ANIMAL BEHAVIOUR, 2008, 75, 823e826

doi:10.1016/j.anbehav.2007.07.002

Available online at www.sciencedirect.com

Hairworm response to notonectid attacks

MA RTA I. SÁ NC H EZ* , FLEUR PON TON *, DOR OTHÉE M IS SÉ* , D AVI D P. HU G H ES† & FR ÉD ÉRIC THOMAS*
*GEMI, UMR CNRS/IRD, Montpellier
yCentre for Social Evolution, Institute of Biology, Universitetsparken, Copenhagen

(Received 20 March 2007; initial acceptance 11 May 2007;

final acceptance 4 July 2007; published online 24 October 2007; MS. number: 9319)

Very few parasite species are directly predated but most of them inherit the predators of their host. We
explored the behavioural response of nematomorph hairworms when their hosts are preyed upon by
one of the commonest invertebrate predators in the aquatic habitat of hairworms, notonectids. The hair-
worm Paragordius tricuspidatus can alter the behaviour of its terrestrial insect host (the cricket Nemobius syl-
vestris), causing it to jump into the water; an aquatic habitat is required for the adult free-living stage of the
parasite. We predicted that hairworms whose hosts are captured by a notonectid should accelerate their
emergence to leave the host before being killed. As predicted, the emergence length of the worm was sig-
nificantly shortened in cases of notonectid predation, but the exact reason of this response seems to be
more complex than expected. Indeed, experimental manipulations revealed that hairworms are remark-
ably insensitive to a prolonged exposure to predator effluvia which notonectids inject into prey, so accel-
erated emergence is not a protective response against digestive enzymes. We discuss other possibilities for
the accelerated exit observed, ranging from unspecific stress responses to other scenarios requiring consid-
eration of the ecological context.

Ó 2007 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

Keywords: crickets; hairworms; manipulation; Nemobius sylvestris; Notonecta glauca; notonectids; Paragordius tricuspidatus;
predation

Very few parasite species have direct predators but most cycles (Poulin 1998; Lafferty 1999; Parker et al. 2003; see
inherit those of their hosts (Thomas et al. 2002a). This pe- Choisy et al. 2003 for review).
culiar ecological context has favoured the evolution of di- Recently, a novel antipredator strategy by parasites was
verse adaptations by parasites to avoid succumbing to found in the hairworm Paragordius tricuspidatus (Nemato-
predation upon their host. The principal and most com- morpha: Gordiida) parasitizing orthoptera. The larval
mon response is to reduce the encounter rate with poten- stages of this parasite develop in the cricket Nemobius syl-
tial predators by altering the behaviour of the host vestris, which is terrestrial, but the adult phase is free living
(Brodeur & McNeil 1994; Levri 1998; Lafferty et al. 2000; and aquatic in ponds and streams of southern France. To
Thomas et al. 2002b; Haine & Rigaud 2005). Where preda- exit the cricket and enter the water, the mature parasite al-
tion is unavoidable, certain parasites have developed the ters the behaviour of the insect host, making it seek out
capacity to encyst in the predator until a new favourable and jump into water (i.e. induced host ‘suicide’; Thomas
event occurs (e.g. Robert et al. 1988; Pampoulie et al. et al. 2002c). These water areas are frequently inhabited
2000) or resist a complete transit in the predator gut by both vertebrate and invertebrate predators. Ponton
(McFarland et al. 2003). Over evolutionary time, parasites et al. (2006a, b) showed that if the crickets that enter the
have also evolved the capacity to colonize and exploit the water are eaten by fish or frogs then the hairworm is able
predators of their host, thereby evolving complex life to escape not only from its insect host but also from the di-
gestive tract of the predator. The worm emerges alive from
the mouth, gills or nose of the predators and continues its
Correspondence: M. I. Sánchez, GEMI, UMR CNRS/IRD 2724, IRD,
life cycle without any fitness costs. This escape response
911 av. Agropolis, BP 64501, 34394 Montpellier cedex 5, France was the first example of a parasite, or any organism, surviv-
(email: marta.sanchez@cefe.cnrs.fr). D. Hughes is at the Centre for ing predation in this way (Ponton et al. 2006a).
Social Evolution, Institute of Biology, Universitetsparken 15, In the forest ponds of southern France, predators
DK-21000 Copenhagen, Denmark. include not only vertebrates but also several predatory
823
0003e 3472/08/$32.00/0 Ó 2007 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
824 ANIMAL BEHAVIOUR, 75, 3

