1Despite its wide ecological relevance, we know little about the physiological mechanisms underlying the growth vs. mortality by predation trade-off. Here, we test for two costly, potential physiological correlates of the fight-or-flight response that may contribute to the growth reduction under predation risk: induction of stress proteins (Hsp60 and Hsp70) and of antioxidant enzymes (superoxide dismutase, SOD and catalase, CAT), in larvae of the damselfly Enallagma cyathigerum.
2Under predation risk, there was a growth reduction and an increase in oxygen consumption, indicative of the fight-or-flight response. Predation risk did not affect Hsp60 levels but induced an increase in energetically costly Hsp70 levels.
3Under predation risk, levels of SOD remained constant and those of CAT decreased. Together with the increase in respiration, this should inevitably result in oxidative stress.
4Our results suggest that induction of stress proteins may contribute to the partly physiologically mediated growth reduction under predation risk and that oxidative stress is a novel cost of predation risk that may have important long-term negative fitness consequences for the prey. The latter adds to the recent insight that costs of stressors and life-history trade-offs may not always directly operate through increased energy consumption and differential allocation, but, may also work through the increased production of reactive oxygen species.
There is a growing awareness that predators may not only impact prey populations through predatory consumption, but also through non-consumptive prey intimidation by changing their behaviour, morphology and life history (Luttbeg & Kerby 2005). The impact of prey intimidation on prey demographics can be as strong as the one of direct consumption (Werner & Peacor 2003). Effects of prey intimidation on life history often take the form of reduced growth rates (Lima 1998; Benard 2004), which are assumed to be behaviourally mediated in line with the so-called growth vs. mortality by predation trade-off (Werner & Anholt 1993). There is ample evidence that prey organisms indeed reduce foraging under predation risk (Lima 1998).
Stress physiology suggests several mechanisms which may underlie a predator-induced growth reduction. Under predator stress, prey organisms typically show a series of physiological effects including an increase of respiration which enable them to react to the predator (i.e. ‘fight-or-flight’ response; Arun 2004). These effects are intended to increase short-term survival by the mobilization and shunting of energy to muscles for the actual fight or flight (Sapolsky 2002). This may have a direct energetic cost in the sense that less energy is allocated to growth thereby causing a growth reduction under predation risk. Additionally, indirect energetic costs may result from the up-regulation of the cellular metabolism. More specifically, increased metabolism likely leads to a higher need for stress proteins (Sørensen, Kristensen & Loeschcke 2003), which are involved in the housekeeping functions of the cell, and for antioxidant enzymes to deal with the increased production of reactive oxygen species (Loft et al. 1994). Reactive oxygen species lead to oxidative stress resulting in cytotoxicity and damage of cellular structure unless they are neutralized by antioxidant enzymes (Korsloot, van Gestel & van Straalen 2004).
Under predation risk, an induction of stress proteins (Hsp60 and Hsp70) has already been demonstrated in some vertebrates (Kagawa, Ryo & Mugiya 1999; Fleshner et al. 2004) and invertebrates (Pauwels, Stoks & De Meester 2005). Yet, this induction is not general (Pijanowska & Kloc 2004; Pauwels et al. 2007) and may therefore not always contribute to the predator-induced growth reduction. Unfortunately, none of the previous studies jointly evaluated an induction of stress proteins and growth reduction under predation risk. So far, no studies explored the effect of predation risk on investment in antioxidant enzymes. On the one hand, the induction of antioxidant enzymes under predation risk may contribute to a growth reduction. On the other hand, the absence of such induction or even a down-regulation of antioxidant defense to shunt energy to the fight-or-flight response may cause oxidative stress. In short, to shed light on the physiological underpinnings of the growth vs. mortality by predation trade-off we critically need studies jointly considering the costly potential by-products of the fight-or-flight response and the predator-induced growth reduction.
