- Email: [email protected]
Nutritional stress and behavioural immunity of damselflies
ANIMAL BEHAVIOUR, 2001, 61, 1093–1099 doi:10.1006/anbe.2001.1693, available online at http://www.idealibrary.com on
Nutritional stress and behavioural immunity of damselflies B. LEUNG*, M. R. FORBES† & R. L. BAKER‡
*Department of Zoology, Cambridge University †Department of Biology, Carleton University ‡Department of Zoology, University of Toronto at Mississauga (Received 25 September 2000; initial acceptance 14 November 2000; final acceptance 18 January 2001; MS. number: 6699)
Increased mortality in the presence of stress may result from stress-reduced availability of energy for immune function, coupled with the presence of pathogens or parasites. We tested the hypothesis that stress reduces antiparasite responses of damselflies Ischnura verticalis (Hagen) to their ectoparasitic mites Arrenurus pseudosuperior (Marshall). Numbers of colonizing mites did not differ between nutritionally stressed and unstressed damselflies. However, unstressed damselflies successfully removed more attached mites than nutritionally stressed host larvae. Furthermore, certain damselfly behaviours increased in the presence of nonfeeding mite larvae. Some of these behaviours were effective in defending against mites, but were reduced by nutritional stress. These results are sufficient to explain inverse relations found between damselfly condition and intensity of mite parasitism seen in nature, and are likely to be applicable to other host–ectoparasite associations.
parasitism may cause population declines (e.g. amphibians: Carey 1993) or regulate populations (e.g. insect pest species: Anderson & May 1981). Thus, it is important to know which stresses (natural and anthropogenic) compromise immune function in which organisms and whether these stresses can act concurrently with challenge by parasites or pathogens. In both vertebrates and invertebrates, inverse relations are often found between the condition of animals and their numbers of ectoparasites (e.g. reviewed by Marshall 1981). These relations could be due to direct costs of parasitism, stress-related reductions in immune ability resulting in greater colonization by parasites, or to both factors. Furthermore, it is known that ectoparasitic arthropods elicit grooming behaviours (e.g. vertebrates: Marshall 1981; Mooring & Hart 1995; invertebrates: Smith 1988). Yet, we often do not know the direct costs of these behaviours or whether their frequency, duration or efficacy is affected by stress. In damselflies specifically, inverse relations have been found between wet mass at emergence (a metric of host condition) and numbers of ectoparasitic mites (Forbes & Baker 1990; Leung & Forbes 1997). However, a direct cost of parasitism does not account for these inverse relations because they exist before the actual onset of parasitism (while the larval mites are phoretic on newly emerged damselflies and before mites parasitize adult damselflies, e.g. Forbes & Baker 1990). These findings do not preclude hypotheses based on costs to resisting mite colonization or costs to actual phoresy (cf. Elzinga & Broce 1988), but
Researchers in the emerging field of ecological immunology wish to understand factors affecting allocation of host energy and resources to resisting parasites and pathogens (Sheldon & Verhulst 1996). Researchers in this and related fields are also interested in understanding the importance of stress to immune function. Resistance is known to be compromised during intrinsic hormonally mediated stress associated with breeding in mammals (lactating pregnant females: Agyemang et al. 1992; males in rut: Folstad et al. 1989). Extrinsic stresses are also associated with reduced immune function in fish (e.g. trout, Oncorhynchus mykiss, in polluted waters have more fungal infections, Carballo et al. 1995; cf. Barker et al. 1994; Evans et al. 1995). For invertebrates, activities such as foraging and mating are known to be costly, diminishing immune responses of insects to novel foreign objects used to simulate parasitism (e.g. Ko ¨ nig & Schmid-Hempel 1995; Siva-Jothy et al. 1998). Such constraints are important to understand given that elevated mortality from stress-related increases in susceptibility to parasitic or pathogenic organisms may transcend into population effects. Some researchers have suggested that synergistic interactions between stress and Correspondence: B. Leung, Department of Zoology, Cambridge University, Downing Street, Cambridge CB2 3EJ, U.K. (email: [email protected]). M. R. Forbes is at the Department of Biology, 2240 Herzberg Laboratories, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario K1S 5B6, Canada. R. L. Baker is at the Department of Zoology, University of Toronto at Mississauga, 3359 Mississauga Road, Mississauga, Ontario L5L 1C6, Canada. 0003–3472/01/061093+07 $35.00/0
2001 The Association for the Study of Animal Behaviour
1093
2001 The Association for the Study of Animal Behaviour
1094
ANIMAL BEHAVIOUR, 61, 6
only suggest that individuals already in poor condition may be subjected to greater eventual parasitism. We carried out the present study to test this hypothesis. We conducted four experiments to determine the nature of relations between nutritional stress, behavioural immunity and parasitism for an invertebrate host. We used the damselfly Ischnura verticalis (Hagen) and its parasitic mite Arrenurus pseudosuperior (Marshall): a model study system that has proved amenable to laboratory studies (Baker & Smith 1997; Leung et al. 1999). We first examined whether nutritionally stressed larval damselflies accumulated more mites over time than unstressed damselflies. We designed this first experiment to see if we could produce a pattern in the laboratory that has been observed in nature, before conducting more detailed experiments on the potential causes of those patterns. We next examined which behaviours of a suite of known larval behaviours were explicitly involved in defence against mites and how expression of various behaviours covaried. We reasoned that defensive behaviours should be elevated in the presence of mites, and that they should be inversely related to the numbers of mites colonizing the host or positively related to the numbers of mites lost or removed from hosts (experiments 2 and 3). We also examined whether these specific behaviours versus other behaviours were diminished for nutritionally stressed larval damselflies (experiment 4). We conducted these experiments sequentially, each building on former experiments. In summary, we explored whether factors that reduce the condition of damselflies may compromise their ability to defend against mites. METHODS All experiments were done in June and July 1997 and 1998. In both summers, we collected damselflies as F-1 larvae (where F means final instar) from Burns Conservation Area (4329 N, 8002 W; near Mississauga, Ontario, Canada). We reared larval mites in the laboratory from eggs readily found at Charleston Lake (4432 N, 7601 W; near Gananoque, Ontario, Canada). We kept damselflies at low temperatures (10C) to slow their development, such that enough larvae moulted to their final instar at about the same time the mite eggs were hatching. Generally, once some clutches of mite eggs hatched, we housed the damselflies individually in 75-ml vials at room temperature (ca. 22C) and fed them a diet of white worms (enchytraeid worms) and Daphnia magna (L.). From then on, we checked them for moulting every 24 h in the early morning. After they moulted into their final instar (wing pads reaching abdominal segment four or five; Baker 1986), equal numbers of final-instar larvae were placed into nutritional or other groups. To determine whether nutritional stress affected numbers of mites attending damselflies over time (experiment 1), we fed some damselflies one white worm twice weekly and others one white worm each day, for a week. We paired a nutritionally stressed and unstressed damselfly of the same sex and placed each pair into a common 75-ml
vial (N=46 for each group). We tested the assumption that our nutritional stress had measurable consequences on condition (stressed damselflies were significantly lighter than unstressed damselflies; paired t test: t22 =5.08, P<0.01). Subsequently, we introduced >100 larval mites into the vial. Clutches of A. pseudosuperior are extremely clumped in nature with over 10 000 eggs in <1 m2 (M. R. Forbes, personal observation); hence, some larval damselflies are likely to experience high numbers of mites challenging them. We controlled for potential variability in laboratory conditions, mite numbers and variation in mite vigour in attacking hosts by using a matched-pair design. Furthermore, this design allowed the mites to have a reasonable probability of encountering both damselflies, and thus allowed them to make a choice between hosts if they possessed this ability. In a previous study, we found that A. pseudosuperior had a slight tendency to attach to larval I. verticalis that were close to emergence (Leung et al. 1999). After 1–2 h, we removed the damselflies, counted the number of attached mites, and placed the larval damselflies individually into separate vials filled with water that was free of mites. After 24 h, we counted the number of mites remaining on each damselfly. The change in the number of mites could indicate the efficiency with which stressed and unstressed damselflies removed mites. To determine specific behaviours associated with mite presence or possible defence against mite colonization (experiment 2), we first kept damselflies at ca. 22C and fed them ad libitum with enchytraeid worms or D. magna every 1–2 days. After they had spent ca. 1 week as a final instar, we transferred them to new vials