Time-lagged effect of predators on tadpole behaviour and parasite infection

Abstract

Prey should adjust their defences against natural enemies to match their current level of risk and balance other needs. This is particularly important when optimal defences represent trade-offs, as is the case with many predator-induced trait-mediated indirect effects (TMIEs) that are antagonistic to those promoting host resistance to parasites and pathogens. However, trade-offs may depend on whether different natural enemies are present simultaneously or represent temporally discrete threats. We found that larval amphibians (Anaxyrus americanus) previously exposed to predator cues did not engage in anti-parasite behaviours (activity increases) in response to a current risk of infection by a pathogenic trematode parasite compared to controls, resulting in higher infection intensities. This suggests that the memory of the likely more lethal threat (predation) had greater influence, maladaptively dampening tadpole activity. Incorporating complexity inherent in natural systems, including spatial and temporal overlap, is necessary to better understand natural enemy ecology and how TMIEs relate to infectious diseases.

1. Introduction

Beyond direct effects on prey populations through consumption, predators can also affect infectious disease dynamics. Best known is the presumed greater predation upon, and thus transmission by, prey that exhibit behavioural alterations stemming from infection with trophically transmitted parasites [1], and the ‘healthy herds’ hypothesis posits a key role for predation through the culling of infected prey, reducing pathogen transmission [2]. Apart from direct consumption, predators can also have trait-mediated indirect effects (TMIEs), including life history and behavioural alterations whose consequences for prey may be more ecologically significant than those resulting from consumption [3,4]. Because parasites are ubiquitous in most natural systems, affecting hosts in many ways [5], and emerging infectious diseases have become a critical concern for humans and wildlife [6], there is increasing recognition that predator TMIEs may have important influences in this context [7]. Notably, behavioural defences against parasites can conflict with anti-predator behaviours [7–10]: many prey species decrease their activity when there is a risk of predation [11–13], but parasite evasion or removal (e.g. grooming) often requires increased activity [7–10,12,13].

Anti-predator behaviours should be matched to predation risk so as to not adversely affect critical activities including foraging and mating (the ‘threat-sensitive predator avoidance hypothesis’), with various observations of such plasticity [14,15]; however, different natural enemies do not necessarily overlap in space or time, thus examinations of predator TMIEs on host parasitism must reflect this disassociation. In the few studies to date, cues separately representing risk of parasitism and predation tend to elicit similar spatial avoidance responses (e.g. [16,17]). By contrast, hosts/prey typically exhibit predator avoidance behaviours when presented with a simultaneous threat of predation and parasitism, or often suffer from increased predation (e.g. [8,12,13,17]). This presumably reflects that a one-time encounter with most predators is likely more lethal relative to parasites [7,18], especially because parasite-induced pathology is often dose-dependent. While there are too few studies of predator-induced TMIEs on host infection susceptibility to make generalizations, such efforts are critical to unifying natural enemy theory [7], and ecologically realistic considerations such as temporal disassociation of natural enemies are necessary.

Larval amphibians represent one of the best-studied groups with respect to plasticity in response to predation threat, with a remarkable ability to learn, selectively forget and fine-tune their anti-predator behaviours corresponding to variation in risk (e.g. [15,19]). While various aquatic predators present a hazard for tadpoles [11], so do pathogenic parasites, such as trematodes (flatworms) [20]. Tadpoles exhibit effective behavioural defences against free-swimming trematode infectious stages (cercariae), often increasing their activity to evade and remove these [8,9,21]. When simultaneously facing threats of parasitism and predation, tadpoles generally respond more strongly to the latter via activity reductions that increase their infection loads [9]. Because prey should match their anti-predator behaviours to the current degree of threat, this may extend to other natural enemies such as parasites. Here, we temporally disassociated the risk posed by predators and parasites to tadpoles to investigate whether hosts/prey responded more strongly to an immediate risk of parasitism or to the memory of predation threat.

