Differential influence of two acanthocephalan parasites on the antipredator behaviour of their common intermediate host

Abstract

Fish acanthocephalans can modify the antipredator behaviour of their intermediate hosts in response to cues from fish predators. However, it is still unclear whether such behavioural changes are adaptive, or are just the consequence of infection. We addressed this question through studying two acanthocephalans, Pomphorhynchus laevis and Polymorphus minutus, and their intermediate host, the amphipod Gammarus pulex. Pomphorhynchus laevis completes its cycle in a freshwater fish, whereas P. minutus exploits waterbirds as final hosts. We first assessed vulnerability of infected and uninfected gammarids to predation by bullheads, Cottus gobio. Pomphorhynchus laevis-infected gammarids were more susceptible to predation than uninfected ones when a refuge was available, whereas no selective predation on P. minutus-infected individuals was recorded, independently of refuge availability. We then quantified refuge use when a bullhead was present in an enclosure or when the enclosure was empty. Individuals of each group significantly increased refuge use in the presence of a bullhead. However, a larger proportion of P. laevis-infected gammarids remained out of the refuge in the presence of a predator, compared with uninfected controls, whereas no such effect was observed in P. minutus-infected ones. Finally, we assessed reaction to bullhead olfactory cues, using a Y-maze apparatus. Pomphorhynchus laevis-infected gammarids spent significantly more time in the predator-scented arm, whereas the reverse was observed in uninfected ones. Polymorphus minutus-infected individuals, however, did not differ from uninfected controls. We discuss our results in relation with the adaptiveness of host manipulation by parasites.

Full text

Page 1

Differential influence of two acanthocephalan parasites on the
antipredator behaviour of their common intermediate host
NICOLAS KALDONSKI, MARIE-JEANNE PERROT-MINNOT & FRANK CE´ZILLY
Equipe Ecologie-Evolutive, UMR CNRS 5561 Bioge ´osciences, Universite ´ de Bourgogne
(Received 22 December 2006; initial acceptance 12 February 2007;
final acceptance 19 February 2007; published online 3 October 2007; MS. number: 9225R1)
Fish acanthocephalans can modify the antipredator behaviour of their intermediate hosts in response to
cues from fish predators. However, it is still unclear whether such behavioural changes are adaptive, or
are just the consequence of infection. We addressed this question through studying two acanthocepha-
lans, Pomphorhynchus laevis and Polymorphus minutus, and their intermediate host, the amphipod Gamma-
rus pulex. Pomphorhynchus laevis completes its cycle in a freshwater fish, whereas P. minutus exploits
waterbirds as final hosts. We first assessed vulnerability of infected and uninfected gammarids to predation
by bullheads, Cottus gobio. Pomphorhynchus laevis-infected gammarids were more susceptible to predation
than uninfected ones when a refuge was available, whereas no selective predation on P. minutus-infected
individuals was recorded, independently of refuge availability. We then quantified refuge use when a bull-
head was present in an enclosure or when the enclosure was empty. Individuals of each group significantly
increased refuge use in the presence of a bullhead. However, a larger proportion of P. laevis-infected gam-
marids remained out of the refuge in the presence of a predator, compared with uninfected controls,
whereas no such effect was observed in P. minutus-infected ones. Finally, we assessed reaction to bullhead
olfactory cues, using a Y-maze apparatus. Pomphorhynchus laevis-infected gammarids spent significantly
more time in the predator-scented arm, whereas the reverse was observed in uninfected ones. Polymorphus
minutus-infected individuals, however, did not differ from uninfected controls. We discuss our results in
relation with the adaptiveness of host manipulation by parasites.
? 2007 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Keywords: Acanthocephalan; antipredator behaviour; Cottus gobio; host-manipulation; olfaction; Polymorphus minutus;
Pomphorynchus laevis; parasites
Several species of parasites critically rely on a phase of
trophic transmission from an intermediate host to a final
host to complete their complex life cycle (Parker et al.
2003). Any alteration in the phenotype of the intermedi-
ate host that would result in increased vulnerability to pre-
dation by final hosts is thus supposed to benefit the
parasite. Over the last 30 years, empirical evidence dem-
onstrating the ability of some parasites to manipulate
the phenotype of their intermediate hosts has accumu-
lated (Moore 2002; Thomas et al. 2005). Parasites
may change host appearance (Camp & Huizinga 1979;
Oetinger & Nickol 1981; Bakker et al. 1997; Fuller et al.
2003),microhabitatchoice(MacNeil etal. 2003;
Miura et al. 2006), behaviour (Bethel & Holmes 1973;
Ce ´zilly et al. 2000) and physiology (Plaistow et al. 2001;
Tain et al. 2006). However, only a few empirical studies
have convincingly demonstrated that changes in the phe-
notype of infected hosts, rather than a general decrease in
stamina, are indeed responsible for the increased vulnera-
bility of intermediate hosts to predation by appropriate fi-
nal hosts (Ce ´zilly & Perrot-Minnot 2005).
A particularly efficient way for trophically transmitted
parasites to achieve specificity in manipulation could be to
target the antipredator behaviour of their intermediate
hosts (Hechtel et al. 1993; Berdoy et al. 2000). Many prey
species have evolved adaptations to reduce predation risk,
such as morphological structures, chemical repellents,
crypsis and avoidance behaviours (Endler 1986; Sih 1987;
Dodson et al. 1994; Kats & Dill 1998). In the presence of
a specific predator, prey may alter their behaviour so that
they are more difficult to encounter, detect or capture
Correspondence: F. Ce ´zilly, Equipe Ecologie-Evolutive, UMR CNRS
5561 Bioge ´osciences, Universite ´ de Bourgogne, 6 Blvd Gabriel, 21000
Dijon, France (email: frank.cezilly@u-bourgogne.fr).
1311
0003e3472/07/$30.00/0
? 2007 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
ANIMAL BEHAVIOUR, 2007, 74, 1311e1317
doi:10.1016/j.anbehav.2007.02.027

