Social wasps desert the colony and aggregate outside if parasitized: parasite manipulation?

Abstract

Infection of the paper wasp, Polistes dominulus (Christ), by the strepsipteran parasite Xenos vesparum Rossi results in a dramatic behavioral change, which culminates in colony desertion and the formation of extranidal aggregations, in which up to 98% of occupants are parasitized females. Aggregations formed on prominent vegetation, traditional lek-sites of Polistes males, and on buildings, which were later adopted as hibernating sites by future queens. First discovered by W.D. Hamilton, these aberrant aggregations are an overlooked phenomenon of the behavioral ecology of this intensively studied wasp. For 3 months in the summer of 2000, during the peak of colony development, we sampled 91 extranidal aggregations from seven areas, numbering 1322 wasps. These wasps were parasitized by both sexes of X. vesparum, but males were more frequent from July until mid-August, during the mating season of the parasite. Aggregations were present for days at the same sites (in one case a leaf was occupied for 36 consecutive days) and were characterized by extreme inactivity. After artificial infection, parasitized "workers" deserted the nest 1 week after emergence from their cell and before the extrusion of the parasite through the host cuticle. Infected individuals did not work, were more inactive, and did not receive more aggression than did controls. We suggest that early nest desertion and subsequent aggregations by parasitized nominal workers and "future queens" is adaptive manipulation of host behavior by the parasite to promote the completion of its life cycle. Copyright 2004.