and later filmed them to record their behaviour for 15 min in the absence of mites (as a control). Then, using the same damselfly larvae, we introduced 30 mites, and recorded behaviour for an additional 45 min. After the 45-min filming period, we added another 60 mites and left the damselflies for an additional 1 h before they were removed and their numbers of attending mites counted. We assumed that relative times spent in various behaviours during the initial 45 min were representative of behaviours for the next 1 h. To record damselfly behaviour, we used two video cameras and recorders and placed three damselflies in the view of each camera; white paper was placed between the vials so the damselflies could not see each other (thereby affecting their behaviour for reasons unrelated to challenge with mites; e.g. conspecific interactions). We recorded the following behaviours (after Baker 1981; Forbes & Baker 1990; Baker & Smith 1997): time spent in Crawl forward, Swim, Groom and Abdomen wave, and frequency of Slash, Wiggle and Labial strike (Table 1). Because of the rapidity of Slash and Wiggle and the fact that we lost resolution when we played tapes at slow speed, it was often difficult to distinguish clearly between the two. Therefore we chose to combine the occurrences of these two behaviours (henceforth referred to as Slash for convenience). We added 60 mites after the initial filming period because we wanted to increase our range in numbers of colonizing mites and thus our ability to detect relations between specific behaviours and numbers
LEUNG ET AL.: STRESS AND BEHAVIOURAL IMMUNITY
Table 1. Behaviours of larval damselflies Behaviour
Abdomen wave
Crawl forward Groom
Labial strike Slash Swim Wiggle
Description and significance
The animal repeatedly performed a low-amplitude (at an angle ca 10° from the rest of the body) transverse motion of the last six or seven abdominal segments and lamellae. Unlike previous investigations (Baker & Smith 1997), single abdomen bends were not recorded Any forward movement in which the animal maintained continuous limb contact with the substrate All movements in which either the tarsal claws or distal segments of a leg or pair of legs were rubbed against any part of the body. Although grooming can remove mites (Forbes & Baker 1990), its efficacy in mite removal has not previously been examined. Grooming can be variously termed scratch and rub, but in this study it included another behaviour involving a lateral bending of the distal tip of the abdomen and lamellae to a point where they were brought over the head and rubbed against the dorsum of the head and thorax Rapid extension of labium, generally in the direction of prey or debris on the floor of the vial. Strike was often directed at swimming larval mites A rapid lateral bending of the abdomen, resulting in caudal lamella being displaced ≥90° Any movement through the water in which the animal did not maintain contact with the substrate. The movement was usually accomplished by a transverse undulation of the abdomen and lamellae A rapid lateral undulation of the entire abdomen resulting in a series of waves passing down the body; each occurrence lasted only 1–2 s. Wiggle and Slash were combined and are simply referred to as Slash
After Forbes & Baker (1990) and Baker & Smith (1997).
of mites attaching, should these relations exist. We identified behaviours as defensive if they increased in the presence of mites, and if duration performing a behaviour was inversely correlated with number of attaching mites. To control for differences in duration of observation, we expressed time spent performing a behaviour as a proportion of the total time the damselfly was watched. To determine whether specific behaviours were associated with mite removal over a longer duration after their initial attachment, we exposed each of 16 damselflies to 75 mites (experiment 3: each damselfly was housed in a smaller 15-ml vial with a perch). We allowed mites to colonize damselflies for 1–2 h and then counted the number of mites on the damselflies. Afterwards, we placed the damselflies in vials with mite-free water, to prevent further attachment. As above, we used two video cameras with a maximum of three damselflies in view of each camera. We recorded behaviour for 1 h. After recording, we recounted the number of mites on each damselfly to determine the number of mites removed. We further identified behaviours as defensive if they had increased in the presence of mites (in the previous experiment), and if duration performing a behaviour was positively correlated with the number of mites removed over time. In all cases of filming, we could not directly record the number of mite contacts and number of mite removals without much more sophisticated equipment and fewer mites to enable tracking of individual mites. To determine if specific defensive behaviours were reduced in the presence of nutritional stress (experiment 4), we fed stressed damselfly larvae six D. magna or one small enchytraeid worm every day and unstressed damselflies ad libitum D. magna or a mixture of enchytraeid worms and D. magna each day. Again, this difference in diet resulted in higher nutritional status of the unstressed group, as measured by wet mass after emergence (twoway ANOVA controlling for sex: F1,20 =7.92, P=0.01). We also used principal components analysis (PCA) to
determine whether specific behaviours covaried with one another such that changes that were due to stress could be viewed as a single correlated response. After the damselflies moulted into final instars, we allowed them to develop for 6–7 days (15 in stressed group and 16 in unstressed group). In so doing, we ensured that individuals in the stressed group received far fewer feedings over a fairly long period. We then challenged the damselflies by placing ca. 20 mites into their vials/day. On the fifth or sixth day, we filmed them for 1 h, after introducing mites into their vials. We then recorded behaviours as above. We transformed all proportions using arcsine(x0.5) before conducting statistical tests (Zar 1984). However, we used the raw proportions for descriptive purposes (i.e. meanSE) as this would be biologically more informative. RESULTS We used proportions to remove variability in the total number of mites attaching to damselfly larvae in each matched pair. The proportion of mites attaching to stressed versus unstressed animals did not differ (paired t test: t45 =0.94, P=0.35; Fig. 1). However, the proportion of mites removed over a 24-h period differed significantly in the predicted direction, that is, unstressed animals lost or removed proportionately more mites than stressed animals (paired t test: t45 =3.62, P<0.001; Fig. 1). This difference in removal rates (see below) resulted in a difference in the proportion of mites remaining on stressed versus unstressed animals (paired t test: t45 =3.03, P=0.004; Fig. 1). After 24 h, stressed animals had 17% more mites than unstressed animals. We used paired t tests to examine which behaviours increased in the presence of mites. Time spent in Groom and Abdomen wave and frequency of Slash and Labial strike all increased in the presence of mites and remained significant after sequential Bonferroni adjustment (Rice
1095
ANIMAL BEHAVIOUR, 61, 6
1.00
0.75 Proportion
1096
0.50
0.25
0.00
Colonizing
Removed
After 24 h
Figure 1. Differences between stressed (h) and unstressed (") damselflies in terms of the proportion of mites initially colonizing, the proportion of mites removed over 24 h, and the proportion of mites remaining after 24 h. Stressed and unstressed damselflies were paired; thus individuals within a pair were subjected to identical parasite conditions (46 pairs). Proportions were determined for each pair, and the average proportions are shown. Statistical tests (see text) were conducted on arcsine(x0.5) transformed proportions (Zar 1984). However, the figure shows raw untransformed values. Error bars represent 1 SE.
1989), but time spent in Crawl forward and Swim did not relate to mite presence (Table 2). In brief, damselfly larvae exposed to mites were more active. Only Abdomen wave, Labial strike and Swim were related to the number of attaching mites (Table 2). However, Abdomen wave was positively related to the number of mites that attached, whereas Labial strike was inversely related, as predicted if this behaviour was important in defence. Swim became nonsignificant after sequential Bonferroni correction. In the third experiment, where larvae were provided with more time to remove mites, Groom and Labial strike were positively related to the proportion of mites removed, but Crawl, Swim, Abdomen wave and Slash were not (Table 2). Groom remained significant after sequential Bonferroni correction, but Labial strike did not. Taken together, these results suggest that only Labial strike and Groom fulfil the criteria of defensive behaviours. These behaviours increased in the presence of mites, but one appeared effective when mites were colonizing hosts (Labial strikes) whereas Groom had an effect on mite removal. Groom may be combined with Labial strike to reduce the threat from mites, by first dislodging and then attacking them (as we occasionally observed during this study). In experiment 4, unstressed damselflies performed Groom, Abdomen wave, Slash and Swim more often than stressed damselflies (Table 2). These behaviours remained significant after sequential Bonferroni corrections, except for Swim. However, stressed and unstressed damselflies did not differ in Crawl forward or Labial strike. In brief, stressed damselflies were considerably less active than unstressed damselflies; however, whereas some behaviours remained unaffected, the defensive behaviour Groom was reduced considerably.
