Abstract
Individuals can alter their behaviour and other traits to reduce threats from predators and parasites. However, predators and parasites likely elicit different responses, which subsequently lead to different non-lethal effects. We created a sequentially structured framework to examine trait responses to distinct predatory and parasitic consumers. We predicted that parasites with strong negative effects on host fitness should act like predators and elicit strong responses before attack. We also predicted that less damaging parasites and micropredators should elicit diverse responses across multiple interaction stages, because their hosts and prey remain alive while being eaten. A meta-analysis indicated that predators do tend to elicit stronger responses than parasites before attack, whereas parasites generally elicit responses after attack, albeit weaker than pre-attack responses to predators. Organisms exposed simultaneously to predator and parasite cues responded similarly when exposed to predator cues alone, suggesting that individuals prioritize anti-predator responses over responses to less harmful parasites. Extending these findings requires addressing knowledge gaps concerning responses to different consumer types, costs of immune responses, and cumulative effects of repeated responses. Expanding research beyond the predator vs. parasite dichotomy toward a broader consumer-resource perspective will facilitate understanding of non-lethal effects in complex, multi-trophic food webs.
Introduction
“Whenever I swim in the ocean… I feel increasingly panicky and … I must leave the water” is a typical response to the 1975 film Jaws (Cantor 2004). As with moviegoers, many species respond to predators by changing behaviours, physiology, or even appearance to avoid being eaten. These non-lethal effects of predators, known as ‘trait responses’, are pervasive and take many forms, such as seeking shelter (Creel et al. 2005) or maturing faster to reach less vulnerable life stages (Raffel et al. 2010). Such trait responses influence how individuals interact with the broader community, driving ‘trait-mediated effects’ that range from reduced individual fitness to trophic cascades (Werner & Peacor 2003; Ritchie & Johnson 2009; Buck & Ripple 2017) that can destabilize communities (Pringle et al. 2019). Wolves, for example, frighten elk away from exposed foraging grounds into sheltered habitats with less nutritious vegetation, which then reduces elk birth rates (Creel et al. 2007) and alters vegetation structure (Fortin et al. 2005). Predators can therefore impact species and communities without directly killing prey, just as Jaws kept many people from swimming in the summer of 1975.
Perhaps less well recognised is that parasites also elicit trait responses in hosts with associated non-lethal effects. To reduce infection risk, hosts may avoid infected conspecifics (Milinski & Bakker 1990; Kavaliers et al. 2003a; Behringer et al. 2006), defend against infectious propagule attack (Sears et al. 2013), or avoid risky areas, such as faeces representing a hot spot of undetectable nematode eggs (Hart 1994; Curtis 2014; Weinstein et al. 2018). Furthermore, in stark contrast to predation, parasitism is not immediately lethal, so hosts can also respond after successful parasite attack through various physiological and behavioural responses (Rigby et al. 2002; Raberg et al. 2009; Buck 2019). For instance, a caterpillar can initiate an immune response to prevent being killed by a parasitoid wasp (Abram et al. 2019). Basic emotions like “disgust” (Curtis & de Barra 2018; Tybur et al. 2018; Weinstein et al. 2018) and the age-old cliché “avoid like the plague” suggest that parasite avoidance is interwoven in our own history as much as is our fear of predators. The diverse trait responses elicited by parasites has led some to hypothesize that parasites actually impose stronger cumulative non-lethal effects than predators (Rohr et al. 2009; Buck & Ripple 2017).
In this review, we compare trait responses to predation and parasitism, considering how they may overlap and differ. Although predators and parasites threaten most species in natural ecosystems, trait responses to predators and parasites have been largely studied in isolation. As a result, how trait responses to parasites compare with trait responses to predators is still unclear. We used a general consumer-resource model to develop hypotheses and predictions for how key life history differences among predators and parasites, such as the number of attacks they make in a lifetime and whether they kill organisms while eating them, should influence the likelihood and magnitude of trait responses at different interaction stages. We then conducted a systematic review and meta-analysis to: (a) assess the literature that compares trait responses to different forms of predation and parasitism, (b) compare average response magnitudes between predation and parasitism, and (c) test how factors related to resources, consumers, and study designs influence trait responses. We conclude by pointing to several unresolved questions concerning how non-lethal species interactions affect community and ecosystem dynamics.