Page 2

(Lima 1998). In the aquatic environment, in particular, ol-
factory assessment of predation risk can occur at distance
and allow prey to react to minimize the risk of encounter
withpredators(Peacor&Werner1997;Turner1997;Peckar-
sky & McIntosh 1998; Huhta et al. 2000; Wisenden 2000).
Aquatic prey organisms respond to predator-associated
chemical cues in a number of ways, most often by reducing
activity level and increasing the use of refuges (for a review,
see Kats & Dill 1998). In this context, any change induced
byparasiteonchemicallybasedpredatorrecognitionmight
be a fine-tuned strategy to reach a final host.
Recently, two independent studies (Baldauf et al. 2007;
Perrot-Minnot et al. 2007) have provided direct evidence
for manipulation of the antipredator behaviour of the
crustacean amphipod Gammarus pulex by acanthocepha-
lan parasites. Baldauf et al. (2007) first showed that gam-
marids infected with the fish acanthocephalan parasite
Pomphorhynchus laevis were attracted to the smell of
a perch Perca fluviatilis, whereas uninfected individuals
were repulsed by the same odour, independently of visual
cues. Perrot-Minnot et al. (2007) went further by showing
that infection of G. pulex with the congeneric parasite
Pomphorhynchus tereticollis resulted in both increased vul-
nerability to predation by bullheads, Cottus gobio, in the
field, and reversed antipredator behaviour (as measured
by use of refuges and reaction to chemical cues from the
predator). However, whether such an alteration of host an-
tipredator behaviour is peculiar to fish parasites, or is just
a by-product of infection with acanthocephalans remains
so far unclear.
We addressed this question in a series of experiments
using two species of acanthocephalan parasites and their
common intermediate host, the crustacean amphipod
G.pulex.InBurgundy(easternFrance),bothP.laevisandPol-
ymorphus minutus, a bird acanthocephalan, exploit the am-
phipod G. pulex as intermediate host. Each parasite is
knowntoinducespecificbehaviouralmodificationsintheir
intermediate host, with P. laevis affecting mainly phototac-
tism, whereas P. minutus reverses geotactism (Ce ´zilly et al.
2000). Pomphorhynchus laevis is known to establish in sev-
eral fish species, including the bullhead (Rumpus 1975;
Kennedy 1999). By contrast, being predated by a bullhead
represents a dead-end for P. minutus. We first compared in
aseriesofcontrolledexperimentsthevulnerabilityof P.lae-
vis-infected gammarids and P.minutus-infected ones to pre-
dation by bullhead, which is both an appropriate final host
fortheparasiteandoneofthemainpredatorsofamphipods
in freshwater ecosystems (Andersson et al. 1986; Andersen
et al. 1993; MacNeil et al. 1999). Second, as G. pulex is
knowntoshowmarkedantipredatorbehaviourinresponse
to predators’ cues (Williams & Moore 1985; Holomuzki &
Hoyle 1990; Wudkevich et al. 1997; Dahl et al. 1998; Bal-
dauf et al. 2007; Perrot-Minnot et al. 2007), we assessed
the influence of each parasite on the antipredator behav-
iouroftheircommonintermediatehostusingtwodifferent
methods.
METHODS
Uninfected, P. laevis- and P. minutus-infected G. pulex were
collected from July to September in 2004 and 2005 in the