It is relevant that only grooming was affected appreciably by nutritional stress, because we found appreciable covariation between behaviours. In the PCA, the first axis accounted for ca. 45% of the combined variation in behaviour. The percentage time spent grooming, waving, slashing and striking all correlated strongly on the first axis (factor loadings >0.70 or < 0.70); the first three behaviours were positively related, whereas striking was inversely related, to that first axis. Swimming and crawling were not strongly related to the first axis (factor loadings=0.10 and 0.35, respectively). However, these two behaviours loaded strongly and inversely on the second PC axis (factor loadings < 0.80) which accounted for ca. 25% of the remaining variation in behaviour. DISCUSSION Synergism between stress and parasitism may occur if stress reduces the ability of hosts to defend against parasites or affects exposure to parasites, thereby resulting in a higher intensity of parasites on a given host. In summarizing the ecology of ectoparasitic insects, Marshall (1981) argued that high intensities of ectoparasitism on hosts might be more a consequence of poor condition of hosts, rather than a cause of that poor condition. Yet scant empirical evidence has addressed whether stressed invertebrate hosts are more likely to be parasitized than unstressed hosts. The majority of studies on stress– immunocompetence relations have been done on vertebrates (see references in Introduction). Our experiments were designed to determine if there was a potential for stress to affect susceptibility of an invertebrate host to ectoparasitism. Our results provide a sufficient explanation for hosts in lower condition having more ectoparasites. In brief, stressed hosts accumulated more mites, grooming appeared effective in removing mites, and grooming was reduced in the presence of stress. We are confident that mites induced increased grooming, rather than the alternative of damselflies taking time to acclimate to vials and then expressing ‘normal’ behaviours. We believe this because grooming occured immediately after mite introductions, and in another paired trial where mites were added versus not after the same period of acclimation in vials, only damselflies with mites added increased grooming (Forbes & Baker 1990). We argue further that nutritional stress, as examined in our study, is likely to play an important role under field conditions. In summarizing studies on food limitation in larval damselflies Corbet (1999) noted that feeding rates in nature are generally less than maximum possible feeding rates in captivity. In our experiments, feeding rates were made to be high and low: the low-condition treatment reflected what can occur in nature (since damselflies could still emerge at this low condition). Such results can explain the observed inverse relations between mite number and damselfly condition, as seen in other field studies (Forbes & Baker 1990; Leung & Forbes 1997). It is perhaps not surprising that labial strike was not reduced in the presence of nutritional stress since this is a
LEUNG ET AL.: STRESS AND BEHAVIOURAL IMMUNITY
Table 2. Effect of mite presence on damselfly behaviour (experiment 2), effect of damselfly behaviour on numbers of mites that attached initially (experiment 2) and on the proportion of mites removed after colonization (experiment 3), and effect of nutritional stress on behaviour (experiment 4) Behaviour
Experiment 2 Mite presence
Experiment 2 Mite attachment
Experiment 3 Mites removed
Experiment 4 Nutritional stress
6.46±2.97 29.26±3.08 F1,27 =25.70
23.36±3.69 17.73±3.81 t24 =2.59
r=0.55 F1,21 =9.25
r= −0.21 F1,13 =0.58
Crawl forward
0.72±0.19 1.20±0.37 t24 =0.55
r=0.18 F1,21 =0.76
r=0.36 F1,13 =1.98
1.20±0.27 1.12±0.27 F1,27 =0.07
Groom
7.70±1.05 1.35±0.38 t24 =6.55
r= −0.07 F1,21 =0.12
r=0.65 F1,13 =9.70
3.36±0.49 7.57±1.06 F1,27 =10.98
Labial strike
0.27±0.05 0.0±0.00 t24 =7.08
r= −0.61 F1,21 =12.22
r=0.53 F1,13 =5.14
1.23±0.33 0.60±0.16 F1,27 =3.45
Slash
0.53±0.11 0.05±0.02 t24 =6.06
r= −0.03 F1,21 =0.023
r=0.22 F1,13 =0.64
0.16±0.06 1.16±0.30 F1,27 =14.89
Swim
0.54±0.16 0.33±0.12 t24 =1.17
r=0.45 F1,21 =5.41
r=0.18 F1,13 =0.44
0.07±0.04 1.15±0.82 F1,27 =4.24
Abdomen wave
For experiment 2, effect of mite presence on behaviour was analysed with sequential Bonferroni-corrected paired t tests, predicting increases in activity in the presence of mites. First and second entries refer to mean percentage of time±SE spent in that behaviour in the presence and absence of mites, respectively. We also give t values. Column 2 shows the correlation (correlation coefficient and F value) between behaviour and proportion of mites attaching initially, analysed with ANCOVA models to control for sex. For experiment 3, the effect of behaviour on the proportion of mites removed is given, analysed with ANCOVA models to control for sex. For experiment 4, the effect of stress on damselfly behaviour was analysed with a two-way ANOVA to control for sex. First and second entries refer to mean percentage of time±SE spent in a behaviour by stressed and unstressed larval damselflies, respectively. Behaviours did not differ between sexes in any test and are not presented. All percentages were first transformed with arcsine(x0.5) before statistical analyses were conducted. However, we present raw untransformed values for means±SE. Bold values indicate significance at P<0.05 after sequential Bonferroni correction.