A general trait-response framework for examining non-lethal effects
Theoretical framework
Predators and parasites employ various ‘consumer strategies’, that is, how individuals find, attack, and consume organisms (Lafferty & Kuris 2002; Lafferty et al. 2015). For example, predators have short feeding times (i.e., seconds to days) and eat multiple organisms in a lifetime, whereas parasites feed on hosts for up to months or even years, but die or transform after a single feeding interaction. Predators kill prey before or while consuming them, but mosquitoes and other micropredators do not. Although some parasites eliminate host fitness, exemplified by parasitoids and parasitic castrators, others infect hosts without substantial fitness impacts. These and other differences in consumer strategies likely affect how, and to what extent, organisms respond to them (Buck 2019). To account for these differences when predicting the non-lethal effects of predation and parasitism, we draw on consumer-resource theory to develop a general trait-response framework that applies across both host-parasite and predator-prey systems.
We use, as a scaffolding, the general model for consumer-resource population dynamics developed by Lafferty et al. (2015), which collates the key elements of all consumer-resource models into a single, temporally compartmentalized structure (Fig. 1, Box 1). Briefly, predators and parasites are ‘consumers’, whereas their prey and hosts are ‘resources’. Interactions are broken into sequential transitions between up to three discrete consumer states and up to four corresponding resource states (circles in Box 1). State transitions — i.e., mortality, contact, attack failure and success, and feeding (arrows in Box 1) — occur at various rates. Basic differences among consumer – resource systems are incorporated with a set of binary parameters that alter model structure (Box 1).
Overview of the general consumer resource model
The Lafferty et al. (2015) model helps identify and compare different consumer-resource interactions using a singular and formalized mathematical logic. During interactions the consumer can be in one of three states: “questing”, “attacking”, or “consuming”. The resource, here being a prey or host, moves through a corresponding four states; they are “exposed” when under attack, being “ingested” while consumers are consuming them, and can either be “susceptible” or “resistant” to new attack. State transitions are driven by multiple biological processes, which can be represented by simple per capita rates or complex functions. Consumers and resources transition from questing and susceptible states, respectively, according to detection rates of each organism and attack rates of the consumer, while they transition to consuming and ingested states, respectively, according to attack success rates of the consumer (Fig. 1a). Handling rates determine transitions back to questing and susceptible states or alternatively, death rates of consumers and resources following consumption. All states have death rates that allow for specific costs to be specified. The expanded model shown here can be formally simplified (through setting some rates to zero to subsume states) to represent nearly all classic consumer-resource models.
The expanded model does not represent a specific consumer-resource interaction. Instead, a set of binary parameters (switches) makes it easy to specify distinct consumer strategies. For instance, the “fatal attack” parameter (f) defines whether resources are dead (f = 1) or alive (f = 0) after consumption, which distinguishes predators and parasitoids from other parasites and micropredators. The “joint death” parameter defines whether the consumer dies if its resource dies (j = 1) or remains living (j = 0), which distinguishes all parasites (which have an intimate relationship with their host) from predators and micropredators. Finally, the “multiple attacks” parameter defines whether the consumer attacks once in a lifetime (m = 0), as parasites and parasitoids do, or more than once in a lifetime (m = 1), as micropredators and predators do.