river Ouche at Dijon (Burgundy, eastern France) using the
kick sampling method (Hynes 1954). Infected gammarids
can be readily distinguished from uninfected ones since
parasites can be seen through the host’s translucid cuticle
as a yellow-orange dot for P. laevis and as an orange-red
dot for P. minutus. Adult males were selected for the exper-
iments, within sizes ranging from 11 to 15 mm long. They
were maintained in several separate tanks, well aerated,
and filled with dechlorinated tap water. Gammarids were
fed with leaves and fish food, for at least 1 week of accli-
matization to room and water condition before experi-
ments. Bullheads, with a size ranging from 75 to 85 mm
were caught with a dipnet, kept in the room for 1 month
before the experiments and fed with gammarids. They
were starved for 24 h prior to the experiments. Room tem-
perature was kept constant at 15?C with a light:dark cycle
of 12:12 h.
Selectivity of Bullheads towards Infected Preys
The selectivity of bullheads towards gammarids preys
infected with either fish or bird acanthocephalans was
assessed during predation tests performed in a controlled
lab-microcosm. Predation tests were carried out in aquaria
measuring 32? 20 ?20 cm, filled with 7-litre of dechlori-
nated and oxygenated tap water. A simple habitat set-up
consisted of washed river sand substrate, whereas a com-
plex habitat set-up was composed of a piece of air brick
(21.5 ?10 ? 5 cm) placed on the sand substrate plus two
Apiacea plants; both air brick and plants could be used
by gammarids as refuges. Aquaria were screened with
brown plastic on four sides and at the bottom to avoid
any lateral disturbance. Overhead solar spectra fluorescent
tubes (JBL Solar Natur, 25 Wand 9000 K) provided an illu-
mination of 700 lux. This light intensity is equivalent to
dawn or dusk light conditions, under which bullheads
show peak predatory activity in the field (Andreasson
1969). Predators were offered a 30:70 infected:uninfected
prey ratio. To that end, we introduced 42 uninfected gam-
marids and either 18 P. laevis- or 18 P. minutus-infected
gammarids in the aquarium 15 min before introducing
one individual bullhead. Each individual bullhead was
used for only one test. Following preliminary experi-
ments, the duration of each trial was set at 90 min. At
the end of each trial, the bullhead and the remaining
prey were removed, surviving gammarids counted, and
we recorded the percentage of surviving infected prey.
All material was then carefully rinsed with tap water. Bull-
heads were released in their habitat of origin at the end of
the experiments.
Following Seppa ¨la ¨ et al. (2004), differential of preda-
tion between the two prey types was assessed using the
probabilistic approach described in Manly (1974) which
allows for the depletion of prey during the course of
the trial, and thus for changes in the proportions of avail-
able prey classes as prey are eaten. Only experiments
where more than six preys had been eaten were kept
for analysis (see Manly 1974). To determine predator se-
lectivity for infected gammarids, we calculated Manly’s
alpha (ai) for variable prey population using the equation
ANIMAL BEHAVIOUR, 74, 5
1312