foraging behaviour. While it may be effective in reducing challenge from mites, these reductions are not enough to prevent associations between condition and ectoparasitism. As mentioned previously, these findings do not preclude hypotheses based on costs to resisting mite colonization or costs to actual phoresy (cf. Elzinga & Broce 1988), but only suggest that individuals already in poor condition may be subjected to greater eventual parasitism. We reason that inverse relations between wet mass at emergence and number of attaching mites would be due to damselflies in poor condition being less likely to engage in antimite behaviours. However, alternative explanations exist. For example, in nature, those damselflies in areas most frequented by mites may groom more, lose mass and still have more mites attach to them. Another possible mechanism is that mites might have preferentially selected poor-condition hosts over others or may have abandoned well-fed hosts more frequently. We deal with each of these alternatives in turn. Grooming in damselflies is likely to be costly. It may be energetically expensive or may detract needed time from other activities such as foraging. Thus, one might expect damselflies that encounter more mites to have lower rates of mass gain. We cannot fully address this alternative at this time, but a few points are worth mentioning. Since,
in nature, mites are probably only locally numerous, dispersal of larval damselflies may be an additional line of defence. The effects of mites on dispersal tendencies and the relation between success in dispersal and condition of larval hosts needs to be examined. There is the additional potential complication of predation. Like many other prey, damselflies trade movement associated with feeding against risk of predation (Dixon & Baker 1988). Because grooming also contributes to movement and potential detection by fish predators, we might expect damselflies confronted with high predation risk in nature to reduce both feeding and antiparasite behaviours, thus producing low-condition animals with relatively high parasitism compared with other animals with locally lower risk of predation. However, Baker & Smith (1997) showed that damselflies continue to groom when confronted with mites even when fish predators are present and this could result in higher predation risk, but not low-condition animals with high parasitism. Two other points are worth considering. First, mites may select stressed over unstressed hosts (we did not consider the reverse scenario given inverse relations between condition and mite numbers observed in the field). This could occur, for instance, if a chemical cue indicated that stressed individuals were unable to mount an effective response. However, the proportion of mites
1097
1098
ANIMAL BEHAVIOUR, 61, 6
initially attaching to stressed hosts equalled those of unstressed hosts, suggesting that they were not choosing their hosts based on host condition. Furthermore, in another study using these same species and a similar laboratory set-up, we found that mites preferentially colonized old larvae close to emergence (Leung et al. 1999) suggesting that if mites were making a choice in the present study, we should have been able to document it. Although stressed and unstressed damselflies are likely to have differed in degree of development in this study, variation in developmental stage was much greater in Leung et al.’s (1999) study , that is, none of the damselflies in the present study was within 1–2 days of emergence. Second, mites may also have differentially abandoned well-fed hosts. While we cannot explicitly address this possibility, we do not believe that it is likely. Damselflies in better condition live for longer after emergence. Thus, mites attached to these damselflies have a greater probability of returning to water and completing their life cycle (i.e. mites should prefer well-fed damselflies). Furthermore, mites may be able to obtain nutrients more quickly on well-fed adult damselflies. Thus, quicker development time (or better mite condition) also would not be a reason for higher abandonment of well-fed hosts. These observations increase our general understanding of parasite–host relations. In nature, stressed hosts are likely to have higher rates of mite attachment and lowered fitness as adults. Degree of mite parasitism on adults has been related to reduced longevity (Forbes & Baker 1991; Leung & Forbes 1997), reduced fecundity in fieldcaught females (Forbes & Baker 1991) and reduced mating success in males (Forbes 1991; but see Andre´ s & Cordero 1995; Rolff et al. 2000). However, positive life history trade-offs are also possible (e.g. Rolff 1999 showed that mite parasitism may increase offspring size in damselflies). In previous correlative studies showing a fitness association with mites, the impact of parasites could not be decoupled from the impact of lower condition potentially resulting in higher parasitism in the first place. Here, we suggest that low condition, in fact, does result in higher parasitism and provides at least a partial mechanism for relations observed in nature. Such studies now have to be combined with the fitness functions in relation to mite intensity for both host and mites, to ascertain what effect such synergisms will have on both populations (cf. Jaenike 1996). Acknowledgments We thank C. Elkin for all his help in the laboratory, B. P. Smith for help with collecting mites and two anonymous referees for comments on the manuscript. This research was supported by a Natural Sciences and Engineering Research Council (NSERC) Scholarship to B.L., and NSERC grants to R.L.B. and M.R.F. References Agyemang, K., Dwinger, R. H., Little, D. A., Leperre, P. & Grieve, A. S. 1992. Interaction between physiological status in N’Dama
cows and trypanosome infections and its effect on health and productivity of cattle in Gambia. Acta Tropica, 50, 91–99. Anderson, R. M. & May, R. M. 1981. The population dynamics of microparasites and their invertebrate hosts. Philosophical Transactions of the Royal Society of London, Series B, 291, 451–524. Andre´ s, J. A. & Cordero, A. 1998. Effects of water mites on the damselfly, Ceriagrion tenellum. Ecological Entomology, 23, 103– 109. Baker, R. L. 1981. Behavioral interactions and use of feeding areas by nymphs of Coenagrion resolutum (Coengrionidae: Odonata). Oecologia, 49, 353–358. Baker, R. L. 1986. Developmental stages and the analysis of zygopteran life histories. Freshwater Ecology, 3, 325–332. Baker, R. L. & Smith, B. P. 1997. Conflict between antipredator and antiparasite behavior in larval damselflies. Oecologia, 109, 622– 628. Barker, D. E., Khan, R. A. & Hooper, R. 1994. Bioindicators of stress in winter flounder, Pleuronectes americanus, captured adjacent to a pulp and paper mill in St. George’s Bay, Newfoundland. Canadian Journal of Fisheries and Aquatic Sciences, 51, 2203–2209. Carballo, M., Munox, M. J., Cuellar, M. & Tarazona, J. V. 1995. Effects of waterborne copper, cyanide, ammonia, and nitrate on stress parameters and changes in susceptibility to Saprolegniosis in rainbow trout (Oncorhynchus mykiss). Applied and Environmental Microbiology, 61, 2108–2112. Carey, C. 1993. Hypothesis concerning the causes of the disappearance of boreal toads from the mountains of Colorado. Conservation Biology, 7, 355–362. Corbet, P. S. 1999. Dragonflies: Behavior and Ecology of Odonata. Ithaca, New York: Cornell University Press. Dixon, S. M. & Baker, R. L. 1988. Effects of size on predation risk, behavioural response to fish, and cost of reduced feeding in larval Ischnura verticalis (Coenagrionidae: Odonata). Oecologia, 76, 200– 205. Elzinga, R. J. & Broce, A. B. 1988. Hypopi (Acari: Histiostomatidae) on house flies (Diptera:Muscidae): a case of detrimental phoresy. Journal of the Kansas Entomological Society, 61, 208–213. Evans, S. M., Dawson, M., Day, J., Frid, C. L. J., Gill, M. E., Pattisina, L. A. & Porter, J. 1995. Domestic waste and TBT pollution in coastal areas of Ambon Island (eastern Indonesia). Marine Pollution Bulletin, 30, 109–115. Folstad, I., Nilssen, A. C., Halvorsen, O. & Andersen, J. 1989. Why do male reindeer (Rangifer tarandus tarandus) have higher abundance of second and third instar larvae of Hypoderma tarandi than females? Oikos, 55, 87–92. Forbes, M. R. L. 1991. Ectoparasites and mating success of male Enallagma ebrium damselflies (Odonata: Coenagrionidae). Oikos, 60, 336–342. Forbes, M. R. L. & Baker, R. L. 1990. Susceptibility to parasitism: experiments with the damselfly Enallagma ebrium (Odonata: Coenarionidae) and larval water mites, Arrenurus spp. (Acari: Arrenuridae). Oikos, 58, 335–341. Forbes, M. R. L. & Baker, R. L. 1991. Condition and fecundity of the damselfly, Enallagma ebrium (Hagen): the importance of ectoparasites. Oecologia, 86, 335–341. Jaenike, J. 1996. Population consequences of parasite aggregation. Oikos, 76, 155–160. Ko ¨ nig, C. & Schmid-Hempel, P. 1995. Foraging activity and immunocompetence in workers of the bumble bee, Bombus terrestris L. Proceedings of the Royal Society of London, Series B, 260, 225–227. Leung, B. & Forbes, M. R. 1997. Fluctuating asymmetry in relation to indices of quality and fitness in the damselfly, Enallagma ebrium (Hagen). Oecologia, 110, 472–477.
LEUNG ET AL.: STRESS AND BEHAVIOURAL IMMUNITY
Leung, B., Baker, R. L. & Forbes, M. R. 1999. Grooming decisions by damselflies, age-specific colonization by water mites, and the probability of successful parasitism. International Journal for Parasitology, 29, 397–402. Marshall, A. G. 1981. The Ecology of Ectoparasitic Insects. London: Academic Press. Mooring, M. S. & Hart, B. L. 1995. Differential grooming rate and tick load of territorial male and female impala, Aepyceros melampus. Behavioral Ecology, 6, 94–101. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution, 43, 223–225. Rolff, J. 1999. Parasitism increases offspring size in a damselfly: experimental evidence for parasite-mediated maternal effects. Animal Behaviour, 58, 1105–1108.
Rolff, J., Antvogel, H. & Schrimpf, I. 2000. No correlation between ectoparasitism and male mating success in a damselfly: why behavior matters. Journal of Insect Behavior, 13, 563–571. Sheldon, B. C. & Verhulst, S. 1996. Ecological immunity: costly parasite defences and trade-offs in evolutionary ecology. Trends in Ecology and Evolution, 11, 317–321. Siva-Jothy, M. T., Tsubaki, Y. & Hooper, R. E. 1998. Decreased immune response as a proximate cost of copulation and oviposition in a damselfly. Physiological Entomology, 23, 274–277. Smith, B. P. 1988. Host-parasite interaction and impact of larval water mites on insects. Annual Review of Entomology, 33, 487–507. Zar, J. H. 1984. Biostatistical Analysis. 2nd edn. Upper Saddle River, New Jersey: Prentice Hall.
