Oecologia (1997) 112:12±16
Ó Springer-Verlag 1997
Joanna Pijanowska
Alarm signals in Daphnia?
Received: 10 October 1996 / Accepted: 20 May 1997
Abstract Daphnia magna can respond to chemical cues from freshly crushed conspeci®cs with various behavioural reactions. A shift in vertical distribution towards the bottom, the formation of aggregations and direct escape responses can all be induced by water-borne signals released from crushed Daphnia. The pattern and strength of the ®rst two behavioural responses (i.e. the persistent tendency to occupy deeper strata in the experimental columns and to stay within patches) indicate that Daphnia perceive the signal from crushed conspeci®cs as nonspeci®c information, not necessarily associated with any particular kind of danger from either vertebrate or invertebrate predators. The adaptive value and possible costs associated with performing these two behavioural reactions are discussed. The adaptive value of the induced escape response was directly tested: Daphnia which had experienced the presence of a cue from crushed conspeci®cs avoided attacks by common bream more eciently than naive Daphnia. The recognition of the signal originating from crushed conspeci®cs can be especially adaptive in encounters with unfamiliar predators and with predators that undergo ontogenetic shifts in their diet. Under natural conditions, the combination of such a signal with a predator cue can, very reliably, advertise the local scale of the predatory impact. Key words Daphnia á Predation á Behavioural defences á Inducibility á Alarm signals
J Pijanowska1 Max-Planck Institut fuÈr Limnologie, Postfach 165, D-24306 PloÈn, Germany Present address: Department of Hydrobiology, University of Warsaw, Banacha 2, 02-097 Warsaw, Poland fax: (48 22) 22 47 04; e-mail:
[email protected] 1
Introduction Planktonic prey can reduce the risk of both invertebrate and vertebrate predation by employing various behavioral responses (for review see Ohman 1988). Predator avoidance in space and/or time may be achieved by vertical or horizontal displacements. Predators confusion and a lowered probability of being captured may be gained by forming dense aggregations (swarming). Finally, alertness and direct locomotory escape may be important mechanisms of capture avoidance. All these behavioural responses can be directly induced by the presence of chemical substances released to the environment by either vertebrate or invertebrate predators (for review see Larsson and Dodson 1993). Though some progress has been achieved in determining the properties of one such substance (Von Elert and Loose 1996), its chemical character still remains unknown. The presence of a predator in the environment does not, however, necessarily imply a direct danger to each potential prey item. The danger to the particular individual can vary not only with relative densities of predator and prey (Morin 1985), but also with their relative body sizes, i.e. relative prey attractiveness (Werner et al. 1983; Semlitsch and Gibbons 1988; Semlitsch 1990). Furthermore, the diversity and complexity of the habitat (Wilzbach et al. 1986; Figiel and Semlitsch 1991) may be crucial for the outcome of prey-predator interactions. The pursuit of very rare or very small prey, or hunting where there is very high refuge availability, can be too costly for a predator in terms of time and/or energy expenditure. Thus, even at high predator density some potential prey can be relatively safe and do not need to pay the high costs of unnecessary defence. Direct information on predation risk can be gathered via the recognition of alarm signals originating from conspeci®cs that are subjected to predation and lethally or non-lethally injured. Such a signal would inform potential prey not only of the presence of a predator in the environment but also of the exact predatory impact.