invertebrates, the most common of which are notonectids experiment. Crickets were collectively kept in aquaria
or backswimmers (Hemiptera, Notonectidae, Notonecta (30  25 cm, height 16 cm) provided with ad libitum
glauca). Notonectids are voracious generalist predators food (in equal proportions: cereals, fish food Tetra Ani
that attack just about any prey that they can overpower Min, dry gammarids and dry tubifex) and humidified cot-
ranging from mosquito larvae to pike fry. Notonectids ton. All individuals, notonectids and crickets, were placed
are known to structure ecological communities (Murdoch in undisturbed rooms which had a 16:8 h light:dark cycle
& Scott 1984; Murdoch et al. 1984; Geddes 1986; Arner that mimicked the natural photoperiod at capture period.
et al. 1998; Blaustein 1998; Pace et al. 1999) and influence The analysis was based on 34 infected crickets in the first
the oviposition behaviour of mosquitoes (Chesson 1984). experiment (14 tested in presence of a notonectid and 20
They will attack orthoptera that accidentally fall into without predator) and 35 in the second experiment (20
water (F. Thomas, personal observations). Like all hemip- crickets in presence of predator and 15 without predator).
terans, notonectids lack chewing mouthparts and feed
using a rostrum, or ‘sucking-beak’. These predatory he-
mipterans inject digestive juices down one canal of the Experimental Procedure
rostrum and suck up the digested prey through another We presented infected crickets to a notonectid to
canal. The significance of such feeding for any parasite determine the hairworm’s response to predation. Experi-
of the prey item is two-fold. First it means that, unlike ver- ments were performed during the afternoon (between
tebrate predation, the body of the prey is not physically 1400 and 2000 hours). Infected crickets were gently placed
ingested inside the predator; instead it stays outside and into a tank of water containing a notonectid. Control
is released when empty. Second, for parasites occupying infected crickets were placed into a tank without a noto-
the host’s haemocoel (as hairworms do), it means that di- nectid. We considered a predation test valid only if the
gestive juices will be encountered immediately. notonectid attacked the cricket immediately after its
The aim of this study was to determine whether hair- entrance in the water. In no cases had worms begun to
worms display antipredator behaviour against notonectids emerge at the moment the cricket was attacked; that is,
and, if so, to examine how it differs from antipredator the parasite was fully inside the cricket.
behaviour against vertebrates. We predicted that hairworms To assess whether there were negative effects on noto-
would avoid notonectid predation by means of a more rapid nectids because of predation on their host we experimen-
emergence from the cricket host when the host was tally prevented hairworm emergence by covering the
attacked. We also wanted to assess the cost of predation terminal part of the cricket’s abdomen with superglue
by notonectids on hairworms so we experimentally pre- (i.e. openings were blocked). Once the attack had finished
vented worms from escaping their host following a noto- (the notonectid released the dead cricket) we gently
nectid attack. We discuss the relevance of the hairworm opened the abdomen to liberate the worm inside and
response in the context of antipredator strategies. examine its state. We determined whether the worm was
dead or alive. If alive we determined whether it could still
METHODS swim and whether it was able to reproduce (lay eggs for
females and donate a spermatophore for males). Worms
Sampling from crickets treated with superglue in the absence of
predators were used as a control. To avoid confounding
As in Thomas et al. (2002c), infected N. sylvestris were effects of multiple infection, only individuals singly
captured at night (between 2200 and 0100 hours) around infected were used for the analysis.
a private swimming pool (15  10 m) and on a parking
area located in Avènes les Bains (southern France, 70 km
north of Montpellier). All specimens were collected during Statistical Analysis
July 2006. The swimming pool and the parking area are
beside a forest that is crisscrossed by small streams in Statistical tests were performed following Sokal & Rohlf
which adult P. tricuspidatus were commonly found during (1981) and Siegel & Castellan (1988). Homogeneity of var-
the summer. Paved areas allowed direct observation and iance between groups was tested using the Levene statis-
capture of infected crickets moving from the forest. Previ- tic. Since variance heterogeneity between the attacked
ous observations (Thomas et al. 2002c) revealed that and the not-attacked groups was one of the predictions
crickets detected on the concrete area were always infected concerning the length of the emergence (the time that it
by at least one worm. takes to emerge), we used a Welch ANOVA on untrans-
Notonectids in surrounding ponds were sampled on the formed and ln-transformed data (Welch 1951) to compare
same date using a net. Captured individuals were kept groups. Welch ANOVA is suitable since it allows compari-
singly in plastic bottles (8 cm diameter, 20 cm height) that sons when variances are unequal and the data are approx-
were placed in aquaria (60 cm length, 30 cm height, 30 cm imately normalized. All tests were two tailed.
width) and filled with constantly aerated water. The bot-
toms of the bottles were covered by a net (2 mm mesh RESULTS
size), allowing water from the tank to circulate freely
through all the compartments. Notonectids were acclima- As predicted, worms that were inside cricket hosts
tized for a period of 4 days during which no food was pro- that were predated by notonectids emerged signi-
vided to induce a fast attack response required for the ficantly faster than controls (110 versus 380 s; Welch
 