We derive three basic trait responses from the model: avoid contact, counter attack, combat consumption (Fig. 1)1. This temporal sequence implies that trait responses to minimize consumption can be driven by multiple biological mechanisms. In initial interactions with questing consumers, susceptible resources may avoid contact in various ways. Avoidance serves to reduce the rate that questing consumers transition to attacking, with the benefit that susceptible resources transition more slowly to exposed states (Table 1). Avoidance responses may manifest as trait adaptations [i.e. constitutive responses (Westra et al. 2015)], such as camouflaged coloration shown by many species to reduce visibility to predators (e.g. Stevens & Merilaita 2009), or as induced plastic traits, exemplified by herbivores moving away from nematode-infected faeces (Hutchings et al. 2001; Weinstein et al. 2018) or wolf scents (Creel et al. 2005). Resources that become exposed to attacking consumers may counter attack to increase attack failure (Box 1). Countering attack includes “fight or flight” responses, like hares sprinting to burrows when being chased by lynx, or tadpoles jolting their bodies when being attacked by trematode cercariae (Sears et al. 2013). Finally, resources—particularly hosts of parasites—that become ingested may combat consumption. Combating consumption shortens or slows consumer feeding rates [i.e. ‘resistance’ in parasitology (Rigby et al. 2002)], or lessens the damage of being eaten [i.e. ‘tolerance’ in parasitology (Raberg et al. 2009)]. Responses that shorten or slow feeding include behaviours like social grooming by primates (Hart & Hart 2018) and adaptive immune responses to parasitism (Hawley & Altizer 2011). Increasing tissue repair and protecting high-risk areas of the body from feeding, as tadpoles do for trematodes (Sears et al. 2013), are ways that resources combat consumption by reducing damage without affecting consumer feeding rates. Whether, and to what degree, resources combat, avoid, or counter, will depend on the different pressures imposed by different consumers, which are modelled by altering rates of contact, attack, and/or consumption. Distinguishing these three model-derived resource trait responses makes it easier to compare, contrast, and make predictions about the different ways that consumers exert non-lethal effects on resources.
Constraints on trait responses
Despite the clear benefits that trait responses provide to resources, various constraints can limit a resource’s ability to avoid, counter, and combat its consumers. Hence, predicting the timing and magnitude of trait responses also requires accounting for constraints on resources. Constraints can arise from basic limitations on detecting consumers and mounting responses, or from trade-offs. First, resources can mount responses only if they can detect consumer threats. Resources use visual and non-visual cues to detect predation and parasitism risk, making sensory limitations - e.g., sight, hearing, and smell – a potential constraint on trait responses. Second, resources must possess the morphology, physiology, and energy level to mount specific responses to consumers. For instance, tadpoles cannot physically leave ponds when predators are present as can adult amphibians, so they may instead reduce activity levels to avoid consumer contact (Hossie et al. 2017). Third, because trait responses can compete with essential activities like feeding, reproducing, or maintenance (Dröge et al. 2017; Hart & Hart 2018), trade-offs can also constrain resource responses. For example, moose can afford to avoid wolves more in summer than in winter when food is scarce (Oates et al. 2019). These three constraints likely interact to jointly influence response timing and magnitudes. For example, food scarcity may interact with moose body size or background nutritional levels to determine their ability to avoid contact with wolves. Considering these constraints, along with the potential benefits of response, leads to several hypotheses and associated predictions for how resources should respond to different consumer threats.
Hypotheses and predictions concerning trait responses against predators and parasites
Hypothesis 1: Severe fitness consequences of consumption will favour strong trait responses—and strong non-lethal effects—at early interaction stages
Severe fitness consequences of consumption make combat responses very risky, placing a premium on mounting defensive responses at earlier interaction stages when consumers are questing or attacking. This leads to the intuitive prediction that predators should elicit strong avoidance and counter responses because successful predation leads to death. Perhaps less intuitively, this hypothesis also leads to the prediction that parasitoids, which regularly kill their hosts (Lafferty & Kuris 2002; Abram et al. 2019), and parasitic castrators, which reproductively kill their hosts, will be similar to predators in eliciting strong responses at the two early interaction stages. Other types of parasites with strong negative fitness impacts, such as certain pathogens, may place similar pressures on hosts by producing detrimental infections. In contrast, micropredators and less harmful parasites should elicit the weakest avoidance and counter responses of all consumer types.
Hypothesis 2: Resources that remain alive while being eaten can implement and concentrate responses—and incur non-lethal effects—during consumption
In stark contrast to most prey of predators, hosts are alive while parasites feed, and prey are alive when micropredators feed. Those resources can therefore mount combat responses while being ingested. This leads to the prediction that, compared to predators, many parasites and micropredators will evoke more types of responses with a more even distribution among the three interaction stages. Because hosts of castrators and many parasitoids remain alive while being eaten, these predictions apply for these types of parasites as well. A less obvious prediction is that the ability to combat consumption may lead to resources concentrating responses in the third interaction stage for parasites – including parasitoid and parasitic castrators - and micropredators, particularly in cases where constraints or trade-offs limit avoidance and combat responses.