Damselflies of the genus Enallagma typically show a growth reduction in response to predation risk which can not be fully explained by a decreased food intake as they also show a decreased efficiency to convert assimilated food into body mass (McPeek et al. 2001; McPeek 2004). This is associated with a lower investment in energy storage (Stoks et al. 2005). In this study, we will jointly study the effects of predation risk on growth rate and evaluate the above-mentioned indirect type of costs associated with the fight-or-flight response, that is, the induction of stress proteins and of antioxidant enzymes in larvae of the damselfly Enallagma cyathigerum.
experimental set up
To study the effect of predation risk we ran a 5-day laboratory experiment where individual damselfly larvae were kept in the absence of any mortality threat or in the combined non-lethal presence of conspecifics and a fish predator. Damselfly larvae are cannibalistic and previous research has shown that they impose predator stress on each other (McPeek et al. 2001). Final instar (F0) and penultimate instar (F1) larvae of E. cyathigerum were collected in June in a pond in Opglabbeek (Belgium) containing dragonfly larvae and fish as predators. Prior to the experiment larvae were housed individually in 200 mL cups (∅ 7·5 cm) with 50 ml of filtered pond water and daily fed ad libitum with Daphnia. To avoid predator stress, larvae were visually isolated from each other. The cups were held in an incubator at 21 °C on a 14 L : 10D photoperiod for at least 10 days to allow acclimation to laboratory conditions. Previous experiments on Enallagma damselflies showed clear effects of similar laboratory predation risk treatments on growth rates after shorter acclimation periods, and this irrespective of prior experienced field predation risk (McPeek et al. 2001; McPeek 2004).
At the start of the exposure experiment, each larva was randomly allocated to the no predator treatment or the predator treatment and individually placed in a glass vial (30 mm diameter, 50 mm high) that was kept floating in a 2-L aquarium by a small piece of Styrofoam. A wooden dowel was provided as a perch for the damselfly. Four vials were kept floating per aquarium. For the predator treatment, one stickleback Gasterosteus aculeatus (standard length 5 cm), a common predator of damselfly larvae (Stoks & De Block 2000), was placed in the aquarium. This way, larvae could see each other and the fish predator but could not be eaten. Moreover, larvae received chemical cues from the fish and conspecific alarm pheromones. Enallagma larvae react to visual and chemical cues from predators and prey (Mortensen & Richardson 2008). Therefore, we daily added to each vial of the predator treatment one ml of water of the corresponding aquarium (fish cues) and 25 µL of the medium obtained by mixing one larva in 1 mL of pond water (alarm pheromones). For the no predator treatment, glass vials were taped so that larvae were visually isolated. These larvae received daily 1·025 ml of water, free of any predator cues. Note that as our aim was not to determine which predator cue caused the response, we combined these cues to mimic a relevant predation risk in the combined presence of conspecifics and fish.
During the experiment the aquaria (ten per treatment) were placed on a shelf in a controlled temperature room (21 °C, 14 L : 10D photoperiod). To ensure that each larva had ad libitum food, we adjusted the daily food ration so that at least one or two uneaten Daphnia were left the next day. To minimize any effects of aquarium, vials with larvae were daily randomly redistributed among aquaria of the same treatment.
Animals were frozen immediately in liquid nitrogen and conserved at –80 °C for further analysis. We quantified growth rate on 36 larvae per predation risk treatment, and physiological variables on a subset of these. Not all endpoints could be measured on the same larvae. On 16 larvae we measured both stress proteins and on another 14 larvae we measured oxygen consumption, and both antioxidant enzymes. Between one and six larvae from the different treatments generated zero readings for Hsp60 or Hsp70. Incubating a subset of the gels that contained zero readings with antibodies for actine (Stressgen®, Ann Arbor, MI), a positive control, suggested these zeroes were artefacts and they were excluded from further analyses (Merino et al. 2004). Final sample sizes per end point are given in Fig. 1.
