SEARCH

SEARCH BY CITATION

Keywords:

  • adaptation;
  • anti-predator behaviour;
  • body shape;
  • induced defence;
  • phenotypic plasticity

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    Many biological textbooks present predator-induced morphological changes in prey species as an example of an adaptive response, because the morphological change is associated with lower predation risk. Here we show that the adaptive morphological response observed in many systems may actually be an indirect effect of decreased activity – which reduces the predation risk – rather than a direct adaptive response.
  • 2
    One of the classical examples comes from crucian carp, where the presence of pike leads to a deeper body. We manipulated pike cues (presence and absence) and water current (standing and running water) and found that both standing water and pike cues similarly and independently induced a deeper body.
  • 3
    Since the presence of pike cues as well as standing water might be associated with low swimming activity, we suggest that the presence of pike causes a reduction in activity (antipredator behaviour). Reduced activity subsequently induces a deeper body, possibly because the energy saved is allocated to a higher growth rate.
  • 4
    Our result suggests that even if morphological change is adaptive, it might be induced indirectly via activity. This important conceptual difference may be similar in many other systems.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Phenotypic plasticity is the ability of an organism to express different phenotypes in response to different environmental conditions (Pigliucci 2001). In some species, plasticity can explain much of the variation in morphology among individuals (Schmitt 1993; Mittelbach, Osenberg & Wainwright 1999). Predator-induced morphological defence is a well-studied aspect of phenotypic plasticity, where the presence of a predator induces morphological changes, such as protective spines, armour and body shape. Several such induced morphological changes have been shown to be adaptive: they increase the probability of survival from attacking predators (Tollrian & Harvell 1999). In many predator–prey systems, the general interpretation of the induced responses is that they are directly induced by the predators’ presence since manipulations of chemical compounds are sufficient to induce the defence (Tollrian & Harvell 1999).

We suggest that induced morphological defence in animals may actually be an indirect effect of changed behaviour, rather than a direct adaptive response, because activity, growth rate and body morphology are sometimes positively correlated. The logic behind our suggestion rest on the fact that, in addition to changes in shape, predators also commonly cause antipredator behaviour of their prey, and a common response is a decrease in activity (Lima & Dill 1990). Reduced activity is often adaptive simply by reducing the predator–prey encounter rate, which decreases the probability of a prey being killed (Werner & Anholt 1993) but may also save energy, which then can be allocated to somatic growth (Tolley & Torres 2002; Donovan & Gleeson 2006) or development of defensive structures, such as a higher body depth, stronger or larger defensive spines etc. For example, under some environmental conditions a reduction in activity of tadpoles in the presence of predators is associated with a deeper tail (which reduces predation risk) and a higher growth rate (Van Buskirk & McCollum 2000; Relyea 2004; Laurila, Lindgren & Laugen 2008).

A textbook example of predator-induced morphological change is the deeper body of crucian carp (Carassius carassius Linnaeus) in the presence of pike (Esox lucius Linnaeus) (Brönmark & Miner 1992; Brönmark & Pettersson 1994; Vøllestad, Varreng & Poleo 2004; Andersson, Söderström & Johansson 2006). The deeper body is assumed to be adaptive by decreasing the risk of pike predation (Nilsson, Brönmark & Pettersson 1995). However, it has also been argued that the deeper body could be an activity-mediated indirect consequence, of the energy saved by decreased activity being allocated to a higher growth rate (Holopainen et al. 1997; Vøllestad et al. 2004; Andersson et al. 2006). At high growth rates, fish typically show a relatively greater increase in height than in length (Vøllestad et al. 2004; Charo-Karisa et al. 2007, but see Olsson, Svanbäck & Eklöv 2007).

To examine whether a deeper body of crucian carp is induced by predator cues or by activity, we independently manipulated swimming activity by subjecting crucian carp to a water current treatment, which was crossed with a pike cue treatment. Our 2 × 2 design allows us to separate the effects of activity and pike cues: if the induced deeper body shape is an indirect effect mediated through activity, we expect standing water and the presence of pike cues to similarly (and independently) change the body shape of carp.

Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Crucian carp in the size range of 6·8–9·2 cm were collected from a small pond in Umeå (63°48′N, 20°14′E), northern Sweden. Crucian carp was the only fish species present in the pond, so this population was naive to predators. To examine the effect of water current and pike cues on body shape, the fish were raised in 65 L circular tanks (diameter 43 cm, height 35 cm) for 67 days during the summer. The experiment was run in a greenhouse which provided natural light conditions and a temperature range of 15–25 °C. A plastic cylinder (diameter 24 cm, height 37 cm) was positioned in the centre of each tank and the fish were held outside this cylinder, giving them a circular-shaped arena.

The fish were subjected to two water current treatments, standing water and water current, and two predator treatments, presence and absence of chemical cues from a pike predator, giving a 2 × 2 factorial design. Dead chironomids bought from a commercial dealer were added daily as food for the crucians at an amount corresponding to 6% of their body weight. All chironomids were eaten within 30 s and were readily taken in the water column. This suggests that difference in energy expenditure due to foraging efficiency that might occur due to water current is small relative to the total energy expenditure difference between crucian carp swimming in water current and no water current. Three individuals were held in each tank and they were weighed before the start of the experiment. The tanks were ordered in eight replicates of 2 × 2 groups, giving a total of 32 tanks, and 96 crucian carp. Within each group, two tanks had standing water and two tanks had water current. Half of the groups received chemical cues from a pike predator while half received no chemical cues from a pike predator. Water was pumped from four large 150-L tanks and then drained back into the large tanks. Each large tank supported two groups of small tanks with water. Two of the large tanks held one pike each (size 1 kg) which provided the pike cues. The pike were fed two crucian carp twice a week. In tanks with water current, the water was let in from an incoming plastic tube, attached along the side of the tank in a horizontal position, close to the water surface. This arrangement created a slow, directional water current (0·035 ms−1) within the tanks. The water current is probably within the range of what crucian carp encounter in natural waters because they are known to occur in rivers with flowing water (e.g. Pollux et al. 2006). In addition, our water current is within the range of water current that can be found in lakes as a result of wind (Wetzel 1975). In tanks without water current, the incoming tube entered through the central cylinder and the water trickled out through a gap between the bottom of the cylinder and the floor of the tank. Thus, no directional water current was present in these tanks. Water from all tanks was drained back to the large containers through a pipe situated at the water surface, and the same amount of water was let in and out in the tanks with and without directional water current. At the end of the experiment, all crucian carp were taken out of the tanks, anaesthetized (MS 222) and weighed. The fish were then placed on a styrofoam plate with the fins fixed to the plate, and photographed to provide digital photos for morphological analysis.

It could be argued that our treatment has some degree of pseudoreplication because we used only four large water source tanks. However, the body shape changes induced by pike in the crucian carp has been shown in many studies; see references in Andersson et al. (2006), and we therefore feel confident that the tank effect did not bias our results. In addition, we found no significant effect of pike tank and tank group in our analysis on shape (results not given).

Body shape was analysed using geometric morphometrics. Ten landmarks (Fig 1) were sampled from digital photos using the software tps-dig (Rohlf 2003a). Shape variables and centroid sizes were generated with the software tps-relw (Rohlf 2003b), see Andersson et al. (2006) for further details on methods and logic behind this process. The effect of water current and pike cues on body shape was tested with a nested mancova using the following model:

image

Figure 1. Landmark configuration (dots) on the crucian carp used for the morphometric analysis. Note than one landmark is positioned at the centre of the eye.

Download figure to PowerPoint

  • body shape = predator treatment + water current treatment + size + (predator × water) + tanks(predator × water),