1099
Nutritional stress and behavioural immunity of damselflies B. LEUNG*, M. R. FORBES† & R. L. BAKER‡
*Department of Zoology, Cambridge University †Department of Biology, Carleton University ‡Department of Zoology, University of Toronto at Mississauga (Received 25 September 2000; initial acceptance 14 November 2000; final acceptance 18 January 2001; MS. number: 6699)
Increased mortality in the presence of stress may result from stress-reduced availability of energy for immune function, coupled with the presence of pathogens or parasites. We tested the hypothesis that stress reduces antiparasite responses of damselflies Ischnura verticalis (Hagen) to their ectoparasitic mites Arrenurus pseudosuperior (Marshall). Numbers of colonizing mites did not differ between nutritionally stressed and unstressed damselflies. However, unstressed damselflies successfully removed more attached mites than nutritionally stressed host larvae. Furthermore, certain damselfly behaviours increased in the presence of nonfeeding mite larvae. Some of these behaviours were effective in defending against mites, but were reduced by nutritional stress. These results are sufficient to explain inverse relations found between damselfly condition and intensity of mite parasitism seen in nature, and are likely to be applicable to other host–ectoparasite associations.
parasitism may cause population declines (e.g. amphibians: Carey 1993) or regulate populations (e.g. insect pest species: Anderson & May 1981). Thus, it is important to know which stresses (natural and anthropogenic) compromise immune function in which organisms and whether these stresses can act concurrently with challenge by parasites or pathogens. In both vertebrates and invertebrates, inverse relations are often found between the condition of animals and their numbers of ectoparasites (e.g. reviewed by Marshall 1981). These relations could be due to direct costs of parasitism, stress-related reductions in immune ability resulting in greater colonization by parasites, or to both factors. Furthermore, it is known that ectoparasitic arthropods elicit grooming behaviours (e.g. vertebrates: Marshall 1981; Mooring & Hart 1995; invertebrates: Smith 1988). Yet, we often do not know the direct costs of these behaviours or whether their frequency, duration or efficacy is affected by stress. In damselflies specifically, inverse relations have been found between wet mass at emergence (a metric of host condition) and numbers of ectoparasitic mites (Forbes & Baker 1990; Leung & Forbes 1997). However, a direct cost of parasitism does not account for these inverse relations because they exist before the actual onset of parasitism (while the larval mites are phoretic on newly emerged damselflies and before mites parasitize adult damselflies, e.g. Forbes & Baker 1990). These findings do not preclude hypotheses based on costs to resisting mite colonization or costs to actual phoresy (cf. Elzinga & Broce 1988), but
Researchers in the emerging field of ecological immunology wish to understand factors affecting allocation of host energy and resources to resisting parasites and pathogens (Sheldon & Verhulst 1996). Researchers in this and related fields are also interested in understanding the importance of stress to immune function. Resistance is known to be compromised during intrinsic hormonally mediated stress associated with breeding in mammals (lactating pregnant females: Agyemang et al. 1992; males in rut: Folstad et al. 1989). Extrinsic stresses are also associated with reduced immune function in fish (e.g. trout, Oncorhynchus mykiss, in polluted waters have more fungal infections, Carballo et al. 1995; cf. Barker et al. 1994; Evans et al. 1995). For invertebrates, activities such as foraging and mating are known to be costly, diminishing immune responses of insects to novel foreign objects used to simulate parasitism (e.g. Ko ¨ nig & Schmid-Hempel 1995; Siva-Jothy et al. 1998). Such constraints are important to understand given that elevated mortality from stress-related increases in susceptibility to parasitic or pathogenic organisms may transcend into population effects. Some researchers have suggested that synergistic interactions between stress and Correspondence: B. Leung, Department of Zoology, Cambridge University, Downing Street, Cambridge CB2 3EJ, U.K. (email: [email protected]). M. R. Forbes is at the Department of Biology, 2240 Herzberg Laboratories, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario K1S 5B6, Canada. R. L. Baker is at the Department of Zoology, University of Toronto at Mississauga, 3359 Mississauga Road, Mississauga, Ontario L5L 1C6, Canada. 0003–3472/01/061093+07 $35.00/0
2001 The Association for the Study of Animal Behaviour
1093
2001 The Association for the Study of Animal Behaviour
1094
ANIMAL BEHAVIOUR, 61, 6
only suggest that individuals already in poor condition may be subjected to greater eventual parasitism. We carried out the present study to test this hypothesis. We conducted four experiments to determine the nature of relations between nutritional stress, behavioural immunity and parasitism for an invertebrate host. We used the damselfly Ischnura verticalis (Hagen) and its parasitic mite Arrenurus pseudosuperior (Marshall): a model study system that has proved amenable to laboratory studies (Baker & Smith 1997; Leung et al. 1999). We first examined whether nutritionally stressed larval damselflies accumulated more mites over time than unstressed damselflies. We designed this first experiment to see if we could produce a pattern in the laboratory that has been observed in nature, before conducting more detailed experiments on the potential causes of those patterns. We next examined which behaviours of a suite of known larval behaviours were explicitly involved in defence against mites and how expression of various behaviours covaried. We reasoned that defensive behaviours should be elevated in the presence of mites, and that they should be inversely related to the numbers of mites colonizing the host or positively related to the numbers of mites lost or removed from hosts (experiments 2 and 3). We also examined whether these specific behaviours versus other behaviours were diminished for nutritionally stressed larval damselflies (experiment 4). We conducted these experiments sequentially, each building on former experiments. In summary, we explored whether factors that reduce the condition of damselflies may compromise their ability to defend against mites. METHODS All experiments were done in June and July 1997 and 1998. In both summers, we collected damselflies as F-1 larvae (where F means final instar) from Burns Conservation Area (4329 N, 8002 W; near Mississauga, Ontario, Canada). We reared larval mites in the laboratory from eggs readily found at Charleston Lake (4432 N, 7601 W; near Gananoque, Ontario, Canada). We kept damselflies at low temperatures (10C) to slow their development, such that enough larvae moulted to their final instar at about the same time the mite eggs were hatching. Generally, once some clutches of mite eggs hatched, we housed the damselflies individually in 75-ml vials at room temperature (ca. 22C) and fed them a diet of white worms (enchytraeid worms) and Daphnia magna (L.). From then on, we checked them for moulting every 24 h in the early morning. After they moulted into their final instar (wing pads reaching abdominal segment four or five; Baker 1986), equal numbers of final-instar larvae were placed into nutritional or other groups. To determine whether nutritional stress affected numbers of mites attending damselflies over time (experiment 1), we fed some damselflies one white worm twice weekly and others one white worm each day, for a week. We paired a nutritionally stressed and unstressed damselfly of the same sex and placed each pair into a common 75-ml
vial (N=46 for each group). We tested the assumption that our nutritional stress had measurable consequences on condition (stressed damselflies were significantly lighter than unstressed damselflies; paired t test: t22 =5.08, P<0.01). Subsequently, we introduced >100 larval mites into the vial. Clutches of A. pseudosuperior are extremely clumped in nature with over 10 000 eggs in <1 m2 (M. R. Forbes, personal observation); hence, some larval damselflies are likely to experience high numbers of mites challenging them. We controlled for potential variability in laboratory conditions, mite numbers and variation in mite vigour in attacking hosts by using a matched-pair design. Furthermore, this design allowed the mites to have a reasonable probability of encountering both damselflies, and thus allowed them to make a choice between hosts if they possessed this ability. In a previous study, we found that A. pseudosuperior had a slight tendency to attach to larval I. verticalis that were close to emergence (Leung et al. 1999). After 1–2 h, we removed the damselflies, counted the number of attached mites, and placed the larval damselflies individually into separate vials filled with water that was free of mites. After 24 h, we counted the number of mites remaining on each damselfly. The change in the number of mites could indicate the efficiency with which stressed and unstressed damselflies removed mites. To determine specific behaviours associated with mite presence or possible defence against mite colonization (experiment 2), we first kept damselflies at ca. 22C and fed them ad libitum with enchytraeid worms or D. magna every 1–2 days. After they had spent ca. 1 week as a final instar, we transferred them to new vials and later filmed them to record their behaviour for 15 min in the absence of mites (as a