13
Alarm signals (Schreckstoe) in aquatic habitats have been shown to announce the risk of predation in ®sh (for a review see Smith 1992), amphibians (Petranka 1989) and some invertebrates, e.g. mosquito larvae (Sih 1986), gyrinid beetles (Henrikson and Stenson 1993), crabs (Appleton and Palmer 1988) and cray®sh (Hazlett 1994). There is no convincing evidence that such a substance aects the behaviour of cladocerans, though it should be present in the water. Some invertebrates do not swallow prey items whole, but crush them, and at least some of the content is released into the water. Young-of-theyear ®sh, still learning how to forage on plankton, have a high rate of unsuccessful attacks and long handling of a prey item before it is ingested (Win®eld et al. 1983; Ibrahim and Huntingford 1992). As a result, some prey can be damaged. The aim of this study was to investigate whether a substance originating from crushed cladocerans can be perceived by Daphnia and whether it can provide a signal inducing behavioural response in survivors, similar to that induced by the chemicals exuded by predators. Three behavioural responses which are known to be inducible by predatory cues were analysed ± diel patterns in vertical distribution, formation of aggregations and a direct escape response, each of them requiring speci®c methods.
Materials and methods I used D. magna from a clone isolated from Binnensee, a shallow brackish-water lake in north-eastern Germany, where they coexist with both ®sh and invertebrate predators (Lampert 1991). This clone has been shown to be sensitive to predatory cues from ®sh and invertebrate predators (Dawidowicz and Loose 1992; Loose et al. 1993; Weider and Pijanowska 1993). Daphnia used in all experimental trials were the third-clutch ospring of mothers which were themselves third-clutch ospring of one single female. The grandmother and mothers of experimental animals were fed the green alga Scenedesmus acutus, at a concentration daily adjusted to 1.0 mg C á l)1. The control medium for Daphnia in all tests was lake water, aged and aerated for about 1 week. The experimental medium was lake water with the addition of Daphnia homogenate prepared by crushing a known number of Daphnia in a small volume of water. Both media were passed through 0.45-lm ®lters prior to use. Vertical distribution The vertical distribution was followed in a system of 1.0-m vertical Plexiglas tubes of 1.5 cm diameter and 200 ml capacity (see Dawidowicz and Loose 1992; Dawidowicz 1993). The tubes were placed vertically in a water bath with thermal strati®cation from 23°C at the surface to 10°C at the bottom. The tubes were illuminated from above by halogen lamps (20 W, 12 V), the photoperiod was 16 h light:8 h dark and the light gradient was from 17.9 lE á cm)2 á s)1 at the surface to 1.3 lE á cm)2 á s)1 at the bottom. Six tubes with ®ve Daphnia in each were used as a control, and six with ®ve Daphnia in each as an experimental treatment. The Daphnia specimens used in the experiment were 48±72 h old, and their size was 1.64 0.07 mm (mean SD). A suspension of S. acutus was added daily to the surface of the tubes to obtain a ®nal concentration of 2.0 mg C á l)1. In the experimental treatment, a 0.45-lm ®ltrate of Daphnia homogenate was added daily with the algae. The concentration used was ®ve crushed Daphnia in 200 ml or 25 per litre. The vertical distribution of Daphnia was followed for almost 100 h,