where shape variables [uniform components (2 variables) and partial warps (14 variables)] were used as dependent variables. Size is the centroid size measured at the end of the experiment. Size was used as a covariate and tanks were nested within the interaction of predator and water treatments (Langerhans & DeWitt 2004). In an initial analysis, the interactions between size and water current and size and predator treatment was included but these factors were dropped from the model since they were nonsignificant. To visualize shape differences between treatments, we produced deformation grids using the software tps-spline (Rohlf 2003c). Main effects were interpreted by producing deformation grids describing the consensus fish (mean fish shape) for the two main factors. We did this by pooling all individuals from the pike treatment and all individuals from the nonpike treatment, which gave us two consensus fish shapes which we compared: one for the pike treatment and one for nonpike treatment. Similarly, consensus fish from the water current treatments were produced and compared by pooling all individuals from the water current and all individuals from the tanks without water current. In addition, we also produced consensus shapes of all the four treatment combinations (i.e. standing water fish with and without pike cues and current water fish with and without pike cues), which allowed us to interpret the interaction terms. Weight was analysed with a two-way anova using predator and water current as factors. We used start weight to compare initial weight differences and we subtracted final weight from start weight to estimate mass gain.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Both predation treatment and water current had a significant effect on body shape (Table 1). Graphical interpretation of the deformation grids (Fig. 2) revealed that pike cues and standing water induced a higher body. Similarly, water current induced a shallow body similar to that observed in the absence of pike cues. The interaction factor (water current × pike cue) was not significant (Table 1), indicating that the two factors were additive. This additive effect is reflected in the deepest body in fish from standing water with pike cues (panel D) and the shallowest body shape in fish from running water without pike cues (panel A). Body shape of fish in the water current treatment with pike cues (B), and in the nonpike treatment with standing water (D), where intermediate in shape with regard to the two extreme body shapes (A and D). Size had no significant effect on shape, since we found no significant effect of treatment on centroid size.

Table 1.  Results from a mancova on shape variables examining the shape variation between the pike and water current treatment
FactorWilk's λF valued.f.P value
Pike0·552·3216,45 0·013
Water0·325·8516·45< 0·001
Size0·701·2216,45 0·29
Water × Pike0·661·4516,45 0·16
Tanks (Water × Pike)< 0·0011·12448,689 0·093
image

Figure 2. Deformation grids showing shape changes induced by water current and pike cues. The central landmark configuration shows the mean fish shape (consensus fish). The head of the fish is to the left and the lines between the outer landmarks were added to help the interpretation of fish shape changes. The vertical central column shows the shape changes when shifting from water current to standing water. The horizontal central row shows shape changes when shifting from no pike cues to pike cues. The corners show the actual shape of fish from the 2 × 2 treatment, i.e. water current and no pike cues (A), water current and pikes cues (B), and so on.

Download figure to PowerPoint

At the start of the experiment, mass of crucian carp did not differ significantly between treatments (P > 0·10 for both factors and interaction term), but after the experiment a two-way anova on weight change showed a significant effect of pike cues (P = 0·027, F1,28 = 5·41) as well as water current (P < 0·001, F1,28 = 22·47) (Fig. 3). In standing water, crucians increased their weight, and this was increased further in the presence of pike cues. In water current, crucians lost weight, but the weight loss was less in the presence of pike cues. Again, no significant interaction between pike cues and water current was found (P = 0·12, F1,28 = 1·71), suggesting that the effects on weight are additive.

image

Figure 3. Change in mass (g) of the crucian carp between the start and end of the experiment, in the presence and absence of pike cues, and under standing water (solid line) and water current (hatched line). Error bars denote SE.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The deeper body of crucian carp is a classical example of predator-induced defence, cited in many textbooks as a direct effect of predators on prey morphology (Wootton 1999; Allen 2006). We have shown here that a similar change in body shape is induced by water current, which has nothing to do with predators. Because the effects of pike and water current treatments appear statistically independent, our results show that the same shape change is induced by these two treatments. Since predation risk and water current probably have similar effects on activity, we find little evidence for the widely held assumption that shape change would be directly induced by the presence of the predator. Instead, we suggest it can be explained as a consequence of antipredator behaviour.

In a separate experiment (Supporting information), we found that pike cues led to reduced activity in the crucian carp population we used. At least six other independent studies have shown that crucians reduced their swimming activity in the presence of pike cues (Holopainen, Tonn & Paszkowski 1991; Pettersson & Brönmark 1993; Pettersson, Andersson & Nilsson 2001; Holopainen et al. 1997; Vøllestad et al. 2004; Andersson et al. 2006). Furthermore, these five studies also found that reduced activity in the presence of pike cues – whether defined as refuge use as in Vøllestad et al. (2004) or as swimming speed as in Andersson et al. (2006) – was associated with a high body shape in crucians.