Hypothesis 3: Detection ability determines trait response timing and magnitude
Regardless of the fitness consequences of consumption, an inability to detect questing predators and parasites will preclude mounting avoidance responses. Under this hypothesis, the strength of responses should increase with the ease of detecting consumers. For instance, to the extent that visual detection is important to elicit defensive responses, we predict that resources avoid questing predators more than questing micropredators and parasites, because the generally larger size of questing predators relative to questing micropredators and parasites make them easier to see. Exceptions will exist, however; some parasites have large searching stages (e.g., some hymenopteran wasp parasitoids), which would permit easy detection and subsequent avoidance by hosts. Further, certain predators, like some ambush predators and filter feeders, are actually not detectable while questing, but only during attack, which precludes avoiding contact but favours countering attack. Hence, this hypothesis does not predict consistent differences among consumer strategies (e.g. predators, parasites, micropredators, etc.), and could perhaps be most strongly tested using predator and parasite species that span a range of detectability.
These three hypotheses are not mutually exclusive and can be integrated to predict how trait responses vary among the different types of predation and parasitism in realistic situations (Fig. 1b-d). For instance, foraging mice will avoid questing owls by hiding in burrows (Fig. 1b). Burrow use may vary with flyover frequency, which predicts mouse-owl contact rates. Burrow use is traded-off against the cost of reducing mouse feeding, and consequently likely depends on mouse nourishment. Mice may also detect infective nematode eggs in feeding fields and then avoid contact by moving to other locations (Fig. 1c). Avoiding contact with nematodes is likely constrained by mouse ability to detect eggs and should also depend on movement costs and food availability in new locations. Unlike avoiding owls, the strength with which mice avoid nematodes may also depend on their ability to combat infection (consumption by the nematode), perhaps by mounting an immune response. Effective immune function would favour mice avoiding substantial avoidance costs (e.g., if mice avoid rich feeding areas with nematodes for poor feeding areas lacking nematodes), which would concentrate non-lethal effects in consumption stages. All three responses may also be expected from caterpillar hosts of wasp parasitoids, with the difference that high risk of death via successful consumption may favour a shift to earlier avoidance and counter responses (Fig. 1d). What remains unclear is whether the broader range of responses, and potential concentration of combat responses, to less virulent parasites like nematodes together might, when combined, rival the magnitude of the avoidance and counter responses to consumers with severe fitness consequences. To compare the overall magnitude of trait responses to parasites and predators, we conducted a systematic review of and meta-analysis on available literature.
Systematic review and meta-analysis of trait responses to predation and parasitism
We systematically reviewed studies that measured the magnitude of trait responses elicited by predators and parasites. We then performed a meta-analysis on the compiled data to test the predictions established above. Our broad goal with the meta-analysis was to assess how the type and magnitude of trait responses vary by consumer strategy as defined by Lafferty and Kuris (2002): solitary predators, trophically transmitted parasites, typical parasites, and pathogens, parasitoids, parasitic castrators, micropredators, or social predators. Many studies have measured trait responses to predators and parasites alone, but we limited our review to studies that measured comparable responses to both a predator and a parasite for the same resource species. To focus on trait responses conferring defence, we did not consider trait changes originating from general parasite pathology, or from parasite adaptive manipulation of host traits (Poulin et al. 1994; Moore 2002; Lafferty & Shaw 2013), though our framework can accommodate such effects.