at 1000, 1200, 1400 and 1600 hours during the day, and once or twice in the night, at 2100 and 0000 hours. Data representing the daytime distribution were pooled for graphic presentation, but were treated separately in the repeated-measures ANOVA model (Sokal and Rohlf 1981) applied to estimate the eect of treatment, the duration of the experiment and the time of day on Daphnia distribution. Aggregation behaviour Aggregation behaviour was examined in six 12-l cylindrical plastic bowls (20 cm diameter), which were placed individually in black wooden compartments and illuminated from above, each bowl by its own halogen lamp (see Pijanowska 1993). Frosted glass placed under the lamp provided a homogeneous light distribution. The light intensity was 0.85 lE á cm)2 á s)1. Each bowl was equipped with a plastic grid of 60 ®elds. For the short time needed to estimate the horizontal distribution of Daphnia, the grid was placed in the bowl and Daphnia were trapped in the individual compartments, allowing fast counting. The grid was then removed, and Daphnia allowed to redistribute in the bowls for the next 2 h. The bowls were ®lled with 6 l water, three with control water and three with treatment water. Sixty Daphnia (1.72 0.23 mm; mean SD) were introduced into each bowl. Prior to the experiment, control Daphnia were kept for 24 h in the control water, while treated Daphnia were acclimated to the treatment water with Daphnia ®ltrate added. The ®ltrate concentration was ten homogenized Daphnia per 1 litre water. The food (S. acutus) concentration was initially set at 0.75 mg C á l)1. Daphnia horizontal distribution was followed over an 18-h period, every 2 h. For the ®rst 8 h, a constant light intensity of 0.85 lE á cm)2 á s)1 was maintained, and for the next 8 h, complete darkness prevailed (Daphnia were then counted in red light). The degree of patchiness, PI, was calculated according to Lloyd's (1967) formula PI
s=x2 ÿ
1=x 1, where x is the sample mean, and s is the sample variance. Index values above 1 indicate a patchy distribution, those below 1 indicate an even distribution, and those equal to 1 indicate a random distribution. Repeated measurements of ANOVA were applied to test for the eects of treatment, light and time on Daphnia horizontal distribution. Escape abilities The escape abilities of non-treated and treated Daphnia were compared by measuring the frequency of successful and failed attacks by a ®sh predator within 1 h starting from the ®rst successful attack. A successful attack was one which ended with prey ingestion, while a failed one ended with prey escape. Fifty Daphnia (size 1.39 0.19 mm; mean SD) were oered to young-of-theyear common bream (Abramis brama L.) of mean (SD) size 2.2 0.4 cm, in 2.5-l glass jars. Three ®shes (each in a separate jar) were fed Daphnia which were kept in aged lake water for 24 h prior to the experiment (control) and three other ®shes were fed Daphnia kept for 24 h in lake water with ®ltrate of freshly homogenized Daphnia added (20 crushed Daphnia á l)1). The number of successful and failed attacks was compared between hungry ®sh (starved for 24 h prior to feeding trials) and well-fed ®sh (which, after 24 h starvation, were fed with 50 Daphnia each prior to the experimental trials). Feeding experiments were run at 19°C and a subdued light intensity of 3.5 lE á cm)2 á s)1. The eect of treatment and ®sh hunger level on the number of successful and failed attacks was analysed by two-way ANOVA.
Results Vertical migration Daphnia exposed to a chemical cue from freshly crushed conspeci®cs remained signi®cantly deeper in the
14
Fig. 1 Daphnia depth distribution in control (dashed line) and in crushed-Daphnia treatments (mean 1 SD) (open squares day depth, ®lled squares night depth)
water column than control individuals P 0:00001, (F1;479 280:55, repeated-measures ANOVA, Fig. 1). In no treatment did animals perform regular diel displacements in the water column. The mean depth of control animals was relatively stable within the 96-h period of observation, and it never exceeded 20 cm. The mean depth of treated animals averaged between 40 and 80 cm, and it tended to increase through the experiment. The eect of time was signi®cant (P 0:0001; F3;479 12:84. The treatment ´ time interaction was also signi®cant (P 0:0438; F3;479 2:71). The broad within-treatment variation (high standard deviations in Fig. 1) resulted from a clear tendency for Daphnia to stay closer to the surface in the morning (counting at 1000 hours) and to descend towards the bottom of the columns later in the day, after the daily addition of