We quantified growth rate based on the increase in wet mass during the 5 days of the exposure experiment. Animals were weighed at the start and the end of the experiment to the nearest 0·001 mg, 24 h after their last meal to ensure empty guts (Stoks, De Block & McPeek 2006). Daily growth rate was calculated as [ln(final wet mass) – ln (initial wet mass)]/5 days. This measure of growth rate takes into account the exponential growth curves of Enallagma larvae (McPeek et al. 2001).
Respiration rates were measured on day 4 of the exposure experiment based on oxygen consumption during 24 h, following the method of Stoks et al. (2006). Oxygen consumption was expressed as milligrams of oxygen consumed per hour per mg body mass.
We separately quantified the levels of two stress proteins, Hsp60 and Hsp70, using an immunoblot assay, following the protocol of Pauwels et al. (2007). Briefly, single larvae were homogenised in a proteinase inhibitor cocktail (Sigma®, St Louis, MO P2714) and a sample corresponding with 20 µg of protein was separated using SDS-polyacrylamide gel electrophoresis (PAGE). Stress proteins were detected using monoclonal primary antibodies (dilution 1:1000, SPA 805, SPA 820, Stressgen®) and a AP-conjugated secondary antibody (dilution 1:1000, D0486, DakoCytomation®, Glostrup, Denmark). The optical density (OD) of stress protein bands on the membrane was quantified on digitized images using the software package Image ProPlus®. To correct for variation between blots, we ran a control sample of 1 µL HeLa Cell Lysate (Heat shocked; Stressgen®) on every blot. For both Hsp60 and Hsp70 we could show that the response curve for optical density against concentration was linear (both R2 > 0·99, P < 0·05).
We quantified the activity of superoxide dismutase (SOD) and catalase (CAT). These are the two most important antioxidant enzymes in insects involved in the defense against two important reactive oxygen species: superoxide anions and hydrogen peroxide (Felton & Summers 1995; Korsloot et al. 2004). Under oxidative stress, it is expected that the activities of both antioxidant enzymes increase (e.g. Mittapalli, Neal & Shukle 2007). Both enzymes were scored spectrophotometrically following the protocol of De Block and Stoks (2008). One unit of CAT activity was defined as the amount that decomposes 1 µmol H2O2/min at 30 °C and pH 7·4 per mg protein. One SOD unit is defined as the amount of enzyme necessary to decrease the colorimetric reaction to 50% of maximum at 37 °C per mg protein.
We ran separate anovas for each response variable with predation risk as the independent variable. Larval instar and its interaction with predation risk were also included to correct for potential differences between instars but were never significant (all P > 0·19). Initially, the models for both stress proteins contained two covariates. The first covariate, protein content, corrected for any remaining differences in the amount of protein loaded. The second covariate, optical densities of the HeLa samples, corrected for variation among blots. As none of the covariates were significant (both P > 0·19), these covariates were removed from final analyses.
The damselfly larvae grew about 47·4% more slowly under predation risk (F1,69 = 4·65, P = 0·034, Fig. 1a). Under predation risk larvae had a higher respiration rate, as indicated by a 36·1% increased oxygen consumption (anova, F1,24 = 9·92, P = 0·0043, Fig. 1b).
Of the two stress proteins, Hsp70 levels were 40·4% higher under predation risk (F1,19 = 5·76, P = 0·027, Fig. 1d). Hsp60 levels were not affected (F1,23 = 0·25, P = 0·62, Fig. 1c). The activity of CAT was 23·2% lower under predation risk (F1,24 = 4·63, P = 0·042, Fig. 1f), but there was no detectable change in the activity of SOD in response to predation risk (F1,24 = 1·52, P = 0·23, Fig. 1e).