S ANCHEZ ET AL.: HAIRWORM RESPONSE TO NOTONECTID ATTACKS 825

ANOVA: F1,19.38 ¼ 8.34, P ¼ 0.0093; Fig. 1). The variance aquatic habitat needed by adult hairworms, we predicted
(Levene test: P < 0.0001) was significantly shorter when that evolving antipredator responses to other kinds of nat-
the host was attacked by a notonectid than when there ural enemies should also be a part of the wider strategy of
was no predator (Fig. 1). Once it had emerged, the adult hairworms for the completion of their life cycle.
hairworm was never directly attacked by the notonectid. Our experiments indicated that the time taken by the
The second experiment revealed that the use of super- worm for full emergence from its host was significantly
glue prevented hairworm emergence so that the worm was reduced in case of notonectid attack. Such an accelerated
constrained to remain inside the cricket host throughout exit might be interpreted as an antipredator response from
the entire predation event (approximately 5 min). Once the worm. Ponton et al. (2006a) showed that time was in-
the dead cricket was released we examined the worm deed a key component in the antipredator response of
and found that all worms in the predated group were alive hairworms towards vertebrates: if a worm was not observed
(as were those in the control). There was no significant coming out the fish within 5 min, it never exited, presum-
size difference between worms from the two groups ably because it died in the hostile environment of the pred-
(Student’s t test: t27 ¼ 0.34, P ¼ 0.74) nor observable ator’s stomach. Notonectids do not ingest crickets as fish
difference in their swimming ability. Finally, there was do but instead pump digestive juices into the host, and
no significant difference between the two groups in the this, we speculate, would be hazardous to the hairworm
proportion of worms mating (80% for predated worms, that occupies the cricket’s abdomen. We reasoned a priori
67% for control; Fisher exact test: P ¼ 0.405). that the harsh environment would impair the viability
and/or the reproductive capacities of the worm. Against
our expectations, the results of the second experiment
DISCUSSION showed that notonectid attack itself does not appear to
The role of predation in the evolutionary ecology of harm the worm. Hairworms from attacked hosts were not
animal communities has long been a focus of ecological only alive and able to swim as well as controls but they
research but, to date, sparse attention has been given to also reproduced. At the adult stage, hairworms are simply
the responses of parasites when their hosts are victims of a mobile bag of gametes. The length of the worm is highly
predation. Because half of life on earth may be parasitic correlated with the number of gametes and hence to fit-
(Price 1980) and because parasite virulence is shaped by ness (Hanelt et al. 2005). Therefore, in our assessment of
predation (Read 1994), this is a serious oversight. Previous fitness we measured worm length and general vigour.
work (Ponton et al. 2006a, b) showed that hairworms have The conclusion that notonectid attack has no significant
evolved an original solution to the predation of their host effect on worm fitness is based on the measurement of
by vertebrates: they crawl out from the stomach of the the relatively few fitness surrogates that we are able to mea-
predator. Since invertebrate predators also inhabit the sure. Arguably there are potentially other aspects of fitness
that may be negatively impacted by the predation event
which we were not able to measure. For example, the biol-
ogy of this parasite system does not lend itself to other
measurements such as lifetime fecundity or longevity be-
cause egg release from this mobile bag of eggs may occur
even when the worm is apparently dead, and ascertaining
the moment of death is problematic since worms can
remain completely inactive for weeks (F. Thomas &
1400 A. Schmidt-Rhaesa, personal observations). Egg shedding
might even occur from dead worms.
1200 It would appear that the accelerated exit of the worm
when its host is attacked by a notonectid is not a response
to the physiological damage posed by exposure to digestive
1000
juices. This might be due to the highly resilient cuticle that
adult hairworms possess. The physical and chemical re-
800 sistance capabilities of the skin lie in its ultrastructural
Time (s)

