Enemy-risk effects and predator-predator interactions
Insect herbivores face the dual challenge of well-defended host plants and natural enemies (Polis 1999). It has become increasingly well established that predators must also forage for defended food resources (their prey) under risk of predation. Enemy risk can stem from specialist higher-order enemies (e.g., obligate hyperparasitoids); intraguild predators (competitors that also engage in uni- or bi-directional predation with the focal predator); or cannibalistic conspecifics (Polis 1981; Polis et al. 1989; Rosenheim et al. 1995; Rosenheim 1998; Schausberger 2003; Wise 2006). And, just as for herbivores, the impacts of higher-order predators, intraguild predators, and cannibals can be both consumptive and non-consumptive (reviewed by Snyder & Ives 2008; Frago 2016). Although enemy-risk effects expressed by predators reacting to other predators are generally viewed as adaptations to reduce their own risk of predation, in most cases it is difficult to separate benefits from reducing the costs of predation versus reducing the costs of competition, or even other costs of high density, such as transmission of diseases that have broad host ranges. Predation risk reduction can, however, be clearly identified as the driver when competition and disease can be ruled out, such as when a primary parasitoid abandons host patches where it detects pheromones produced by an obligate hyperparasitoid (Höller et al. 1994).
                  Natural enemies express a broad array of responses to their own predators. A common response is to move away from areas where predator risk is perceived; this may be measured experimentally as shorter patch residency times (Nakashima & Senoo 2003; Meisner et al. 2011; Frago & Godfray 2014), reduced oviposition or prey consumption (Agarwala et al. 2003; Magalhães et al. 2004; Meisner et al. 2011; Choh et al. 2015), or outright avoidance of patches where predators or predator-associated cues are detected (Magalhães et al. 2004; Choh et al. 2015; Cotes et al. 2015; Seiter & Schausberger 2015). Occasionally, parasitoids have been found to increase, rather than decrease, their oviposition activity in host patches with elevated predation risk, likely due to high patch quality even considering predator presence (e.g., Velasco-Hernández et al. 2013). Other common responses include modulation of overall foraging activity (either increased or decreased; Magalhães et al. 2004; Bucher et al. 2014; Walzer et al. 2015; Hentley et al. 2016) and increased use of refuges (Venzon et al. 2000). Developmental effects include increased mortality, delayed (or sometimes accelerated) development, decreased (or sometimes increased) adult body size, and shortened pre-oviposition periods for adults (Walzer et al. 2015; Michaud et al. 2016). Compensatory growth has been recorded following periods of elevated predation risk that slowed growth (Walzer et al. 2015). In many cases, predators respond not to reduce their own risk of predation, but rather to reduce the likelihood that their more vulnerable offspring will be attacked. Transgenerational phenotypic plasticity in response to predation risk has been recorded (Seiter & Schausberger 2015), and in cases where predator-prey role reversals are possible, adult predators that witness a heterospecific predator attacking juvenile members of its own species may subsequently be more aggressive in reciprocal attacks on juveniles of the attacking species (Choh et al. 2014). Predators may even invade central locations within colonies of their prey to secure the predation risk-reduction benefits of a selfish herd (Dumont et al. 2015).
            What influence these responses have on the overall success of biological control is uncertain. Much of the literature is framed around the idea that anti-enemy behavior of intraguild prey ameliorate the impact of intraguild predation, potentially facilitating the coexistence of multiple natural enemies, and presumably enhancing suppression of pest populations. In the short-term, however, anti-predator responses that reduce potential intraguild predation or cannibalism often results in reduced overall consumption of prey (Sih et al. 1998; Vance-Chalcraft et al. 2007). Localized loss of contributions to biological control ascribed to non-consumptive effects of intraguild predators or hyperparasitoids have indeed been reported (Höller et al. 1994; Raymond et al. 2000; Meisner et al. 2011; Frago & Godfray 2014). But it is easier to record the potential erosion of biocontrol in a focal patch of prey than to document the possibly enhanced biocontrol elsewhere (for one study that investigated but did not find such an outcome, see Frago & Godfray 2014). Predators that abandon patches of rich host/prey resources due to the presence of other natural enemies presumably weaken biocontrol in those patches, but may strengthen biocontrol elsewhere. Furthermore, consumptive and non-consumptive effects have not been separated in these studies, and doing so while still assessing the overall level of biocontrol success would not be easy: treatments (e.g., mouthpart manipulations) that could be applied to an intraguild predator to eliminate CEs imposed on an intermediate predator would also, unfortunately, eliminate CEs on the shared herbivore prey. Studies of hyperparasitoids could avoid this problem. In some cases the herbivores themselves have been shown to recognize localized enemy-free space generated by hyperparasitoids and to respond with elevated per capita reproductive output, perhaps as a consequence of reduced expression of costly anti-predator defenses (van Veen et al. 2001). To our knowledge, no one has attempted to measure or model the global effects of fear-mediated redistribution of natural enemies (but see Northfield et al. 2017 for a model that could provide a useful framework for such an investigation).