Page 3

ai¼logepi
Pm
j¼1pj
where aiis the Manly’s alpha (preference index) for prey
type I, piand pjare the proportion of prey i or j remaining
at the end of the trial, and m is the number of prey types.
Manly’s selectivity index ranges from 0 (when only unin-
fected prey are eaten) to 1 (when only infected prey are
eaten), with a value of 0.5 for absence of preference.
Observed values of aiwere compared with a situation of
equal vulnerability (ai¼ 0.5) using a two-tailed nonpara-
metric sign test, since the assumptions of parametric tests
were not met.
Influence of Infection on Refuge Use
Bullhead are known to emit both acoustic and olfactory
signals. Acoustic ‘knocking’ sounds are produced by the
males to attract a female (Ladich 1989) and have a territo-
rial function in the sexual season, but are rare compared
with chemical cues associated with mucus and urine pro-
duction. Gammarids are known to respond to these stim-
uli, but in greater proportion to the latter (Andersson et al.
1986). The typical reaction is a hiding behaviour under
stones or any refuge around, and diminishing activity
(Wudkevich et al. 1997).
Experiments were performed in the same experimental
set-up than previously, except that this time only the
piece of air brick was made available as refuge. For each
test, 10 individual gammarids of the same type (un-
infected, P. laevis or P. minutus infected) were placed in
the aquarium. The number of gammarids outside the ref-
uge was then recorded after 90 min, either in a predator-
free situation, or when a single bullhead had been placed
in a microperforated enclosure inside the aquarium. Since
data were not normally distributed, we used a Kruskale
Wallis test, followed by post hoc tests for comparison
with a control (Siegel & Castellan 1988), to assess the
effect of each parasitic infection on the use of refuge,
in the presence or in the absence of a predator. Manne
Whitney U tests were subsequently used to compare the
use of refuge in the absence versus in the presence of
a predator within each group (uninfected, P. laevis or
P. minutus infected).
Response to Fish Odour
Water is an excellent medium for the solubilization and
dispersal of chemical signals, and the antipredator re-
sponse of both vertebrate and invertebrate prey to pred-
ator chemical cue is well known to confer a fitness benefit
(Wisenden et al. 1997). Response to fish odour is thus
a critical component of susceptibility to predation. Exper-
iments were carried out in a Y-maze olfactometer made of
clear glass and fitted to a peristaltic pump, as described in
Perrot-Minnot et al. (2007). Experiments were run be-
tween 0900 and 1700 hours. All observations were done
behind a curtain to minimize interference with the ob-
server. Bullhead-conditioned water was obtained through
placing a single fish for 24 h into a clean aquarium with
250 ml of dechlorinated tap water per fish centimetre. To-
tal length of bullheads ranged from 80 to 90 mm. At the
end of each trial, a small quantity of tap water was added
to the aquarium to re-fill it from the lost volume. Oxygen-
ated and dechlorinated tap water was used as control
water.
Each experimental trial lasted 11 min. After a preflow
time of 1 min, a single specimen of G. pulex was intro-
duced in the starting area in the 4 downstream cm of
the Y-maze. Following a 5-min acclimatization period,
scented water was connected to the olfactometer, the
door was simultaneously removed, releasing the gam-
marid that was expected to swim against the mild current
of the inflow due to rheotaxis. Gammarid behaviour was
directly observed for 5 min using a Psion Workabout. Af-
ter each run, the test gammarid was removed and the ol-
factometer and connecting tubes were carefully rinsed
with tap water. The inflow arm of the treatment water
was switched after every five replicates to control for
any side preference, at which moment the olfactometer
was cleaned with ethanol and tap water. We used a differ-
ent gammarid for each replicate and a different bullhead
every 30 gammarids (such that 10 gammarids of each
group were tested with a single bullhead). Trials where
gammarids showed no movement for 2 min following
door-lifting or spent less than 20% of total time in the
two inflow arms were not considered in the analysis. For
each trial, the proportion of total time spent in the treated
arm was calculated. Arcsine square root-transformed pro-
portions met normality (ShapiroeWilk test), thus allow-
ing the use of parametric tests. We calculated 95%
confidence intervals around the mean proportion of
time spent in the scented arm. Comparison between un-
infected, P. laevis- and P. minutus-infected gammarids
was made using an analysis of variance (ANOVA; Sokal
& Rohlf 1995).
All statistical tests were performed with JMP statistical
software v. 5.0 (SAS Institute Inc., Cary, NC, U.S.A.).
RESULTS
Selectivity of Bullhead towards Infected Preys
A total of 823 gammarids were eaten by 62 different
bullheads in the microcosm experiments. Figure 1 shows
variation in specificity in predation towards infected
prey in relation to parasite species and presence/absence
of refuges, as assessed from Manly’s a preference index. Se-
lective predation on parasitized prey was significant in the
case of P. laevis infection only when a refuge was available
(Sign test, without refuge: P ¼ 0.4545; with refuge: P ¼
0.0180). By contrast, P. minutus-infected prey did not dif-
fer from uninfected ones in vulnerability to predation by
bullheads, both in the absence (P ¼ 1.000) and presence
of a refuge (P ¼ 0.6072). Vulnerability of P. laevis-infected
gammarids was higher than that of P. minutus-infected
ones only in the presence of refuge (ManneWhitney U
test: with refuge: U ¼ 25.5, NP. laevis¼ NP. minutus¼ 15, P ¼
0.0003; without refuge: U ¼ 68.5, NP. laevis¼ NP. minutus¼ 16,
P ¼ 0.07).
KALDONSKI ET AL.: MANIPULATION OF HOST ANTIPREDATOR BEHAVIOUR
1313