organization and biochemical composition (Swanson

600 1970; Schmidt-Rhaesa 1996; Brivio et al. 2000). The cuticle
is multilayered and extremely complex (Schmidt-Rhaesa
2004), which may provide effective protection from the
400
harsh environment of the notonectids’ effluvia. The rapid
emergence, despite no obvious mortality risks, may occur
200 because worms cannot discriminate between host attacks
that are dangerous and those that are not. If no proximate
0 factor permits the worm to make such a distinction, it is
Host predated Control undoubtedly better to accelerate the exit in all cases, espe-
Figure 1. Mean  SE time taken to emerge for worms exposed to cially if this has only trivial fitness costs as suggested by our
notonectid predation and those used as a control (in absence of data and by Ponton et al (2006b). It is also possible that the
predators). predation of crickets by notonectids increases the visibility
826 ANIMAL BEHAVIOUR, 75, 3

of the cricket to visually hunting vertebrates such as fish. McFarland, L. H., Mouritsen, K. N. & Poulin, R. 2003. From first to
In this case more rapid emergence would be advantageous; second and back to first intermediate host: the unusual transmis-
although worms can survive predation by vertebrates, it is sion route of Curtuteria australis (Digenea: Echinostomatidae).
fatal in many cases (Ponton et al. 2006a). Finally, because Journal of Parasitology, 89, 625e628.
the contortion performed by emerging worms (Fig. 1) Murdoch, W. W. & Scott, M. A. 1984. Stability and extinction of
laboratory populations of zooplankton preyed on by the back-
make them more visible to generalist predators that could
swimmer Notonecta. Ecology, 65, 1231e1248.
also consume both notonectid and cricket, we cannot fully
Murdoch, W. W., Scott, M. A. & Ebsworth, P. 1984. Effects of the
exclude the possibility that notonectids induce the rapid general predator Notonecta (Hemiptera) upon a freshwater com-
emergence of the worm to reduce their own predation munity. Journal of Animal Ecology, 53, 791e808.
risk. The proximate mechanisms that would sustain such Pace, M. L., Cole, J. J., Carpenter, S. R. & Kitchell, J. F. 1999.
a scenario remain to be determined; it is of interest that Trophic cascades revealed in diverse ecosystems. Trends in Ecology
the cuticle of hairworms has sensory capabilities & Evolution, 14, 483e488.
(Schmidt-Rhaesa 2004). Pampoulie, C., Lambert, A., Rosecchi, E., Crivelli, A. J.,
Bouchereau, J. L. & Morand, S. 2000. Host death: a necessary
condition for the transmission of Aphalloides coelomicola Dollfus,
Acknowledgments
Chabaud, and Golvan, 1957 (Digenea, Cryptogonimidae)? Journal
M.I.S. was supported by a postdoctoral grant from the of Parasitology, 86, 416e417.
Ministerio de Ciencia y Tecnologı́a (Spain). We thank the Parker, G. A., Chubb, J. C., Ball, M. A. & Roberts, G. N. 2003. Evo-
lution of complex life cycles in helminth parasites. Nature, 425,
thermal station of Avènes les Bains for their cooperation
480e484.
during the field study. T. Lefevre and C. Lebarbenchon
Ponton, F., Lebarbenchon, C., Lefèvre, T., Biron, D. G., Duneau,
helped with field work. We thank two anonymous referees
D., Hughest, D. & Thomas, F. 2006a. Parasite survives predation
and the editor, David Hosken, for their constructive on its host. Nature, 440, 756.
comments on the paper. Ponton, F., Lebarbenchon, C., Lefèvre, T., Thomas, F., Duneau,
D., Marché, L., Renault, L., Hughes, D. P. & Biron, D. G.
References 2006b. Hairworms anti-predator strategy: a study of causes and
consequences. Parasitology, 133, 631e638.
Arner, M., Koivosto, S., Norberg, J. & Kautsky, N. 1998. Trophic Poulin, R. 1998. Evolutionary Ecology of Parasites: From Individuals to
interactions in rockpool food webs: regulation of zooplankton Communities. London: Chapman & Hall.