the crushed Daphnia ®ltrate. The eect of daytime was signi®cant at P 0:0029 (F3;479 4:78, repeated-measures ANOVA), and Daphnia remained in the deepest parts of the experimental columns at 1600 hours (d1600 > d1400 > d1200 > d1000, P 0:05, Tukey-HSD test for pairwise comparisons). Though animals apparently tended to descend during the night following the laboratory sunsets, the overall dierence between night and day depths was not signi®cant. Aggregation behaviour Both control and treated Daphnia tended to aggregate in the experimental bowls, and the values of Lloyd's patchiness index never dropped below 1.5. During the 16 h of observation, Daphnia distribution was signi®cantly more patchy in the presence of the ®ltrate from crushed conspeci®cs than in the control (Fig. 2). The treatment eect was signi®cant at P 0:0013 (F1;47 143:37, repeated-measures ANOVA). Horizon-
Fig. 2 Degree of patchiness of control Daphnia (open bars) and of Daphnia exposed to ®ltrate of crushed conspeci®cs (hatched bars) (mean 1 SD). The horizontal black bar indicates total darkness
tal distribution remained stable during the ®rst eight daylight hours. Afterwards, in darkness, the animals distributed themselves less patchily (P 0:001, TukeyHSD test for pairwise comparisons), but the dierence between control and treated Daphnia remained signi®cant. The eect of light regime was signi®cant
P 0:0006; F1;47 247:72. Escape behaviour The total number of successful attacks by young-of-theyear bream (both hungry and well-fed ®sh) was signi®cantly lower when ®sh were fed with Daphnia which had experienced a 24-h exposure to a ®ltrate from freshly crushed conspeci®cs (Fig. 3). The overall dierence was signi®cant at P 0:0472, (F1;11 5:49, two-way ANOVA). Fish made signi®cantly more unsuccessful attacks on Daphnia exposed to ®ltrate from recently crushed conspeci®cs
P 0:0424; F1;11 5:82. Hungry ®sh performed more successful
P 0:0015; F1;11 22:36 and also more failed
P 0:0004; F1;11 33:98 attacks than fed ones. There was no signi®cant dierence between the number of successful attacks on control and treated Daphnia made by hungry ®sh, while the number of failed attacks was signi®cantly higher for treated Daphnia at P 0:0457 (two-sample t-test, Fig. 3a). In contrast, well-fed ®sh preyed more successfully
P 0:0239 on naive Daphnia, but the number of failed attacks did not dier between treated and non-treated Daphnia (Fig. 3b). Hungry ®sh were more active and performed more chaotic movements, while well-fed ®sh made fewer attacks and seemed to focus on particular prey items. On average, hungry ®sh made 0.28 unsuccessful attacks per successful one when feeding on control prey and 0.48 unsuccessful attacks per successful one when feeding on treated Daphnia. Well-fed ®sh
15
Fig. 3 Number of successful and failed attacks made by hungry (a) and well-fed (b) ®sh on control Daphnia (open bars) and on Daphnia treated with the ®ltrate of crushed conspeci®cs (hatched bars)
performed 0.17 unsuccessful attacks per successful one when preying on control Daphnia and 0.43 unsuccessful attacks per successful one while feeding on treated prey.
Discussion Some substances originating from crushed Daphnia induce behavioural responses in conspeci®cs. Apparently, these substances add to the chemical information in the environment and can be used by potential prey to recognize the local scale of predatory risk, leading to behaviours lowering the success of predators. The question arises as to how the Daphnia ``interpret'' the presence of such cues in the environment, i.e. whether it is identi®ed as a general or speci®c threat? From the persistent distribution of Daphnia at the bottom (Fig. 1), and their constant tendency to form aggregations which do not