Full text

Social wasps desert the colony and aggregate
outside if parasitized: parasite manipulation?
David P. Hughes,
a
Jeyaraney Kathirithamby,
a
Stefano Turillazzi,
b
and Laura Beani
b
a
Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK,
and
b
Dipartimento di Biologia Animale e Genetica, Universita` di Firenze,
Via Romana 17, 50125 Florence, Italy
Infection of the paper wasp, Polistes dominulus (Christ), by the strepsipteran parasite Xenos vesparum Rossi results in a dramatic
behavioral change, which culminates in colony desertion and the formation of extranidal aggregations, in which up to 98% of
occupants are parasitized females. Aggregations formed on prominent vegetation, traditional lek-sites of Polistes males, and on
buildings, which were later adopted as hibernating sites by future queens. First discovered by W.D. Hamilton, these aberrant
aggregations are an overlooked phenomenon of the behavioral ecology of this intensively studied wasp. For 3 months in the
summer of 2000, during the peak of colony development, we sampled 91 extranidal aggregations from seven areas, numbering
1322 wasps. These wasps were parasitized by both sexes of X. vesparum, but males were more frequent from July until mid-August,
during the mating season of the parasite. Aggregations were present for days at the same sites (in one case a leaf was occupied for
36 consecutive days) and were characterized by extreme inactivity. After artificial infection, parasitized ‘‘workers’’ deserted the
nest 1 week after emergence from their cell and before the extrusion of the parasite through the host cuticle. Infected individuals
did not work, were more inactive, and did not receive more aggression than did controls. We suggest that early nest desertion and
subsequent aggregations by parasitized nominal workers and ‘‘future queens’’ is adaptive manipulation of host behavior by the
parasite to promote the completion of its life cycle. Key words: aggregation, behavioral manipulation, nest desertion, Polistes,
Strepsiptera. [Behav Ecol 15:1037–1043 (2004)]
Primitively eusocial wasps of the genus Polistes have played
a pivotal role as ‘‘hypothesis generating model organisms’’
(Turillazzi and West-Eberhard, 1996) in sociobiology (Pardi,
1948) and kin selection theories (Hamilton, 1964). Their
annual, unenveloped, small nests, founded by one or more
overwintered females, are relatively easy to study. This, in
addition to their widespread occurrence in temperate zones,
has resulted in a voluminous literature regarding their
behavior (Reeve, 1991). One species in particular, Polistes
dominulus (Christ), has been claimed to the ‘‘most well
studied social wasp’’ (Queller et al., 2000). Therefore, the
discovery by W.D. Hamilton of previously unknown extranidal
aggregations of P. dominulus consisting of females parasitized
by the strepsipteran Xenos vesparum (dissected by J. Kathir-
ithamby) at a time when colonies were fully active (August
1998, Tuscany, Italy, see Hughes, 2002), was novel and
unexpected. Why should infected individuals abandon their
nest and aggregate outside when the colony is a ‘‘fortress’’ for
both parasites and wasps (Schmid-Hempel, 1998)? Extranidal
aggregations, which consist of healthy future queens, are
common only at the end of the reproductive phase (Pardi,
1942; West-Eberhard, 1969) and during diapause (Hunt et al.,
1999; Turillazzi, 1980). Hamilton’s aberrant, early aggrega-
tions of females could be an example of parasite-induced
change in host behavior: ‘‘After all, if they can so completely
castrate the wasps without killing them, an ability obviously
evolved in an association through an immense period of time,
it will not be surprising to find that they also manipulate the
behavior’’ (Hamilton WD, e-mail to Laura Beani, 3-09-1998,
11:35 hrs).
The order Strepsiptera are obligate endoparasitic insects
that are known to parasitize seven insect orders, including
solitary and social Hymenoptera (Kathirithamby, 1989). They
are ‘‘parasitic castrators’’ (Baudoin, 1975), because they
induce sterility in their hosts (Strambi and Strambi, 1973).
These macroparasites exhibit extreme sexual dimorphism
(Figure 1a,b): the short-lived (usually less than 5 h) winged
adult males and the first instar larvae are the only free-living
stages, whereas the neotenic larviform adult females are
permanently parasitic. Infection by X. vesparum (which is
termed stylopization after the family Stylopidae that infects
wasps and bees) begins with the entry of the first instar larva
into a P. dominulus larvae (all host larval stages are susceptible).
After successive endoparasitic stages and host pupation, the
last instar X. vesparum extrudes its anterior region through the
intersegmental membranes of the adult host. The male forms
a cephalotheca and pupates, and the female forms a cephalo-
thorax and becomes a neotenic nonpupating adult. Adult
males emerge from their puparia and fly off to inseminate
a female through the brood canal opening in her cephalo-
thorax. Females are viviparous and embryonic development
occurs within the hemocoel (Kathirithamby, 2000), and the
first instar larvae emerge via the brood canal opening.
Emergence occurs after a stylopized wasp (containing a gravid
female X. vesparum) alights on flowers, and subsequent
transport of first instars to the nest is phoretic, via a foraging
wasp. However, stylopized wasps might land on nests and first
instars emerge here to find new hosts (Hughes et al., 2003).
Only adult female X. vesparum overwinter (with their host);
males die soon after fertilizing the females.
In Tuscany, Italy, fertilized P. dominulus emerge from
diapause in March and begin nest construction, either singly
(haplometrosis) or as a group (pleometrosis), with the first
workers emerging in May/June and the first sexuals emerging
in July/August (Pardi, 1942; Reeve, 1991). Colony decline
occurs in September. Nests are very common on buildings, for
Address correspondence to D. P. Hughes, who is now at De-
partment of Biology, University of Oulu, Oulu 90014, Finland. E-mail:
david.hughes@oulu.fi.
Received 8 September 2003; revised 24 March 2004; accepted 2
April 2004.
Behavioral Ecology Vol. 15 No. 6: 1037–1043
doi:10.1093/beheco/arh111
Advance Access publication on July 7, 2004
Behavioral Ecology vol. 15 no. 6 International Society for Behavioral Ecology 2004; all rights reserved.
by guest on June 3, 2013http://beheco.oxfordjournals.org/Downloaded from