Detailed information about data collection, extraction and analyses are available in the Supplementary Material. Briefly, we compiled 129 entries from 15 studies, which included 44 predator-prey interactions, 44 host-parasite interactions, and 41 simultaneous interactions with predators and parasites (Table S1). Although studies measured several morphological, behavioural, and physiological responses, behavioural traits were most common, with activity level being the most reported trait (Fig. S1). No studies fitting our criteria measured physiological or immunological trait responses. The studies spanned the following consumer strategies: solitary predators, trophically transmitted parasites, typical parasites, and pathogens (Table S1). We therefore could not consider responses to parasitoids, parasitic castrators, micropredators, or social predators. Hereafter, we broadly distinguish between predators (i.e. solitary predators) and parasites (i.e. trophically-transmitted, typical, or pathogens). Predator-induced trait responses were only measured during the questing predator state (Fig. S1), whereas measurements of parasite-induced responses included questing (10), attacking (9), and consuming (25) states (Fig. S1). There were 77 entries for individual-level responses and 49 entries for group-level responses.
We calculated the standardized mean difference (Hedge’s d) (Koricheva et al. 2013) from included studies as the measure of trait response magnitudes, whereby positive effects denoted reductions in trait values (e.g., reduced activity level or mass), except in measures of space use that measured time in a refuge or distance from a consumer cue (e.g. positive effect would mean an increase in refuge use). We reversed the sign of these values so that positive effect sizes would denote reductions in use of risky habitats, indicative of defence (see Table S1 for further details). Because studies often included a treatment containing both a predator and parasite cue, we also estimated the magnitude of trait responses to the combined presence of predators and parasites. We did not have predictions for how these responses would compare to those made to predators or parasites by themselves. We also considered the following factors related to consumers, resources, and experimental designs that may have influenced response magnitudes: a) the type of trait measured to quantify responses (trait type), b) whether studies used a trait value of individuals or proportions of individuals in group that exhibited the focal trait (analysis scale); c) whether responses were elicited under the physical presence of the consumer, or solely by indirect cues such as used media or ingested conspecifics (consumer presence), d) the genus and species of the consumer and resource, and e) whether the consumer was in a questing, attacking or consuming state (Box 1) when the trait response was measured. We also assessed consumer state and consumer strategy effects using only the parasite data because data on responses to predators were limited to one consumer state (questing) and strategy (solitary predator).
Results
We found considerable variation in response magnitude and direction to predators (Fig. 2a), parasites (Fig. 2b), and their combination (Fig. 2c). However, although individual parasite-induced effects were sometimes just as strong as predator-induced effects (Fig. 2), on average and across all stages, predator-based trait responses were stronger than parasite-based trait responses (Table S2, Fig. 3a). These patterns were also evident after controlling for consumer state (i.e., questing predators vs. questing parasites) (Table S2, Fig. 3b). Nevertheless, distinguishing between parasite states (questing, attacking, or consuming) revealed that parasites, on average, did elicit responses, but only while they were consuming (i.e. infecting) their hosts (Table S2. Fig. 3b). The simultaneous presence of predators and parasites also elicited responses on average, and they were similar in magnitude to trait responses elicited by predators alone (z = 0.10, p = 0.476; Fig. 3a). Predators and the simultaneous presence of predators and parasites elicited reductions in activity but, on average, did not influence space use or morphological/physiological traits (Table S2, Fig. 3c). Whether traits were measured at the individual level or group level (e.g. proportions) influenced response magnitudes, with group-level responses being stronger (Table S2, Fig. d). Responses were not contingent on the consumers being physically present; indirect cues of the consumers elicited similar responses (Table S2). Across the host-parasite interactions studied, responses did not depend on the specific strategy of parasites (pathogens, trophically transmitted parasites, or typical parasites; Table S2), and there was insufficient replication to consider how consumer or resource taxon influenced responses. For that reason, these results mostly pertain to amphibians as resources (Table S1).
Discussion
We used a general consumer-resource model to construct a framework that can be broadly applied across many predator-prey and host-parasite systems to predict trait responses and non-lethal effects. The framework identified plausible mechanisms driving the timing and magnitude of trait responses, including the fitness consequences associated with being eaten, whether individuals are alive while being eaten, and the ease of detecting consumers. From these hypothesized mechanisms, we generated testable predictions regarding how trait responses should differ between predator-prey and host-parasite interactions. We generally predict severe fitness consequences of predator consumption to drive strong avoidance and counter responses in prey before any contact is made. We also predict that, in general, host responses to parasites are weaker than those of prey, but also more diverse; a range of behavioural, morphological, and physiological responses can be made throughout all interaction stages (Rigby et al. 2002; Raberg et al. 2009), and potentially concentrated to combat consumption. These general predictions were supported by our meta-analysis on the existing literature to directly compare trait responses to predatory versus parasitic consumers. However, the meta-analysis data predominantly related to a rather narrow taxonomic range of predators and parasites, and several limitations of the included studies (Box 2) suggest ways forward to corroborate and expand the results.