Page 4

Influence of Infection on Refuge Use
Infection status had no significant influence on refuge
use in the absence of a predator (KruskaleWallis test:
H2¼ 3.43, P ¼ 0.18; Fig. 2). The presence of a predator
induced a decrease in the number of gammarids out of
the refuge (ManneWhitney U test: uninfected: U ¼ 0,
Nwithout bullhead¼ Nwith bullhead¼ 8, P < 0.001; P. laevis
infected:
U ¼ 12,
P ¼ 0.033; P. minutus infected: U ¼ 5, Nwithout bullhead¼
Nwith bullhead¼ 8, P ¼ 0.0037), but not to the same extent.
Indeed, refuge use in the presence of a predator differed
Nwithout bullhead¼ Nwithbullhead¼ 8,
between the three groups (KruskaleWallis test: H2¼ 8.25,
P ¼ 0.016). This difference was actually due to a higher
number of P. laevis-infected individuals out of the refuge
compared with uninfected (comparison with uninfected
control:q ¼ 2.891,P<0.01),whereasP.minutus-infectedin-
dividuals and uninfected did not differ (q ¼ 0.535, P >
0.05).
Response to Fish Odour
Response to fish odour in a Y-maze olfactometer differed
significantly between uninfected and infected gammarids
(ANOVA, F2.117¼ 9.58, P < 0.0001; Fig. 3). Percentage of
time spent by uninfected gammarids was significantly
lower in the arm with scented water than expected under
the assumption of random choice (time spent in the
scented arm over the total time spent in the inflow
arms: mean and 95% confidence interval on back-
transformed data ¼ 39.2% (32.3e46.4%), N ¼ 41; Fig. 3).
By contrast, P. laevis-infected gammarids spent signifi-
cantly more time in the arm with bullhead-conditioned
water (mean and 95% confidence interval on back-trans-
formed data ¼ 60.6% (53.5e67.4%), N ¼ 43; Fig. 3). The
difference between the two groups was significant (least-
square contrast post hoc comparison: F1.115¼ 17.00,
P < 0.0001). Polymorphus minutus-infected individuals
spent on average 43.1% (95% confidence interval on
back-transformed data ¼ 35.2e51.2%, N ¼ 34) of their
time in the predator-scented arm, and did not differ
from uninfected individuals (least-square contrast post
hoc comparison: F1.115¼ 0.48, P ¼ 0.49). By contrast,
P. minutus-infected individuals spent significantly less
time than P. laevis-infected ones within the scented
arm(least-squarecontrast
F1.115¼ 10.34, P ¼ 0.002).
posthoccomparison:
DISCUSSION
Overall, our results, together with those of previous
studies (Hechtel et al. 1993; Bakker et al. 1997; Wellnitz
0
1
2
3
4
5
6
7
8
Number of individuals out of the refuge
Uninfected
P. minutus-infected
P. laevis-infected
9
10
******
*
Figure 2. Box plot indicating refuge use by uninfected, Pomphorhyn-
chus laevis-infected, and Polymorphus minutus-infected Gammarus
pulex in a microcosm setting. Light bars: without predator; dark
bars: with predator (N ¼ 8 replicates for each group). Groups con-
nected by above lines are significantly different (see text for
methods). *P < 0.05; **P < 0.01; ***P < 0.001.
50
41
Uninfected
Proportion of time spent
in the scented arm
P. laevis-
infected
43
P. minutus-
infected
34
25
0
75
Figure 3. Time spent by uninfected, Pomphorhynchus laevis-infected,
and Polymorphus minutus-infected gammarids in the predator-
scented arm in the Y-maze experiment. Dashed line represents
random visit of control and stimulus arms. Means are given ?95%
confidence interval. Numbers in the bars are sample size.
Preference index (Manly’s alpha)
0
0.25
0.5
0.75
1
Uninfected versus
P. laevis-infected
Uninfected versus
P. minutus-infected
15
16
15
16
**
Figure 1. Box plot indicating median, interquartile range and range
for differential predation on infected versus uninfected Gammarus
pulex as assessed from Manly’s a preference index (light bars: no ref-
uge available; dark bars: refuge available). Values above the dashed
line indicate overconsumption of infected prey. Numbers above the
bars are sample size. *P < 0.05; **P < 0.01; ***P < 0.001.
ANIMAL BEHAVIOUR, 74, 5
1314