control). Then, using the same damselfly larvae, we introduced 30 mites, and recorded behaviour for an additional 45 min. After the 45-min filming period, we added another 60 mites and left the damselflies for an additional 1 h before they were removed and their numbers of attending mites counted. We assumed that relative times spent in various behaviours during the initial 45 min were representative of behaviours for the next 1 h. To record damselfly behaviour, we used two video cameras and recorders and placed three damselflies in the view of each camera; white paper was placed between the vials so the damselflies could not see each other (thereby affecting their behaviour for reasons unrelated to challenge with mites; e.g. conspecific interactions). We recorded the following behaviours (after Baker 1981; Forbes & Baker 1990; Baker & Smith 1997): time spent in Crawl forward, Swim, Groom and Abdomen wave, and frequency of Slash, Wiggle and Labial strike (Table 1). Because of the rapidity of Slash and Wiggle and the fact that we lost resolution when we played tapes at slow speed, it was often difficult to distinguish clearly between the two. Therefore we chose to combine the occurrences of these two behaviours (henceforth referred to as Slash for convenience). We added 60 mites after the initial filming period because we wanted to increase our range in numbers of colonizing mites and thus our ability to detect relations between specific behaviours and numbers
LEUNG ET AL.: STRESS AND BEHAVIOURAL IMMUNITY
Table 1. Behaviours of larval damselflies Behaviour
Abdomen wave
Crawl forward Groom
Labial strike Slash Swim Wiggle
Description and significance
The animal repeatedly performed a low-amplitude (at an angle ca 10° from the rest of the body) transverse motion of the last six or seven abdominal segments and lamellae. Unlike previous investigations (Baker & Smith 1997), single abdomen bends were not recorded Any forward movement in which the animal maintained continuous limb contact with the substrate All movements in which either the tarsal claws or distal segments of a leg or pair of legs were rubbed against any part of the body. Although grooming can remove mites (Forbes & Baker 1990), its efficacy in mite removal has not previously been examined. Grooming can be variously termed scratch and rub, but in this study it included another behaviour involving a lateral bending of the distal tip of the abdomen and lamellae to a point where they were brought over the head and rubbed against the dorsum of the head and thorax Rapid extension of labium, generally in the direction of prey or debris on the floor of the vial. Strike was often directed at swimming larval mites A rapid lateral bending of the abdomen, resulting in caudal lamella being displaced ≥90° Any movement through the water in which the animal did not maintain contact with the substrate. The movement was usually accomplished by a transverse undulation of the abdomen and lamellae A rapid lateral undulation of the entire abdomen resulting in a series of waves passing down the body; each occurrence lasted only 1–2 s. Wiggle and Slash were combined and are simply referred to as Slash
After Forbes & Baker (1990) and Baker & Smith (1997).
of mites attaching, should these relations exist. We identified behaviours as defensive if they increased in the presence of mites, and if duration performing a behaviour was inversely correlated with number of attaching mites. To control for differences in duration of observation, we expressed time spent performing a behaviour as a proportion of the total time the damselfly was watched. To determine whether specific behaviours were associated with mite removal over a longer duration after their initial attachment, we exposed each of 16 damselflies to 75 mites (experiment 3: each damselfly was housed in a smaller 15-ml vial with a perch). We allowed mites to colonize damselflies for 1–2 h and then counted the number of mites on the damselflies. Afterwards, we placed the damselflies in vials with mite-free water, to prevent further attachment. As above, we used two video cameras with a maximum of three damselflies in view of each camera. We recorded behaviour for 1 h. After recording, we recounted the number of mites on each damselfly to determine the number of mites removed. We further identified behaviours as defensive if they had increased in the presence of mites (in the previous experiment), and if duration performing a behaviour was positively correlated with the number of mites removed over time. In all cases of filming, we could not directly record the number of mite contacts and number of mite removals without much more sophisticated equipment and fewer mites to enable tracking of individual mites. To determine if specific defensive behaviours were reduced in the presence of nutritional stress (experiment 4), we fed stressed damselfly larvae six D. magna or one small enchytraeid worm every day and unstressed damselflies ad libitum D. magna or a mixture of enchytraeid worms and D. magna each day. Again, this difference in diet resulted in higher nutritional status of the unstressed group, as measured by wet mass after emergence (twoway ANOVA controlling for sex: F1,20 =7.92, P=0.01). We also used principal components analysis (PCA) to
determine whether specific behaviours covaried with one another such that changes that were due to stress could be viewed as a single correlated response. After the damselflies moulted into final instars, we allowed them to develop for 6–7 days (15 in stressed group and 16 in unstressed group). In so doing, we ensured that individuals in the stressed group received far fewer feedings over a fairly long period. We then challenged the damselflies by placing ca. 20 mites into their vials/day. On the fifth or sixth day, we filmed them for 1 h, after introducing mites into their vials. We then recorded behaviours as above. We transformed all proportions using arcsine(x0.5) before conducting statistical tests (Zar 1984). However, we used the raw proportions for descriptive purposes (i.e. meanSE) as this would be biologically more informative. RESULTS We used proportions to remove variability in the total number of mites attaching to damselfly larvae in each matched pair. The proportion of mites attaching to stressed versus unstressed animals did not differ (paired t test: t45 =0.94, P=0.35; Fig. 1). However, the proportion of mites removed over a 24-h period differed significantly in the predicted direction, that is, unstressed animals lost or removed proportionately more mites than stressed animals (paired t test: t45 =3.62, P<0.001; Fig. 1). This difference in removal rates (see below) resulted in a difference in the proportion of mites remaining on stressed versus unstressed animals (paired t test: t45 =3.03, P=0.004; Fig. 1). After 24 h, stressed animals had 17% more mites than unstressed animals. We used paired t tests to examine which behaviours increased in the presence of mites. Time spent in Groom and Abdomen wave and frequency of Slash and Labial strike all increased in the presence of mites and remained significant after sequential Bonferroni adjustment (Rice
1095
ANIMAL BEHAVIOUR, 61, 6
1.00
0.75 Proportion
1096
0.50
0.25
0.00
Colonizing
Removed
After 24 h
Figure 1. Differences between stressed (h) and unstressed (") damselflies in terms of the proportion of mites initially colonizing, the proportion of mites removed over 24 h, and the proportion of mites remaining after 24 h. Stressed and unstressed damselflies were paired; thus individuals within a pair were subjected to identical parasite conditions (46 pairs). Proportions were determined for each pair, and the average proportions are shown. Statistical tests (see text) were conducted on arcsine(x0.5) transformed proportions (Zar 1984). However, the figure shows raw untransformed values. Error bars represent 1 SE.
1989), but time spent in Crawl forward and Swim did not relate to mite presence (Table 2). In brief, damselfly larvae exposed to mites were more active. Only Abdomen wave, Labial strike and Swim were related to the number of attaching mites (Table 2). However, Abdomen wave was positively related to the number of mites that attached, whereas Labial strike was inversely related, as predicted if this behaviour was important in defence. Swim became nonsignificant after sequential Bonferroni correction. In the third experiment, where larvae were provided with more time to remove mites, Groom and Labial strike were positively related to the proportion of mites removed, but Crawl, Swim, Abdomen wave and Slash were not (Table 2). Groom remained significant after sequential Bonferroni correction, but Labial strike did not. Taken together, these results suggest that only Labial strike and Groom fulfil the criteria of defensive behaviours. These behaviours increased in the presence of mites, but one appeared effective when mites were colonizing hosts (Labial strikes) whereas Groom had an effect on mite removal. Groom may be combined with Labial strike to reduce the threat from mites, by first dislodging and then attacking them (as we occasionally observed during this study). In experiment 4, unstressed damselflies performed Groom, Abdomen wave, Slash and Swim more often than stressed damselflies (Table 2). These behaviours remained significant after sequential Bonferroni corrections, except for Swim. However, stressed and unstressed damselflies did not differ in Crawl forward or Labial strike. In brief, stressed damselflies were considerably less active than unstressed damselflies; however, whereas some behaviours remained unaffected, the defensive behaviour Groom was reduced considerably.