Future directions for research on non-lethal effects
There remain several limitations in the comparative data on trait responses to, and non-lethal effects of, predators and parasites. Addressing the following limitations will lead to more comprehensive estimates of non-lethal effects in diverse, real ecosystems:
1. Comparative experiments of trait responses that cover a broader range of consumer-resource systems
can determine how generalizable patterns are across systems, and opens avenues to consider factors, like phylogenetics of consumer and resource, on response magnitudes. Most comparative trait response studies identified by our systematic review used larval amphibians and their aquatic consumers. The experimental designs of those studies provide examples of how comparisons of trait responses to different types of consumers can be performed with other species to test the generality of our meta-analysis results.
2. Comparative studies that include responses to parasitic castrators, parasitoids and micropredators
are needed to provide novel tests of the mechanisms, such as detection and fitness consequences, underlying trait responses to consumer threats (see Discussion). Comparisons of trait responses to different consumer types are currently limited to typical and trophically-transmitted parasites, pathogens, and solitary predators – only three of the 10 consumer life histories found in natural ecosystems (Lafferty & Kuris 2002; Lafferty et al. 2015).
3. Considering the full array of parasite non-lethal effects on hosts
will be important to truly compare predators and parasites. For instance, though immune responses are one of the most common forms of anti-parasite defence, we were not able to include immune responses into comparisons of trait responses to predators and parasites. Because immune responses can be costly to initiate and maintain, such host responses certainly cause non-lethal effects. Further, parasites may alter host traits in ways that do not involve defensive responses, including by pathological energy drain or tissue damage, or via adaptive host manipulation of host phenotypes. All these effects occur during the consumption phase and will be uniquely pronounced for parasites compared to predators.
4. Longitudinal data on prey and host responses
are needed to consider how multiple responses elicited by single consumers collectively shape the magnitude of non-lethal effects. Longitudinal data may even reveal interactive effects between trait responses, such that mounting one type of response affects the strength of other responses. For example, parasites whose infections are not highly costly (e.g., many gastrointestinal helminths) may cause hosts to prioritize combat responses to minimize trade-offs of avoiding contact against fitness-related activities, like feeding and reproducing (Moore 2002; Hart & Hart 2018). This occurrence may explain why newts, for example, do not avoid infective questing parasites in breeding ponds, but infections from those parasites drive the same individuals to leave those ponds (Daversa et al. 2018). Prioritization of combat responses to parasitism provides one explanation for our finding that average responses to parasites were strongest after parasites commenced feeding.
5. Linking trait-response magnitudes to trait-mediated effects
can provide estimates of how the non-lethal impacts of consumers on individual traits correspond with the their broader impacts on individual fitness, and community and ecosystem dynamics. Future work can extend the length of prey and host monitoring to link trait responses to individual fitness. Mesocosm and field experiments mirroring the experimental designs of the studies in our meta-analysis can introduce primary producers and other species in food webs into the picture, allowing associations between response magnitudes to predators and parasites and trophic flows and cascades to be quantified.