2. Material and methods

Two clutches of American toad (Anaxyrus americanus) eggs were locally collected from a pond in southern Manitoba that lacked larval odonates or snail hosts of trematodes, hatched and group-housed in the laboratory within 15 l plastic tubs containing dechlorinated water on a 14 L : 10 D cycle at 21°C. Tadpoles were fed boiled spinach ad libitum until reaching approximately Gosner [22] stage 26 when 50 were haphazardly selected and transferred into individual plastic tubs (16.5 cm diameter × 11.4 cm height) containing 1 l of water, with 25 assigned to receive either the control/sham or predator cue. The predator cue mixed strained conspecific alarm cue (10 euthanized and macerated conspecifics) with 500 ml of water from a container housing five larval dragonflies (Anax sp.) and was kept frozen in 150 ml aliquots that were thawed immediately prior to use. The control/sham consisted of thawed water only. Each tadpole received either 5 ml of predator cue or the sham in three separate instances separated by 48 h over 6 days. Tadpoles were maintained as described above within separate tubs during these exposures, followed by another 7 days without any additions after a complete water change and container rinse.

We locally collected Stagnicola sp. snails and screened them for trematode infections based on established procedures [21], using those infected with Echinoparyphium sp., which forms cysts in tadpole nephric systems that can cause mortality and sublethal pathology [20]. Seven days after the last addition of predator cue or sham, we collected emerged cercariae from 25 snails and pooled these before transfer in groups of 25 into separate microcentrifuge tubes. To examine tadpole anti-parasite behaviour, we followed the procedures of previous studies [21]. Briefly, we placed individual tadpole tubs under two digital recording cameras (5/camera) in random order. After a 15 min acclimation period (no recording), we recorded tadpole activity in the absence of cercariae for 15 min, followed by another 15 min of recording after the addition of cercariae (1 tube/tadpole). All recordings occurred on 1 day, and no cercariae were older than 5 h at their time of addition. After recording, tadpoles were maintained for 48 h with cercariae, euthanized in a 1% solution of buffered MS-222, and individually frozen for later necropsy to quantify infection load and Gosner developmental stage.

We evaluated whether each tadpole was active (swimming) at 20 s intervals for the 15 min periods with and without cercariae present, respectively, converting this into the proportion of time points (/45) active. Activity data were arcsine-square-root transformed before analysis with a repeated-measures general linear model (GLM) in SPSS 24.0. We used a factorial design with parasite presence as the within-subjects factor, predator cue exposure as the between-subjects factor and Gosner stage as a covariate. Two separate generalized linear models (GLZM) analysed the relationship between individual tadpole activity with cercariae present and resulting cyst count (Poisson distribution with log link), and the effect of predation cue on infection intensity, with developmental stage as a covariate. Separate GLMs examined the effect of predation cue on stage and mass.

3. Results

There was a significant interaction between predation cue exposure and parasite presence (F_1,42 = 5.51, p = 0.024); tadpoles receiving the sham increased their activity with cercariae present whereas those given the predation cue did not (figure 1). Because tadpoles exposed to the predation cue were less active with and without cercariae present than those receiving the sham (F_1,42 = 10.22, p = 0.003), there was only a strong overall trend towards greater activity by all individuals when cercariae were present (F_1,42 = 3.93, p = 0.054). There was a strong negative relationship between tadpole activity in the presence of cercariae and infection intensity (χ² = 24.16, d.f. = 1, p < 0.001; figure 2). Predator cue-exposed tadpoles had a significantly higher parasite load compared to sham-exposed individuals (χ² = 6.76, d.f. = 1, p = 0.009), but did not differ in their developmental stage (F_1,42 = 0.998, p = 0.323) or mass (F_1,42 = 0.093, p = 0.762). Mean infection intensity (±s.d.) was 14 (±3.1) for sham-exposed tadpoles and 17 (±2.9) for those given the predation cue. Note that 44 of 50 tadpoles survived until the end of the experiment.