Page 5

et al. 2003; Baldauf et al. 2007; Perrot-Minnot et al. 2007),
show that the modification of the antipredator behaviour
of crustacean intermediate hosts by fish acanthocephalans
may contribute to increase in the trophic transmission of
parasites to fish final hosts. In addition, the present study
indicates that, in the same conditions, gammarids in-
fected by a bird acanthocephalan do not show increased
vulnerability to a fish predator. Therefore, taken together,
our results suggest that the alteration of the antipredator
behaviour of intermediate hosts might ensure trophic
transmission to appropriate final hosts, thus providing ev-
idence for some degree of specificity in parasitic manipu-
lation. Such specificity in manipulation supports the
hypothesis of adaptive manipulation according to one of
the key criteria of fitness gain proposed by Poulin (1995).
Compared with previous studies on the influence of
parasites on hosts’ reaction to predators (i.e. Hechtel et al.
1993; Berdoy et al. 2000; Wellnitz et al. 2003; Baldauf et al.
2007; but see Perrot-Minnot et al. 2007), the present study
combined detailed analysis of antipredator behaviour
with predation tests in microcosms. One of the major crit-
icisms of predation experiments made in microcosms is
the use of unrealistic ratio of parasitized prey, hence fa-
vouring their overconsumption (Nickol 2005; Thomas
et al. 2005). Indeed, in most previous studies (e.g. Bethel
& Holmes 1977; Camp & Huizinga 1979; Bakker et al.
1997) infected and uninfected preys were offered in equal
proportions, such that infected preys were overrepre-
sented compared with their natural prevalence. By con-
trast, in the present study, the proportions of infected
and uninfected prey (30:70) were closer to field prevalence
of infected gammarids in the drift (L. Bollache et al., un-
published data).
However, the present study relied on naturally infected
prey items instead of experimentally infected individuals.
It might therefore be argued that gammarids behaved
differently for reasons unrelated to parasitism, and that
the odd behaviour was actually the cause rather than the
consequence of infection with acanthocephalans. Such an
explanation is unlikely for two different reasons. The first
one is the fact that the results are different for the two
species of parasites. The second one corresponds to the
fact that changes in the behaviour of intermediate hosts
infected with acanthocephalans happen only after the
parasite has reached the cystacanth stage (i.e. the stage at
which the parasite becomes infective to the definitive
host), whereas intermediate hosts infected with immature
parasites show no change in behaviour (F. Ce ´zilly & M.-J.
Perrot-Minnot, unpublished results; see also Bethel &
Holmes 1974).
One particular interest of our results is the strong
congruence between the results of predation experiments
and those related to the use of refuge and response to
predator chemical cues by uninfected and infected prey.
No difference was observed in the use of refuge between
uninfected gammarids, P. laevis-infected, and P. minutus-
infected ones in the absence of a predator. A difference
in the use of refuge was only observed in the presence of
a predator. Although P. laevis-infected individuals in-
creased their use of refuges in the presence of a predator,
they did so with less intensity than uninfected and
P. minutus-infected ones. Accordingly, differential preda-
