails (Physa acuta , Helisoma trivolvis ) disappeared when eutrophication of their outdoor mesocosms led to chemosensory impairment (Turner & Chislock 2010). Environmental features that impede predator detection may themselves be recognized as risk cues by prey and thereby result in elevated defensive investment. For example, Embar et al . (2011) demonstrated that gerbils (Gerbillus andersoni allenbyi ) reduced their foraging activity (measured by GUDs) when landscape features blocked sightlines that were necessary for anti-predator vigilance. In general, then, environmental heterogeneity with respect to properties that influence predator detection may mediate substantial inter-individual and population variation in the degree to which prey recognize predation danger and subsequently experience and transmit NCEs.
During phase two, properties of the setting may influence the scope for prey anti-predator behavior in several ways. First, predator and prey habitat domains are shaped in part by environmental context (e.g., thermal conditions, vegetative structure; Schmitz & Barton 2014). Accordingly, environmental factors constraining prey movement or the amount of predator-free space could dictate whether prey manifest chronic anti-predator vigilance, use refugia, or experience consumptive effects under the four contingent scenarios described above. Barton & Schmitz (2009) showed, for example, that experimental warming created enemy-free space by shifting the environment from one where two spider predators were spatially complementary to overlapping. This led to a strictly non-consumptive interaction whereby grasshoppers avoided predators rather than a composite scenario where they avoided the sit-and-wait predator but experienced consumptive effects of the active hunter.
Second, even when predator and prey domains are unaffected by the setting, landscape features can shape NCEs by modifying the efficacy of prey escape behavior. The ability of an individual to escape a predator following an encounter can depend on environmental factors that influence mobility (e.g., terrain) or visibility (e.g., when the background affects prey camouflage) (Wirsing et al . 2010). Thus, areas with properties that render prey escape tactic(s) less effective are likely to be avoided, at least when predators are present, or to elicit other countermeasures that enhance the probability of early predator detection (e.g., vigilance). For instance, reef habitat complexity enhanced and dampened anti-predator behaviors of large and small fishes, respectively, likely because large-bodied fish are less able to flee from predators through obstacle-rich reefscapes than their smaller counterparts (Catano et al . 2016).
Third, food quantity or quality at the landscape scale can shape NCEs by influencing the mean energetic state of prey populations (Heithauset al . 2008; Wirsing & Ripple 2011). In depauperate landscapes, average energetic states will be depressed and the overwhelming necessity of food should drive foraging decisions (Chesson & Kuang 2008), whereas anti-predator investments should increase when resources are plentiful and prey have nutritional reserves (Hopcraft et al . 2010; Matassa & Trussell 2014). For example, Matassa et al . (2016) found that elevated resource (barnacle, Semibalanus balanoides ) density strengthened anti-predator investment (refuge use) by sub-adult snails (Nucella lapillus ) exposed to risk cues from predatory crabs (Carcinus maenas ), presumably by augmenting prey state.
Fourth, interacting predator-prey pairs are unlikely to do so in isolation from other species, which may alter the focal prey species’ responses to perceived risk. For instance, dwarf mongooses (Helogale parvula ) displayed lower rates of anti-predator vigilance when in the presence of an avian co-forager, the drongo (Dicrurus adsimilis ) (Sharpe et al . 2010). The presence of other predators may also affect the transmission of NCEs if prey species with conflicting predator-specific responses consequently reduce their investment in defense (Sih et al . 1998). In accord with predator facilitation (Charnov et al . 1976; Kotler et al . 1992; Korpimaki et al . 1996), for example, Meadows et al . (2017) showed that larval mosquitoes (Culex pipiens ) abandoned diving behavior normally deployed to escape surface-hunting insect mesopredators when also exposed to a benthic predator (dragonfly naiads,Aeshna spp.).
Lastly, landscape properties may mediate how prey are affected by temporal variation in predation risk (Box 2 ). Many nocturnal animals, for example, decrease their activity on moonlit nights because of their increased exposure to visually-orienting predators, and this trend is accentuated in areas dominated by open habitats (Prugh & Golden 2014). Therefore, moonlight exacerbation of NCEs experienced by nocturnal prey species is likely to be inversely proportional to landscape cover availability. Landscapes also may influence temporal patterns of predation risk, and thus NCEs, over longer intervals. Seasonal variation in snow accumulation, for example, can give prey a temporary refuge or heighten vulnerability to predation by restricting mobility (Gorini et al . 2011). Not surprisingly, snow depth has been linked to prey risk taking (e.g., yarding in deer; Nelson & Mech 1991).
Any of these environmental attributes, alone or in concert, can influence the kinds of anti-predator behaviors that manifest during phase two and that precipitate as indirect NCEs during phase three (Heithaus et al . 2009; Wirsing & Ripple 2011). Thus, direct and indirect non-consumptive relationships between the same suites of interacting predator and prey species may differ markedly as a function of landscape type. Trussell et al . (2006) determined, for example, that habitat type (availability of refugia) shaped how risk from crabs (Carcinus maenas ) altered the foraging intensity of a snail (Nucella lapillus ) and, consequently, the levels of consumption of the snails’ resource (S. balanoides ).