disperse at night (Fig. 2), it seems that this cue is interpreted as a general signal of danger, not related to any particular source. If the perceived risk was related to a speci®c predator, one could expect the regular diel rhythms in the pattern of vertical displacements as well as in the formation of aggregations (Larsson and Dodson 1993). When two dierent types of predator which require dierent defensive strategies cause similar injuries, prey response to the unspeci®c signal can be maladaptive if high costs of permanent defence are paid. Recognition of unspeci®c cues should, however, be especially bene®cial in encounters with unfamiliar predators (Mathis and Smith 1993a) or in encounters with predators that undergo shifts in their diet (BroÈnmark and Pettersson 1994). This study directly demonstrated the adaptive value of the induced locomotory response: Daphnia which had experienced the previous exposure to a cue from damaged conspeci®cs avoided attacks by common bream more frequently than naive Daphnia (Fig. 3). This type of response should also be bene®cial against other predators, although common bream is probably one of the most important ®sh predators in eutrophic lakes. The biomass of common bream that may feed on zooplankton during the entire lifespan is invariably found to be very high, amounting to 40±80% of the total ®sh biomass in eutrophic lakes (Lammens et al. 1992). Under natural conditions, a signal from damaged and/or partly ingested prey is probably combined with substances released by predators. Such a combination should provide a more speci®c cue for potential prey. Indeed, as demonstrated by Appleton and Palmer (1988), marine gastropods displayed a weaker morphological response to the predator cue alone than to a cue coming from predatory crabs feeding on conspeci®c snails. Similarly, only pike fed fathead minnows (Mathis and Smith 1993b), perch fed crucian carp (BroÈnmark and Pettersson 1994) and pike fed bleak (Jachner 1997) elicited the defensive (behavioural or morphological) responses in conspeci®c prey ®sh, while a sole cue from piscivores or piscivores fed a no-®sh diet induced no (or a weaker) response in conspeci®c prey ®sh. Stirling (1995) recently reported that D. magna responded to the presence of ®sh by performing vertical migration, regardless of ®sh diet, while another cladoceran, D. galeata mendotae did not respond to the presence of ®sh alone but to a cue from ®sh eating conspeci®cs. These results revealed that closely related species react dierently to the presence of ®sh and that some Daphnia are sensitive not only to the presence of ®sh, but also to the level of predation risk. It has been suggested that an alarm substance released by injured prey may serve as a warning signal to conspeci®cs about danger from predators (e.g. Maynard-Smith 1965; Charnov and Krebs 1975). The recognition of this signal is bene®cial to survivors, as directly demonstrated in my study where threatened Daphnia suered fewer attacks from a ®sh predator. This observation supports the few other direct demon-
16
strations (e.g. Hews 1988; Mathis and Smith 1993a) that receivers of alarm signals experience increased survival chances compared to animals which do not receive the warning. Signal sending may also be bene®cial to senders when animals sending the signals are closely related. Acknowledgements I am grateful to Larry Weider for oering me the facilities of his laboratory during my fellowship stay at the Max-Planck Institute in PloÈn, and to Mrs. Eva Geissler for her assistance during the experimental work. I pro®ted greatly from the exchange of ideas with the participants of the Third International Symposium on Cladocera in Bergen, 1993. Anonymous reviewers are kindly acknowledged for valuable suggestions, criticism and linguistic improvements.