It is relevant that only grooming was affected appreciably by nutritional stress, because we found appreciable covariation between behaviours. In the PCA, the first axis accounted for ca. 45% of the combined variation in behaviour. The percentage time spent grooming, waving, slashing and striking all correlated strongly on the first axis (factor loadings >0.70 or < 0.70); the first three behaviours were positively related, whereas striking was inversely related, to that first axis. Swimming and crawling were not strongly related to the first axis (factor loadings=0.10 and 0.35, respectively). However, these two behaviours loaded strongly and inversely on the second PC axis (factor loadings < 0.80) which accounted for ca. 25% of the remaining variation in behaviour. DISCUSSION Synergism between stress and parasitism may occur if stress reduces the ability of hosts to defend against parasites or affects exposure to parasites, thereby resulting in a higher intensity of parasites on a given host. In summarizing the ecology of ectoparasitic insects, Marshall (1981) argued that high intensities of ectoparasitism on hosts might be more a consequence of poor condition of hosts, rather than a cause of that poor condition. Yet scant empirical evidence has addressed whether stressed invertebrate hosts are more likely to be parasitized than unstressed hosts. The majority of studies on stress– immunocompetence relations have been done on vertebrates (see references in Introduction). Our experiments were designed to determine if there was a potential for stress to affect susceptibility of an invertebrate host to ectoparasitism. Our results provide a sufficient explanation for hosts in lower condition having more ectoparasites. In brief, stressed hosts accumulated more mites, grooming appeared effective in removing mites, and grooming was reduced in the presence of stress. We are confident that mites induced increased grooming, rather than the alternative of damselflies taking time to acclimate to vials and then expressing ‘normal’ behaviours. We believe this because grooming occured immediately after mite introductions, and in another paired trial where mites were added versus not after the same period of acclimation in vials, only damselflies with mites added increased grooming (Forbes & Baker 1990). We argue further that nutritional stress, as examined in our study, is likely to play an important role under field conditions. In summarizing studies on food limitation in larval damselflies Corbet (1999) noted that feeding rates in nature are generally less than maximum possible feeding rates in captivity. In our experiments, feeding rates were made to be high and low: the low-condition treatment reflected what can occur in nature (since damselflies could still emerge at this low condition). Such results can explain the observed inverse relations between mite number and damselfly condition, as seen in other field studies (Forbes & Baker 1990; Leung & Forbes 1997). It is perhaps not surprising that labial strike was not reduced in the presence of nutritional stress since this is a
LEUNG ET AL.: STRESS AND BEHAVIOURAL IMMUNITY
Table 2. Effect of mite presence on damselfly behaviour (experiment 2), effect of damselfly behaviour on numbers of mites that attached initially (experiment 2) and on the proportion of mites removed after colonization (experiment 3), and effect of nutritional stress on behaviour (experiment 4) Behaviour
Experiment 2 Mite presence
Experiment 2 Mite attachment
Experiment 3 Mites removed
Experiment 4 Nutritional stress
6.46±2.97 29.26±3.08 F1,27 =25.70
23.36±3.69 17.73±3.81 t24 =2.59
r=0.55 F1,21 =9.25
r= −0.21 F1,13 =0.58
Crawl forward
0.72±0.19 1.20±0.37 t24 =0.55
r=0.18 F1,21 =0.76
r=0.36 F1,13 =1.98
1.20±0.27 1.12±0.27 F1,27 =0.07
Groom
7.70±1.05 1.35±0.38 t24 =6.55
r= −0.07 F1,21 =0.12
r=0.65 F1,13 =9.70
3.36±0.49 7.57±1.06 F1,27 =10.98
Labial strike
0.27±0.05 0.0±0.00 t24 =7.08
r= −0.61 F1,21 =12.22
r=0.53 F1,13 =5.14
1.23±0.33 0.60±0.16 F1,27 =3.45
Slash
0.53±0.11 0.05±0.02 t24 =6.06
r= −0.03 F1,21 =0.023
r=0.22 F1,13 =0.64
0.16±0.06 1.16±0.30 F1,27 =14.89
Swim
0.54±0.16 0.33±0.12 t24 =1.17
r=0.45 F1,21 =5.41
r=0.18 F1,13 =0.44
0.07±0.04 1.15±0.82 F1,27 =4.24
Abdomen wave
For experiment 2, effect of mite presence on behaviour was analysed with sequential Bonferroni-corrected paired t tests, predicting increases in activity in the presence of mites. First and second entries refer to mean percentage of time±SE spent in that behaviour in the presence and absence of mites, respectively. We also give t values. Column 2 shows the correlation (correlation coefficient and F value) between behaviour and proportion of mites attaching initially, analysed with ANCOVA models to control for sex. For experiment 3, the effect of behaviour on the proportion of mites removed is given, analysed with ANCOVA models to control for sex. For experiment 4, the effect of stress on damselfly behaviour was analysed with a two-way ANOVA to control for sex. First and second entries refer to mean percentage of time±SE spent in a behaviour by stressed and unstressed larval damselflies, respectively. Behaviours did not differ between sexes in any test and are not presented. All percentages were first transformed with arcsine(x0.5) before statistical analyses were conducted. However, we present raw untransformed values for means±SE. Bold values indicate significance at P<0.05 after sequential Bonferroni correction.