Several overlooked distinctions in how organisms respond to parasites likely led to underestimation of overall response magnitudes to parasitism in our meta-analysis. None of the included studies measured individual responses in more than one interaction stage, even though parasites commonly evoke responses in all interaction stages. Longitudinal data on individual responses to multiple consumer states will more comprehensively quantify trait response magnitudes, and may even reveal interactive effects between trait responses (Boxes 2,3). Given that immunological responses may be the most common type of host response to parasitism, non-lethal effects arising from host combat responses, in particular, are likely to be much stronger than our meta-analysis suggests. Additionally, non-lethal effects of parasitism can also arise from host phenotypic changes caused by parasite manipulation (Poulin et al. 1994; Lafferty & Shaw 2013), and even directly from parasite feeding independent of defensive responses. For instance, general energy drain (Munger & Karasov 1989; Delahay et al. 1995) or direct tissue damage caused by parasite infection can substantially impact host performance (Palstra et al. 2007). Such pathological effects are not driven by a host response, but nevertheless represent non-lethal effects. These distinct types of non-lethal effects of parasitism could collectively rival in magnitude the stronger predator avoidance that we observed. To test this hypothesis would require longitudinal data on predator responses as well, which our review indicates is also lacking in the literature. Although we expect combat responses to predation to be rare, there are exceptional cases of prey defending themselves while being ingested, particularly for some slow predators that do not kill their prey before consuming them (e.g., sea stars eating mussels). In such cases, predators could also evoke combat responses.
Considering distinct predator and parasite consumer strategies led to more comprehensive trait response predictions that did not completely align with the predator versus parasite dichotomy. However, our systematic review revealed a paucity of literature to develop this more holistic approach at understanding trait responses, suggesting a fruitful area of future research on non-lethal effects. For example, we predict parasitoids and parasitic castrators should act like predators to elicit strong pre-contact responses, but share with other parasites the ability to elicit combat responses during consumption. Micropredators, by not killing their prey when feeding, should act like parasites to elicit responses across all interaction stages. Studies that focus solely on micropredators (Kavaliers et al. 2003b, 2005) and parasitoids (Abram et al. 2019), do find that resources avoid, counter, and combat these consumers, though we still lack direct comparisons of the magnitudes of these responses with similar responses to predators and typical parasites. Furthermore there remains much to be known regarding the extent to which detectability constrains or enables responses. Testing these more specific, yet meaningful, trait response predictions requires distinguishing not just between predators and parasites, but also different types of predators and parasites. A synthesis of trait responses from multiple single-consumer studies may permit tests of these predictions, but will come with the potential expense of error from inconsistent study designs. To minimize such error, we encourage experiments that directly compare trait responses to a broader range of consumer strategies and resource species.
Regardless of how individual resources respond to predators and parasites alone, risks of predation and parasitism in the wild rarely occur in isolation. The non-additive predator and parasite effects that we observed in the meta-analysis may be indicative of priority effects, whereby the first type of exposure elicits the stronger response, though we could not test for this. Additionally, predators may frequently interfere with parasite responses by imposing stronger immediate threats to survival. Trade-offs between defences against predation and parasitism, such as adjustments in activity by tadpoles (Koprivnikar & Urichuk 2017) and shoaling behaviour by guppies (Stephenson et al. 2015), could also explain why it might be difficult to respond effectively to different simultaneous threats. However, although trade-offs between predator and parasite defences have been considered previously (Orlofske et al. 2012; Stephenson et al. 2015; Koprivnikar & Urichuk 2017), our meta-analysis highlights that anti-predator and anti-parasite responses may also be complementary in that predators and parasites can elicit similar responses, and most of the significant predator and parasite responses reported by studies were in the same direction (i.e. a reduction in the trait expression). A well-demonstrated example comes from tadpoles (Box 3), which reduce activity levels (Marino et al. 2014; Preston et al. 2014; Gallagher et al. 2019), increase refuge use (Han et al. 2011) and reduce time in foraging habitats (Koprivnikar & Penalva 2015) when either exposed to predators or infected with parasites. When responses deter both predators and parasites, avoiding predation may also inadvertently aid in the avoidance of parasites.
Larval anurans as a case study for interactive non-lethal effects of predation and parasitism
The literature on trait responses to consumers is mostly about frogs. Larval anurans (i.e. tadpoles) were the most studied resource species in our meta-analysis (Table S1), and how tadpole behaviour responds to predation risk (Relyea & Werner 1999; Van Buskirk 2001; Hossie et al. 2017) and parasitism (Han et al. 2011; Preston et al. 2014; Gallagher et al. 2019) is well-documented for several species. Questing dragonfly larvae cause tadpoles to reduce activity levels in order to avoid contact, a trait response that can reduce rates at which tadpoles are eaten. Similarly, parasitic trematodes can affect tadpole activity, but tadpole responses seem more variable and depend on the parasite state; questing trematode cercariae either elicit no change in activity (Preston et al. 2014) or increase activity (Rohr et al. 2009; Raffel et al. 2010), whereas attacking cercariae cause strong activity spikes in tadpoles that increase attack failure rates (Sears et al. 2013). However, consuming trematode states (larval metacercarial stages in tissues) cause tadpoles to reduce activity (Preston et al. 2014).