References Appleton RD, Palmer AR (1988) Water-borne stimuli released by predatory crabs and damaged prey induce more predatorresistant shells in a marine gastropod. Proc Natl Acad Sci USA 85:4387±4391 BroÈnmark C, Pettersson LB (1994) Chemical cues from piscivores induce a change in morphology in crucian carp. Oikos 70:396± 402 Charnov EL, Krebs JR (1975) The evolution of alarm calls: altruism or manipulation? Am Nat 109:107±112 Dawidowicz P (1993) Diel vertical migration in Chaoborus ¯avicans: population patterns vs individual tracks. Arch Hydrobiol Beih Ergeb Limnol 3:19±28 Dawidowicz P, Loose CJ (1992) Metabolic costs during predatorinduced diel vertical migration of Daphnia. Limnol Oceanogr 37:1589±1595 Figiel CR, Semlitsch RD (1991) Eects of nonlethal injury and habitat complexity on predation in tadpole populations. Can J Zool 69:830±834 Hazlett BA (1994) Alarm responses in the cray®sh Oronectes virilis and Oronectes propinquus. J Chem Ecol 20:1525±1535 Henrikson B-I, Stenson JAE (1993) Alarm substance in Gyrinus aeratus (Coleoptera, Gyrinidae). Oecologia 93:191±194 Hews DK (1988) Alarm response in larval western toads, Bufo boreas: release of larval chemicals by a natural predator and its eect on predator capture eciency. Anim Behav 36:125-133 Ibrahim AA, Huntingford FA (1992) Experience of natural prey and feeding eciency in three-spined sticklebacks (Gasterosteus aculeatus L.). J Fish Biol 41:619±625 Jachner A (1997) The response of bleak to predator odour of unfed and recently fed pike. J Fish Biol 50:878±886 Lammens EHRR, Frank-Landman A, McGillavry PJ, Vlink B (1992) The role of predation and competition in determining the distribution of common bream, roach and white bream in Dutch eutrophic lakes. Environ Biol Fish 33:195±205
Lampert W (1991) The dynamics of Daphnia magna in a shallow lake. Verh Int Ver Limnol 24:795±798 Larsson P, Dodson SI (1993) Chemical communication in planktonic animals. Arch Hydrobiol 129:129±155 Lloyd M (1967) Mean crowding. J Anim Ecol 36±130 Loose CJ, Von Elert E, Dawidowicz P (1993) Chemically-induced diel vertical migration in Daphnia: a new bioassay for kairomones exuded by ®sh. Arch Hydrobiol 126:329±337 Mathis A, Smith RJF (1993a) Chemical alarm signals increase the survival time of fathead minnows (Pimephales promelas) during encounters with northern pike (Esox lucius). Behav Ecol 4:260± 265 Mathis A, Smith RJF (1993b) Chemical labeling of northern pike (Esox lucius) by the alarm pheromone of fathead minnows (Pimephales promelas). J Chem Ecol 19:1967±1979 Maynard Smith J (1965) The evolution of alarm calls. Am Nat 119:116±120 Morin PJ (1985) Predation intensity, prey survival and injury frequency in an amphibian predator-prey interaction. Copeia 1985:624±628 Ohman MD (1988) Behavioral responses of zooplankton to predation. Bull Mar Sci 43:530±550 Petranka JW (1989) Response of toad tadpoles to con¯icting chemical stimuli: predator avoidance versus ``optimal'' foraging. Herpetologica 45:283±292 Pijanowska J (1993) Fish-enhanced patchiness in Daphnia distribution. Verh Int Ver Limnol 25:2366±2368 Semlitsch RD (1990) Eects of body size, sibship, and tail injury on the susceptibility of tadpoles to dragon¯y predation. Can J Zool 68:1027±1030 Semlitsch RD, Gibbons JW (1988) Fish predation in size-structured populations of treefrog tadpoles. Oecologia 75:321±326 Sih A (1986) Antipredator responses and the perception of danger by mosquito larvae. Ecology 67:434±441 Smith RJF (1992) Alarm signals in ®shes. Rev Fish Biol Fish 2:33± 63 Sokal RR, Rohlf FJ (1981) Biometry. Freeman, San Francisco Stirling G (1995) Daphnia behaviour as a bioassay of ®sh presence or predation. Funct Ecol 9:778±784 Von Elert E, Loose CJ (1996) Predator-induced diel vertical migration in Daphnia: enrichment and preliminary chemical characterization of a kairomone exuded by ®sh. J Chem Ecol 22:885±895 Weider LJ, Pijanowska J (1991) Plasticity of Daphnia life histories in response to chemical cues from predators. Oikos 67:385±392 Werner EE, Gilliam JF, Hall DJ, Mittelbach GG (1983) An experimental test of the eects of predation risk on habitat use in ®sh. Ecology 64:1540±1548 Wilzbach MA, Cummins KW, Hall JD (1986) In¯uence of habitat manipulations on interactions between cutthroat trout and invertebrate drift. Ecology 67:898±911 Win®eld IJ, Peirson G, Cryer M, Townsend CR (1983) The behavioural basis of prey selection by underyearling bream (Abramis brama L.) and roach (Rutilus rutilus L.). Freshwater Biol 13:139±149