foraging behaviour. While it may be effective in reducing challenge from mites, these reductions are not enough to prevent associations between condition and ectoparasitism. As mentioned previously, these findings do not preclude hypotheses based on costs to resisting mite colonization or costs to actual phoresy (cf. Elzinga & Broce 1988), but only suggest that individuals already in poor condition may be subjected to greater eventual parasitism. We reason that inverse relations between wet mass at emergence and number of attaching mites would be due to damselflies in poor condition being less likely to engage in antimite behaviours. However, alternative explanations exist. For example, in nature, those damselflies in areas most frequented by mites may groom more, lose mass and still have more mites attach to them. Another possible mechanism is that mites might have preferentially selected poor-condition hosts over others or may have abandoned well-fed hosts more frequently. We deal with each of these alternatives in turn. Grooming in damselflies is likely to be costly. It may be energetically expensive or may detract needed time from other activities such as foraging. Thus, one might expect damselflies that encounter more mites to have lower rates of mass gain. We cannot fully address this alternative at this time, but a few points are worth mentioning. Since,
in nature, mites are probably only locally numerous, dispersal of larval damselflies may be an additional line of defence. The effects of mites on dispersal tendencies and the relation between success in dispersal and condition of larval hosts needs to be examined. There is the additional potential complication of predation. Like many other prey, damselflies trade movement associated with feeding against risk of predation (Dixon & Baker 1988). Because grooming also contributes to movement and potential detection by fish predators, we might expect damselflies confronted with high predation risk in nature to reduce both feeding and antiparasite behaviours, thus producing low-condition animals with relatively high parasitism compared with other animals with locally lower risk of predation. However, Baker & Smith (1997) showed that damselflies continue to groom when confronted with mites even when fish predators are present and this could result in higher predation risk, but not low-condition animals with high parasitism. Two other points are worth considering. First, mites may select stressed over unstressed hosts (we did not consider the reverse scenario given inverse relations between condition and mite numbers observed in the field). This could occur, for instance, if a chemical cue indicated that stressed individuals were unable to mount an effective response. However, the proportion of mites
1097
1098
ANIMAL BEHAVIOUR, 61, 6
initially attaching to stressed hosts equalled those of unstressed hosts, suggesting that they were not choosing their hosts based on host condition. Furthermore, in another study using these same species and a similar laboratory set-up, we found that mites preferentially colonized old larvae close to emergence (Leung et al. 1999) suggesting that if mites were making a choice in the present study, we should have been able to document it. Although stressed and unstressed damselflies are likely to have differed in degree of development in this study, variation in developmental stage was much greater in Leung et al.’s (1999) study , that is, none of the damselflies in the present study was within 1–2 days of emergence. Second, mites may also have differentially abandoned well-fed hosts. While we cannot explicitly address this possibility, we do not believe that it is likely. Damselflies in better condition live for longer after emergence. Thus, mites attached to these damselflies have a greater probability of returning to water and completing their life cycle (i.e. mites should prefer well-fed damselflies). Furthermore, mites may be able to obtain nutrients more quickly on well-fed adult damselflies. Thus, quicker development time (or better mite condition) also would not be a reason for higher abandonment of well-fed hosts. These observations increase our general understanding of parasite–host relations. In nature, stressed hosts are likely to have higher rates of mite attachment and lowered fitness as adults. Degree of mite parasitism on adults has been related to reduced longevity (Forbes & Baker 1991; Leung & Forbes 1997), reduced fecundity in fieldcaught females (Forbes & Baker 1991) and reduced mating success in males (Forbes 1991; but see Andre´ s & Cordero 1995; Rolff et al. 2000). However, positive life history trade-offs are also possible (e.g. Rolff 1999 showed that mite parasitism may increase offspring size in damselflies). In previous correlative studies showing a fitness association with mites, the impact of parasites could not be decoupled from the impact of lower condition potentially resulting in higher parasitism in the first place. Here, we suggest that low condition, in fact, does result in higher parasitism and provides at least a partial mechanism for relations observed in nature. Such studies now have to be combined with the fitness functions in relation to mite intensity for both host and mites, to ascertain what effect such synergisms will have on both populations (cf. Jaenike 1996). Acknowledgments We thank C. Elkin for all his help in the laboratory, B. P. Smith for help with collecting mites and two anonymous referees for comments on the manuscript. This research was supported by a Natural Sciences and Engineering Research Council (NSERC) Scholarship to B.L., and NSERC grants to R.L.B. and M.R.F. References Agyemang, K., Dwinger, R. H., Little, D. A., Leperre, P. & Grieve, A. S. 1992. Interaction between physiological status in N’Dama
cows and trypanosome infections and its effect on health and productivity of cattle in Gambia. Acta Tropica, 50, 91–99. Anderson, R. M. & May, R. M. 1981. The population dynamics of microparasites and their invertebrate hosts. Philosophical Transactions of the Royal Society of London, Series B, 291, 451–524. Andre´ s, J. A. & Cordero, A. 1998. Effects of water mites on the damselfly, Ceriagrion tenellum. Ecological Entomology, 23, 103– 109. Baker, R. L. 1981. Behavioral interactions and use of feeding areas by nymphs of Coenagrion resolutum (Coengrionidae: Odonata). Oecologia, 49, 353–358. Baker, R. L. 1986. Developmental stages and the analysis of zygopteran life histories. Freshwater Ecology, 3, 325–332. Baker, R. L. & Smith, B. P. 1997. Conflict between antipredator and antiparasite behavior in larval damselflies. Oecologia, 109, 622– 628. Barker, D. E., Khan, R. A. & Hooper, R. 1994. Bioindicators of stress in winter flounder, Pleuronectes americanus, captured adjacent to a pulp and paper mill in St. George’s Bay, Newfoundland. Canadian Journal of Fisheries and Aquatic Sciences, 51, 2203–2209. Carballo, M., Munox, M. J., Cuellar, M. & Tarazona, J. V. 1995. Effects of waterborne copper, cyanide, ammonia, and nitrate on stress parameters and changes in susceptibility to Saprolegniosis in rainbow trout (Oncorhynchus mykiss). Applied and Environmental Microbiology, 61, 2108–2112. Carey, C. 1993. Hypothesis concerning the causes of the disappearance of boreal toads from the mountains of Colorado. Conservation Biology, 7, 355–362. Corbet, P. S. 1999. Dragonflies: Behavior and Ecology of Odonata. Ithaca, New York: Cornell University Press. Dixon, S. M. & Baker, R. L. 1988. Effects of size on predation risk, behavioural response to fish, and cost of reduced feeding in larval Ischnura verticalis (Coenagrionidae: Odonata). Oecologia, 76, 200– 205. Elzinga, R. J. & Broce, A. B. 1988. Hypopi (Acari: Histiostomatidae) on house flies (Diptera:Muscidae): a case of detrimental phoresy. Journal of the Kansas Entomological Society, 61, 208–213. Evans, S. M., Dawson, M., Day, J., Frid, C. L. J., Gill, M. E., Pattisina, L. A. & Porter, J. 1995. Domestic waste and TBT pollution in coastal areas of Ambon Island (eastern Indonesia). Marine Pollution Bulletin, 30, 109–115. Folstad, I., Nilssen, A. C., Halvorsen, O. & Andersen, J. 1989. Why do male reindeer (Rangifer tarandus tarandus) have higher abundance of second and third instar larvae of Hypoderma tarandi than females? Oikos, 55, 87–92. Forbes, M. R. L. 1991. Ectoparasites and mating success of male Enallagma ebrium damselflies (Odonata: Coenagrionidae). Oikos, 60, 336–342. Forbes, M. R. L. & Baker, R. L. 1990. Susceptibility to parasitism: experiments with the damselfly Enallagma ebrium (Odonata: Coenarionidae) and larval water mites, Arrenurus spp. (Acari: Arrenuridae). Oikos, 58, 335–341. Forbes, M. R. L. & Baker, R. L. 1991. Condition and fecundity of the damselfly, Enallagma ebrium (Hagen): the importance of ectoparasites. Oecologia, 86, 335–341. Jaenike, J. 1996. Population consequences of parasite aggregation. Oikos, 76, 155–160. Ko ¨ nig, C. & Schmid-Hempel, P. 1995. Foraging activity and immunocompetence in workers of the bumble bee, Bombus terrestris L. Proceedings of the Royal Society of London, Series B, 260, 225–227. Leung, B. & Forbes, M. R. 1997. Fluctuating asymmetry in relation to indices of quality and fitness in the damselfly, Enallagma ebrium (Hagen). Oecologia, 110, 472–477.
LEUNG ET AL.: STRESS AND BEHAVIOURAL IMMUNITY
Leung, B., Baker, R. L. & Forbes, M. R. 1999. Grooming decisions by damselflies, age-specific colonization by water mites, and the probability of successful parasitism. International Journal for Parasitology, 29, 397–402. Marshall, A. G. 1981. The Ecology of Ectoparasitic Insects. London: Academic Press. Mooring, M. S. & Hart, B. L. 1995. Differential grooming rate and tick load of territorial male and female impala, Aepyceros melampus. Behavioral Ecology, 6, 94–101. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution, 43, 223–225. Rolff, J. 1999. Parasitism increases offspring size in a damselfly: experimental evidence for parasite-mediated maternal effects. Animal Behaviour, 58, 1105–1108.
Rolff, J., Antvogel, H. & Schrimpf, I. 2000. No correlation between ectoparasitism and male mating success in a damselfly: why behavior matters. Journal of Insect Behavior, 13, 563–571. Sheldon, B. C. & Verhulst, S. 1996. Ecological immunity: costly parasite defences and trade-offs in evolutionary ecology. Trends in Ecology and Evolution, 11, 317–321. Siva-Jothy, M. T., Tsubaki, Y. & Hooper, R. E. 1998. Decreased immune response as a proximate cost of copulation and oviposition in a damselfly. Physiological Entomology, 23, 274–277. Smith, B. P. 1988. Host-parasite interaction and impact of larval water mites on insects. Annual Review of Entomology, 33, 487–507. Zar, J. H. 1984. Biostatistical Analysis. 2nd edn. Upper Saddle River, New Jersey: Prentice Hall.
1099