Figure 1.

Mean (±s.e.) activity level in the absence/presence of trematode parasite infectious stages for tadpoles exposed to a predator cue or sham 7 days earlier. (Online version in colour.)

Figure 2.

Significant negative relationship between tadpole activity in the presence of trematode parasite infectious stages and infection load.

4. Discussion

Temporal separation of the risk posed to larval amphibians by two different natural enemies (predators and parasites) resulted in behavioural defences best matched to the enemy likely representing the more lethal single-encounter threat (predators), corresponding to previous studies with simultaneous risk [8,12,17]. As behavioural defences against these two natural enemies are antagonistic here (activity decrease against predators versus increase in the presence of trematode infectious stages), the choice to respond to predation risk rather than the current threat of parasitism resulted in greater parasite loads given the inverse relationship with individual activity when cercariae were present. Higher infections in hosts/prey faced with spatial and temporal overlap in predation and parasitism risk have been previously observed with amphibians and other parasite–host–predator systems, as has increased predation on prey engaging in anti-parasite behaviours at odds with those best-suited for predator avoidance (e.g. [8,9,13]). However, predation threat may also cause immunological and morphological alterations that increase prey infection susceptibility [7], and should be investigated alongside behavioural changes.

The threat-sensitive predator avoidance hypothesis suggests that prey should adjust relatively plastic responses, including behavioural defences, to match current predation risk, and larval amphibians can respond to temporal variation, trends in risk and predator identity (e.g. [14,15,19,23]). While tadpoles retain the memory of predation threat (defined as retention of acquired information, e.g. [19]) for significant periods of time, with even embryonic exposure shaping later behaviour, its influence can also fade if risk is uncertain or low [19,23]. This forgetting is adaptive because the retention of obsolete information regarding predation risk can overestimate its importance, causing prey to engage in risk-aversive behaviours that trade-off with other important activities [19,23]. Larval amphibians can also make context-dependent decisions with respect to anti-predator behaviours, balancing these with other needs such as foraging based on their energetic state [24].

Despite these examples of behavioural defences finely tuned to predation threat, including assessment of current risk, our results suggest that flexibility in response to other natural enemies may be constrained. Although tadpoles had not experienced predator cues for over a week, they still responded more strongly to them than a current threat of parasitism. One possibility is that conflicting behavioural responses to natural enemies are an ‘all or nothing’ decision based on potential for harm regardless of present risk, and a single encounter with a larval odonate is likely more lethal relative to that with trematodes [7,18]. This considerably broadens the scope for predator-induced TMIEs to affect infectious disease dynamics if natural enemies need not be simultaneously present. The temporal window for these TMIEs to affect host susceptibility could thus be much larger than expected. Future work modifying the risk of parasitism by increasing the infectious dose, or employing more pathogenic parasites, will be helpful to elucidate the context-dependency of behavioural defences against natural enemies, as will studies that vary the time elapsed since predation risk. By incorporating greater complexity related to risk, such as the temporal disconnect between threats here, we will achieve a better understanding of natural enemy ecology and the possible joint and separate effects on hosts and prey.

Ethics

All work was carried out in accordance with the guidelines of the Canadian Council for Animal Care with institutional approval by the local animal care committee (permit no. 2012T02).

Data accessibility

Data can be accessed at http://dx.doi.org/10.5061/dryad.157c8 [25].

Authors' contributions

J.K. designed the study, carried out the analysis and drafted the manuscript. T.M.Y.U. carried out data collection, participated in the discussion of results and in writing of the manuscript. J.K. and T.M.Y.U. approved the final version to be published and agreed to be accountable for all aspects of the work.

Competing interests

The authors declare no competing interests.

Funding

This work was supported by a NSERC Discovery (grant no. RGPIN 05566) to J.K. and a NSERC USRA to T.M.Y.U.