Our experiment goes one step further than the preceding studies on this phenomenon, since we simultaneously manipulated pike cues and swimming activity, independent of each other. Although we were unable to quantify swimming activity in our fish shape experiment, qualitative observation during the experiment suggests that fish subjected to the water current always faced in the direction of the flow while fish without a water current were standing close to the bottom facing different directions. This suggests that the water current forced the fish to swim actively. If the induced body shape would (partly) be a direct effect of predator cues, we would have expected a significant (pike × water current) interaction. Of course, despite our eight replicates with four unique treatments and three fish per treatment, we cannot exclude the possibility that with still larger sample sizes a significant interaction term would have appeared.

Fish shape changes could also be adaptations to water currents because a shallow body shape could mean less drag and an adaptation to a water current (e.g. Imre, McLaughlin & Noakes 2002). In our experiment, we cannot distinguish between such an adaptation and an activity-mediated shape changes. But we suggest that energy expenditure of fish subjected to water current resulted in a different growth rate, which in turn affected body shape. Fish have to adjust swimming speed (activity) in order to stay in position at different water velocities, so that fish subjected to different water velocity will differ in activity level. Thus, activity may mediate the effect of water velocity on body shape as well. Hence, it would be interesting to study the effect of growth rate differences resulting from energy expenditure on shape change in fish.

Currently, we have no explanation for why fish should grow more in depth rather than in length when more energy is available for growth. Fish in general grow allometrically, and a higher growth rate results in a higher body depth (Vøllestad et al. 2004; Charo-Karisa et al. 2007). There might be an allometric relationship between length and depth that is caused by genetic constraints related to other factors.

Further support for the association between swimming activity and body shape is available for other fish species. A similar increase in body depth in response to the presence of pike cues has been shown in perch, Perca fluviatilis (Eklöv & Jonsson 2007). In contrast, Eklöv & Jonsson (2007) found that roach, Rutilus rutilus, showed a deeper caudual peduncle in the presence of pike cues. The differences in response between perch and roach can be explained by differences in antipredator behaviour. While perch respond to the presence of pike cues by reducing activity, roach increase their activity in response to pike cues (Christensen & Persson 1993). Interpreted in light of the present results, a decrease in activity in perch results in a deeper body while a tendency for the opposite occurs in roach.

We do not claim that all cases of predator-induced defence are indirect and mediated by reduced activity. But we note that a reduction in activity in the presence of predators is associated with a larger size and/or body shape changes in ciliates (Kusch 1993), amphibians (VanBuskirk & McCollum 2000; Laurila et al. 2008), gastropods (Crowl & Covich 1990), daphnids (Boersma, Spaak & De Meester 1998), and insects (Mikolajewski et al. 2005). For example, in daphnids – another much cited example of predator-induced morphological defence –Scheiner & Berrigan (1998) found that the presence of chemical cues from a predator resulted in reduced metabolic rate (lower oxygen consumption) and increased reproductive rate. Scheiner and Berrigan suggested that the daphnids achieved this by reducing activity or by increasing assimilation efficiency and changing allocation from maintenance to growth and reproduction.

However, the presence of predators may also result in reduced activity and a smaller body (Relyea 2002; Laurila et al. 2008), for example, prey might experience physiological stress in the presence of predators, resulting in reduced growth efficiency (Boonstra, McColl & Karles 2001; Trussell & Nicklin 2002; Stoks et al. 2005). Hence, activity and growth might not be linked physically and can therefore be decoupled resulting in less growth despite a lower activity (McPeek 2004; Stoks et al 2005; Steiner 2007a).

We note that many association between activity, growth rate body size and tail depth are possible. These different associations might be dependent on experimental conditions (e.g. food) or population origin (McCollum & Van Buskirk 1996; Relyea 2004; Steiner 2007b; Laurila et al. 2008), and therefore careful experimental conditions examining our suggestion are called for. Benard (2004) reviewed predator-induced plasticity in life history and morphology in organisms with complex life histories. His review showed that few studies on predator-induced morphological defence have tested for the effect of size at metamorphosis, growth rate, time to metamorphosis, morphology and behaviour, which is unfortunate.