The rich literature using this exemplar system offers guidelines for how studies could be designed to quantify non-lethal effects of predators and parasites in other systems. In general, simultaneous interactions with both predators and parasites elicit reductions in activity similar to encounters with predators alone (Fig. 1a). Although this finding suggests that tadpoles prioritize predator avoidance over parasite avoidance, trait responses elicited during these multi-trophic interactions may be more complex. For example, because trematode-induced changes in tadpole activity vary by parasite state, trait responses and their non-lethal effects may vary depending on the order of encounters, otherwise known as priority effects. Being exposed to attacking trematode cercariae during or after encounters with predatory insects poses a clear trade-off between increasing activity to counter cercariae attack and decreasing activity to avoid contact with the predator. Under the hypothesis that severe fitness consequences of consumption will favour strong trait responses (Hypothesis 1 in main text), the tadpoles should sustain reduced activity levels, giving rise to non-lethal effects in the form of increased infection rates and reduced feeding rates (Fig S4a). By contrast, being exposed to attacking cercariae before encounters with predatory insects should first elicit activity increases, followed by reduced activity in individuals that become infected. If a questing predator is then encountered, infections should facilitate contact avoidance of the predator (Fig. S4b). Nevertheless, the possibility exists that encounters with questing predators while being eaten by parasites elicits additive reductions in tadpole activity that give rise to strong non-lethal effects from reduced feeding. These studies underscore how different consumer states can have different non-lethal effects, and highlight how non-lethal effects of predation and parasitism can interact.
Conclusion
Whether through fear or through infection, consumers elicit costly trait responses in their resources that give rise to non-lethal effects at the level of individuals, communities, and ecosystems. A general consumer-resource model helped to develop a framework to predict trait responses to various consumer types. Owing to differences in consumer strategies that influence when and how strongly they impact resources, we expect different trait responses to different types of predators and parasites, and therefore, different non-lethal effects. However, many consumer strategies have not yet been tested in a comparative fashion. Expanding research of non-lethal interactions beyond the predator vs. parasite dichotomy toward a broader consumer-resource perspective sets the foundation for exploring how non-lethal effects manifest in the complex, multi-trophic food webs found in real ecosystems.
Acknowledgements
We thank Gordon Research Conferences and Andy Sih, chair of the 2016 meeting “Predator-Prey Interactions”, for providing the forum that inspired this work; the EGLIDE group (Amy Pedersen, Amy Sweeny, Saudamini Venkatesan, Dishon Muloi, Alexandra Morris, Shaun Keegan and Kayleigh Gallagher), the EEGID group at University of Liverpool (Mike Begon, Greg Hurst, David Montagnes, Mark Viney, Steve Parratt), the Garner lab at the Zoological Society of London (Trent Garner, Chris Owen, Stephen Price, Goncalo Rosa, Bryony Allan), and Chris Carbone for their comments on earlier versions of the review; and Daniel Noble, Raj Whitlock, and Wolfgang Viechtbauer for advice on the meta-analysis. We also thank an unlisted author for their contributions to the idea formation and manuscript writing. This author will be named in future versions of the manuscript. Any use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government. This manuscript benefited from NSF Ecology of Infectious Diseases grant (OCE-1115965) and a grant from the Natural Environment Research Council UK (NE/N009800/1).
Footnotes
↵** note that there is an additional author that will be listed in a later version
Added note to author list specifying than an additional author will be listed in future versions. Added a complementary sentence about the unlisted author in the acknowledgements
↵1 These classifications are similar to “avoidance”, “escape attack”, and “escape capture” defined by Lima & Dill (1990) for behavioural responses of prey to predator encounters, though we propose this alternative terminology as a more comprehensive categorization of the diversity of trait responses elicited by different predators and parasites.