and phytoplankton by Notonecta and Daphnia. Freshwater Biology, Price, P. W. 1980. Evolutionary Biology of Parasites. Princeton, New
39, 79e90. Jersey: Princeton University Press.
Blaustein, L. 1998. Influence of the predatory backswimmer, Noto- Read, A. F. 1994. The evolution of virulence. Trends in Microbiology,
necta maculata, on invertebrate community structure. Ecological 2, 73e76.
Entomology, 23, 246e252. Robert, F., Renaud, F., Mathieu, E. & Gabrion, C. 1988. Impor-
Brivio, M. F., de Eguileor, M., Grimaldi, A., Vigetti, D., Valvassori, tance of the paratenic host in the biology of Bothriocephalus
R. & Lanzavecchia, G. 2000. Structural and biochemical analysis gregarius (Cestoda, Pseudophyllidea), a parasite of the turbot.
of the parasite Gordius villoti (Nematomorpha, Gordiacea) cuti- International Journal of Parasitology, 18, 611e621.
cule. Tissue and Cell, 32, 366e376. Schmidt-Rhaesa, A. 1996. Zur Morphologie, Biologie und Phylogenie
Brodeur, J. & McNeil, J. N. 1994. Seasonal microhabitat selection by der Nematomorpha e Untersuchungen an Nectonema munidae und
an endoparasitoid through adaptive modification of host behav- Gordius aquaticus. Göttingen: Cuvillier Verlag.
iour. Science, 244, 226e228. Schmidt-Rhaesa, A. 2004. Ultrastructure of an integumental organ
Chesson, J. 1984. Effect of notonectids (Hemiptera: Notonectidae) with probable sensory function in Paragordius varius (Nematomor-
on mosquitoes (Diptera: Culicidae): predation or selective oviposi- pha). Acta Zoologica, 85, 15e19.
tion? Environmental Entomology, 13, 531e538. Siegel, S. & Castellan, N. J. 1988. Nonparametric Statistics for the
Choisy, M., Brown, S. P., Lafferty, K. D. & Thomas, F. 2003. Behavioral Sciences. 2nd edn. New York: McGraw-Hill.
Evolution of trophic transmission in parasites: why add intermedi- Sokal, R. R. & Rohlf, F. J. 1981. Biometry. 2nd edn. New York: W. H.
ate hosts? American Naturalist, 162, 172e181. Freeman.
Geddes, M. C. 1986. Understanding zooplankton communities in Swanson, C.J. 1970. An ultrastructural, biochemical, and functional
farm dams: the importance of predation. In: Limnology in Australia analysis of the body wall of Paragordius varius (Leidy), and Gordius
(Ed. by P. Dekker & W. D. Williams), pp. 387e401. Dordrecht: Junk. robustus (Leidy) (Aschelminthes, Nematomorpha, Gordioidea).
Haine, E. R. & Rigaud, T. 2005. Conflict between parasites with dif- Ph.D. thesis, University of Illinois, Urbana.
ferent transmission strategies infecting an amphipod host. Pro- Thomas, F., Brown, S. P., Sukhdeo, M. & Renaud, F. 2002a.
ceedings of the Royal Society of London, Series B, 272, 2505e2510. Understanding parasite strategies: a state-dependent approach?
Hanelt, B., Thomas, F. & Schmidt-Rhaesa, A. 2005. Biology of the Trends in Parasitology, 18, 387e390.
phylum Nematomorpha. Advances in Parasitology, 59, 243e305. Thomas, F., Fauchier, J. & Lafferty, K. D. 2002b. Conflict of interest
Lafferty, K. D. 1999. The evolution of trophic transmission. Parasi- between a nematode and a trematode in an amphipod host: test
tology Today, 15, 111e115. of the ‘sabotage’ hypothesis. Behavioral Ecology and Sociobiology,
Lafferty, K. D., Thomas, F. & Poulin, R. 2000. Evolution of host 51, 296e301.
phenotype manipulation by parasites and its consequences. In: Thomas, F., Schmidt-Rhaesa, A., Martin, G., Manu, C., Durand,
Evolutionary Biology of HosteParasite Relationships: Theory Meets P. & Renaud, F. 2002c. Do hairworms (Nematomorpha) manipu-
Reality (Ed. by R. Poulin, S. Morand & A. Skorping), pp. 117e late the water seeking behaviour of their terrestrial hosts? Journal of
127. Amsterdam: Elsevier Science. Evolutionary Biology, 15, 356e361.
Levri, E. P. 1998. The influence of non-host predators on parasite in- Welch, B. L. 1951. On the comparison of several mean values: an
duced behavioral changes in a freshwater snail. Oikos, 81, 531e537. alternative approach. Biometrika, 38, 330e336.