As shown before in Enallagma larvae (McPeek et al. 2001; McPeek 2004; Stoks et al. 2005), we found a reduction in growth rate under predation risk. Previous studies also showed that the growth reduction in Enallagma damselflies can not be fully explained by a reduced food intake, suggesting a role for physiology as indicated by the reduced efficiency to convert assimilated food into biomass under predation risk (McPeek et al. 2001; McPeek 2004). This partly physiologically mediated life-history response may be a by-product of an accelerated metabolism under the fight-or-flight response which besides an increase of the heart beat rate and respiration summarizes a number of metabolic effects including the induction of hyperglycaemia, glycolysis and lipolysis (Randal, Burggren & French 2001; Hawkins, Magurran & Armstrong 2007). Surprisingly, only few other studies so far reported an increased oxygen consumption under predation risk (Woodley & Peterson 2003; Beckerman, Wieski & Baird 2007; see also Millidine, Armstrong & Metcalfe 2006). Alternatively, the growth reduction and physiological changes were due to the visual cues which might have transmitted information about densities and therefore induced stress associated with resource competition. This is, however, unlikely because larvae were placed individually in vials with ad libitum food and interference competition has been shown not important in Enallagma larvae (Anholt 1990).
In line with the fight-or-flight response (as indicated by the increased metabolism), there was an induction of Hsp70 under predation risk. Such predator-induced increase of stress proteins has been previously reported but is not general (see Introduction). Stress protein production is a likely response to the accelerated cellular metabolism under predation risk and the associated increased challenge to maintain homeostasis (Sørensen et al. 2003). Additional, non-exclusive, mechanisms may also underlie stress protein induction under predation risk. First, the Hsp70 induction could be part of the generalized stress response that serves to protect the cell from cytotoxic stress response substances (e.g. cytokines) (Fleshner et al. 2004). Alternatively, the release of Hsp70 under predation risk exposure could serve as a danger signal to the immune system, facilitating a faster and more directed response if injury would be followed (Pockley 2003). Finally, stress proteins may also play a role in generating the fight-or-flight response by facilitating the binding of messenger molecules and receptors as suggested by Kagawa and Mugiya (2002).
Several experiments revealed that the expression of stress proteins is costly (overview in Sørensen et al. 2003) and may therefore have contributed to the reduced growth rates. The costs of this induction are caused by the extensive energy needed for production of stress proteins and the negative effects of high stress protein concentrations working through their catabolic function (Feder & Hofmann 1999; Sørensen et al. 2003). It is unlikely that the up-regulation of Hsp70 and the down-regulation of CAT cancelled out energetically. The up-regulation of Hsp70 was much more pronounced (+40·4%) than the down-regulation of CAT (–23·2%). Moreover, Hsp70 is not only costly to produce but also costly to maintain given its catabolic function. Further, the opposite scenario where the growth reduction would have generated an up-regulation of Hsp70 is highly unlikely as, if anything, a growth reduction would not disrupt cellular homeostasis and as a result not increase the need to up-regulate costly stress proteins. Increased energy allocation toward stress proteins therefore likely contributed to the previously observed reduced efficiencies with which assimilated food was converted in body mass in Enallagma damselfly larvae under predation risk (McPeek et al. 2001; McPeek 2004). In line with this, Menge et al. (2002) reported that increased levels of Hsp70 under stressful conditions (high temperature, low moisture) in whelks were associated with a growth reduction. Other studies presented particularly strong evidence for a causal link between Hsp70 up-regulation and growth reduction as they directly manipulated Hsp. For example, Drosophila larvae with extra copies of the Hsp70 gene have a decreased growth rate compared to control larvae (Krebs & Feder 1997), and Drosophila cells engineered to express high Hsp70 levels grow more slowly than control cells (Feder et al. 1992). Given the generality of the fight-or-flight response (Sapolsky 2002; Roeder 2005) and the link between one of its essential components, an increase in metabolism and induction of stress proteins (Sørensen et al. 2003), the latter may be a general mechanism contributing to the partly physiologically mediated growth reduction under predation risk. Of course, under predation risk other costly physiological mechanisms may be initiated that also shunt energy away from growth and thereby further reduce growth efficiency.
In contrast with the predator-induced induction of Hsp70, there was no change in the amount of Hsp60 between the control and the predation risk treatment. This may reflect the different physiological roles of Hsp60 and Hsp70. Although both stress proteins promote efficient folding of newly synthesised proteins, especially Hsp70 is found associated with misfolded and unassembled proteins (Pockley 2003).