If reduced activity results in energy savings allocated to growth, why do not all crucian carps (or prey in general) reduce their activity? Probably reduced activity would in nature result in a competitive disadvantage when resources are scarce. These costs would seem especially high under predator-free conditions, where competitor densities are high. In such circumstances, it probably pays to move actively (resulting in a shallow body) because it allows consumers to find food. When predators are present, competition is reduced, and more prey is available per capita. Thus, under predation risk it seems advantageous to have a low activity (resulting in a deep body). Field data support this because crucian carp show a shallow body shape and a low growth rate in lakes and ponds without pike, whereas in ponds with pike they show a deep body and a large size, indicating a high growth rate (Tonn, Holopainen & Paszkowski 1994; Holopainen et al. 1997). However, it still remains to be elucidated which factor has the strongest impact on the observed pattern in natural ponds and lakes; the differences in activity or the differences in resource density, and they might not be independent. Theoretical models have successfully explored the interaction between predator avoidance and per capita resource gain of the prey (e.g. McNamara & Houston 1987; Werner & Anholt 1993) but modelling the special case of predator-induced defence has reached less interest. Nevertheless, Peacor (2002) showed theoretically that predator-induced reduced activity could potentially increase resource densities of the consumers, resulting in a higher growth rate of the consumers. He found empirical support for this in an experimental study on bullfrog larvae (Peacor 2002).

The difference in activity may have a bigger impact on growth in the laboratory than in the wild, where the effect of competition for limited resources plays a role (Werner & Anholt 1993; Holopainen et al. 1997), because a reduced activity cannot go on forever in the field (McNamara & Houston 1987; Werner & Anholt 1993). Laboratory studies where resource densities as well as predation risk is manipulated would shed light on this issue. Today, very few such studies have been performed and it is premature to draw any conclusive general patterns. Appleton & Palmer (1988) performed such a study and claimed that their study found unambiguous evidence of a direct predator-induced effect, since the presence of predators induced a defence even during starvation. However, they did not measure the activity of the prey, which may have been low under starvation in the presence of predators. Steiner (2007b) did a similar manipulation in tadpoles and found that the behavioural response was strongest under intermediate resource conditions but the morphological response was not affected by resource densities. Relyea (2004) on the other hand found that tadpole showed less morphological response in response to predators under high competition for food (low per-capita resource densities). We suggest that the different correlations among antipredator behaviour, growth rate and induced morphological defence found in, for example, the well-studied amphibian system reviewed above might be a result of laboratory conditions with regard to resource densities used.

Our result also suggest that the frequently found pattern of fish feeding on benthic prey developing a deeper body than fish feeding on pelagic prey (Hjelm et al. 2001; Svanbäck & Eklöv 2003; Andersson et al. 2006) is also mediated by different activity levels caused by prey use. In a previous experiment, we showed that when crucians were given sedentary chironomids as food, a deep body shape was induced, while a shallow body shape was induced when they fed on actively swimming zooplankton. These changes were also correlated with swimming activity, since for example crucians that were given chironomids had a lower swimming activity compared to those that were given zooplankton as food resource (Andersson et al. 2006).

In summary, there appears to be abundant evidence that reduced activity results in body shape changes, and we found evidence for an activity-mediated indirect effect of predator presence. While this indirect effect may not be universal, it should be considered in all future studies of adaptive predator-induced morphological defence. Whether the indirectly induced morphology is adaptive (in the crucian carp as well as in other similar systems) remains subject to discussion. We note that pike prefer shallow-bodied carp (Nilsson et al. 1995), a deep body increase handling time (Nilsson et al. 1995), and improves escape response (Domenici et al. 2008). However, it is the reduction in activity that causes the body shape change.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Many thanks to Folmer Bokma, Christer Brönmark, Peter Eklöv, Tim Hipkiss, Martin Lind, Lennart Persson and Josh Van Buskirk for valuable comments on a previous version of this article. Henrik Jensen and Daniel Lussetti helped with the experiments. Financial support was provided by Swedish Research Council to F.J.

The experiments were performed with the permission (2007-1057) of the Swedish Animal Welfare Agency and the Swedish Board of Agriculture.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Fig. S1. Activity index of crucian carps (no. of squares entered during a 5-min period) in the presence (filled dots) and absence (unfilled dots) of pike cues at day 24 and 45 of the experiment. Error bars denote SE.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JANE_1530_sm_FigS1.doc32KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.