tion on P. laevis-infected individuals was significant only
when refuges were made available. The importance of re-
action to predator chemical cues is further demonstrated
by the experiments using the Y-maze oflactometer. Con-
sidered alone, the lesser use of refuge by P. laevis-infected
gammarids in the presence of predator cues, compared
with P. minutus-infected and uninfected ones, is open to
alternative explanations. Infection with P. laevis may ren-
der gammarids less able to detect fish odour, or infected
gammarids may remain able to detect it but not respond
accordingly (see Wellnitz et al. 2003). Here, however, the
use of a distant system of chemical predator perception al-
lowed us to separate chemical cues from other factors. The
significant attraction of P. laevis-infected individuals to-
wards chemical cues from a bullhead clearly shows that
infected individuals are indeed able to perceive chemical
stimuli, but differ from uninfected ones in their response
to predator scent (see also Perrot-Minnot et al. 2007).
Infection by acanthocephalans can result in various
alterations of the host phenotype (Bakker et al. 1997;
Ce ´zilly et al. 2000; Baldauf et al. 2007; this study) that
may all contribute to increased vulnerability to predation
by final hosts. It has been previously shown that P. laevis
reverses photokinesis in G. pulex, whereas P. minutus
strongly alters geotaxis (Ce ´zilly et al. 2000). Both alter-
ations of host behaviour may contribute to increase in
appropriate transmission to appropriate final hosts.
However, the effect of manipulating parasites on host
responses to light and gravity may not be sufficient to
explain differential susceptibility in the field (Bakker
et al. 1997; Levri & Fisher 2000). Other variables, such as
cues from predators and increased conspicuousness of in-
fected hosts (Camp & Huizinga 1979; Oetinger & Nickol
1981; Fuller et al. 2003), may be important and interact
with physical cues to determine the behaviour of infected
hosts. In this respect, the absence of differential predation
between P. laevis-infected individuals and uninfected ones
in the absence of refuge seems at first sight to contradict
the results of Bakker et al. (1997) who found that the
change in appearance of infected individuals (due to the
orange cystacanth being visible through the host’s trans-
lucid cuticle) contributed directly to their increased vul-
nerability to a fish predator. However, the two studies
used different predator species. Bakker et al. (1997) studied
predation by sticklebacks, Gasterosteus aculeatus, which are
active predators and in which females show a marked
sensory bias for red-orange coloration in relation to male
nuptial ornaments (Smith et al. 2004). By contrast, the
bullhead is an ambush predator with no known sensory
bias in colour perception. The precise importance of
parasite coloration versus host-modified behaviour in
increasing vulnerability to predators deserves further con-
sideration. Cystacanths of P. minutus are bright orange and
P. minutus-infected gammarids appear, at least to the hu-
man eye, even more conspicuous that P. laevis-infected
ones. If cystacanth coloration does increase vulnerability
to fish predators, then the underconsumption of P. minu-
tus-infected preys by bullheads observed in the present
study suggests that parasitic manipulation of G. pulex by
P. minutus may eventually compensate for increased
KALDONSKI ET AL.: MANIPULATION OF HOST ANTIPREDATOR BEHAVIOUR
1315