Despite an increase in oxygen consumption, which is typically linked to an increased production of reactive oxygen species (Loft et al. 1994), there was no up-regulation of the two key antioxidant enzymes under predation risk in insects. Instead we observed constant levels of SOD and even a decrease in CAT levels, which may be explained by the increased need to shunt energy towards the muscles to prepare the body for a fight-or-flight response. The alternative scenario whereby the growth reduction generated the down-regulation of CAT seems unlikely because growth rate was not positively coupled to cellular metabolism and because a growth reduction would, if anything, have increased energy availability for other processes. The different response of SOD and CAT might be due to their different function. SOD catalyses the formation of the superoxide anion into hydrogen peroxide, which is in turn transmutated into water and oxygen by CAT (Felton & Summers 1995). As free radicals are more reactive and thus more dangerous than hydrogen peroxide (Korsloot et al. 2004), animals may have prioritized avoiding a reduction in SOD. Because antioxidant enzymes are costly, there is under standard conditions a balance between their levels and those of reactive oxygen species so that no oxidative stress occurs (Felton & Summers 1995; Korsloot et al. 2004). Therefore, the combination of the increased levels of reactive oxygen species (as suggested by the increased respiration) and the constant (SOD) or even decreased (CAT) levels of antioxidant enzymes should result in oxidative stress. Oxidative stress may negatively affect fecundity and life span in insects (Ahmad 1992). Interestingly, oxidative stress may potentially also reduce growth rate (Davies 1999). Therefore, our results point to a novel cost of predation risk which may have long-term fitness consequences even when predation risk has already disappeared. Given the generality of the fight-or-flight response we hypothesize oxidative stress may be a widespread, yet overlooked cost of predation risk.
There is increasing awareness that predators may strongly affect prey populations not only through direct consumption but also through non-consumptive prey intimidation (Werner & Peacor 2003; Luttbeg & Kerby 2005). Research on the latter mechanism has largely been biased towards studying effects on behaviour, morphology and life history. Our results on physiological prey responses to predation risk have two important implications for the research on non-consumptive predator-prey interactions. First, there is increasing evidence for a combination of behavioural and physiological mediation of the growth vs. mortality by predation trade-off (see Introduction). Our results add to this by identifying a potential mechanism for the latter option working through induced Hsp70 production. Growth reduction under predation risk is not always present (Benard 2004) and an interesting avenue for further research would be to explore to what extent behavioural and physiological mechanisms may predict its occurrence. This has large ecological relevance as the growth vs. mortality by predation trade-off has wide ramifications for populations which may cascade up to the community level (Werner & Peacor 2003). A better mechanistic understanding of this trade-off is needed (Noonburg & Nisbet 2005) and as indicated by our results the identification of costly physiological by-products in response to the fight-or-flight response may play an important role here. Second, we identified a novel cost of predation risk in terms of increased oxidative stress due to a lowered total investment in antioxidant defense under predation risk. This may have several important long-term effects on fitness that need to be addressed. For example, oxidative stress may affect brain and muscle tissue (Felton & Summers 1995; Droge & Schipper 2007), so that predation risk may have the potential to negatively affect the future ability of the prey to recognize and escape predators and to reduce its lifetime reproductive success. More general, this may add to the recent insight that costs of stressors and life-history trade-offs may not always directly operate through increased energy consumption and differential allocation, but, given the associated increase in oxidative stress, may also work through the increased production of reactive oxygen species (Alonso-Alvarez et al. 2007; De Block & Stoks 2008).
We thank Ine Swillen, Luc De Meester, Marjan De Block and two anonymous referees for improving this paper. Cathy Duvivier helped with statistical analyses. SS is a scholar funded by a PhD grant of the Institute for the Promotion of Innovation through Sciences and Technology in Flanders (IWT-Vlaanderen). This study was funded by FWO research grant G.0269·04 and projects OT/04/23 and GOA/2008/06 of the KULeuven Research Fund.