example, under roof tiles and eaves. Mating occurs at lek-sites
in August/September (large trees and poles around houses
where nests are constructed; Beani, 1996), and females over-
winter in crevices of buildings and under roof tiles. Parasitism
of P. dominulus by X. vesparum was common in Tuscany: 58% of
nests were infected; in these nests 28% of brood, but none of
adults, were stylopized (Hughes et al., 2003).
The ‘‘intimate interaction’’ (Combes, 2001) between host
and parasite is particularly long in Xenos/Polistes systems as
infection begins at the immature host stage (larval wasp) and
proceeds into adulthood. Female Xenos remain associated with
their hosts throughout adulthood and, where the female
overwinters, the association can last up to 1 year. Nevertheless,
previous behavioral records are scant and limited to lethargy by
infected workers (Fitzgerald, 1938; Hubbard, 1892), commu-
nal overwintering by healthy and stylopized females (Turillazzi,
1980), and the nonparticipation of infected ‘‘queens’’ in
colony life (Pardi, 1942). However, two reviews of colony
symbionts do not mention them (Nelson, 1968; Yamane,
1996). In the present study, we document the occurrence of
extranidal aggregations of wasps at seven sites in Tuscany,
including Hamilton’s first record area, throughout the season.
We also document the behavior of individuals within these
aggregations through focal animal observations and mark-
recapture techniques. Importantly, we perform controlled
infections of workers under laboratory conditions to de-
termine when they leave the nest and if desertion is preceded
by aggression from nest mates, or if stylopized individuals
contribute to own indirect fitness through work performance.
METHODS
Data collection in the field: single stylopized wasps,
aggregations, and nests
We checked seven field sites within 15 km of Florence, Italy
(43459N, 11189E), from 8 June–23 August 2000 for the
presence of extranidal wasps. For one site (area 1, close to
Florence airport) we sampled aggregations until 23 October,
in order to document the changes in aggregations over the
season. First, we collected single stylopized wasps on
vegetation or in flight (n¼21). After the 16 July, we sampled
‘‘aggregations,’’ here defined as four or more wasps in
contact, from vegetation surrounding buildings with a high
density of nests. We placed an insect sweep net over each
aggregation, and because of the extreme inactivity of the
gathered wasps, escapes were uncommon. Sampling of
aggregations was divided by week, with 16–23 July being week
1 and 7–15 October being week 12. We noted the location,
size, and parasite prevalence (the proportion of infected
individuals) of each aggregation. Collections occurred from
0800–2000 h, plus seven nightly checks (0200–0700 h). Wasps
that were evidently parasitized were checked for sex of both
wasps and parasites and released, whereas a few (n¼27) that
did not appear evidently parasitized, that is, with extruded
cephalothorax and/or cephalothecae, were killed and dis-
sected to identify possible endoparasitic stages. This occurred
until 23 August; after this date, at least one parasite per host
already extruded through the host cuticle, and we were able
to reliably estimate prevalence without any dissection.
Behavior in aggregations was recorded by using tripod-
mounted Sony TR425 Hi 8-mm camcorder positioned approx-
imately 2 m from the aggregation (area 1, weeks 2 and 3). We
observed the behavior of 32 female focal wasps already within
three aggregations and 12 individuals after they joined an
aggregation (2 min/wasps). There was no difference in activity
between the three aggregations (ANOVA on arcsine trans-
formed data, F
2,31
¼0.004, p¼.99) so these data were pooled.
To record the spatial behavior of stylopized wasps in area 1,
we marked 402 stylopized wasps with enamel paint from
aggregations on two mulberry trees and checked the same
sites twice per week from 16 July–31 August. For wasps marked
on the first five capture dates (up to 1 August, five cohorts),
we constructed mark-recapture histories by using the program
MARK (Cormack-Jolly-Seber open-population capture-recap-
ture methods; reduced model of constant recapture proba-
bility; Lebreton et al., 1992). The use of individual covariates,
such as parasite age, could not be included in the model
without host dissection.
The presence on the nest of either parasitized adults or
marked individuals from aggregations was estimated by the
collection and dissection of adults from 21 nests over three
dates (10 and 17 July and 3 August, n¼7 colonies each date).
A further 12 nests were collected from 3–8 August and
examined for the presence of marked wasps only. All nests
were taken under roof tiles of the main building in area 1
early in the morning, before foragers had left the colony.
Laboratory behavioral recording of infected workers
In 2001 and 2002 P. dominulus nests, founded by queens from
our study area, were maintained in the laboratory under 14-h
light/10-h dark and 27C–29C standard conditions. The first
instar strepsipteran larvae, which are the infective stage, were
collected as they emerged live from adult female X. vesparum
parasitic within stylopized wasps that had overwintered (from
the same area). Under a dissecting microscope, a single X.
vesparum first instar larva was transferred by means of a single
hair to the body of a larval wasp (the nest was cooled to 4Cso
that the adult wasps could be removed first). The cell was
marked with paint, and similar-sized larva were sham treated.
Three nests in 2001 (nine infected and nine control workers
in total; three each per nest) and eight nests in 2002 (12
infected and 14 control in total; between two and four per
nest) were successfully infected. All nests had approximately
40 cells and less than eight workers at the time of infection,
Figure 1
(a) A line diagram of an adult female and adult male X. vesparum.
The double-pointed arrow details the female cephalothorax,
which extrudes through the host. The scale bar ¼0.8 mm
(original drawing by Jeyaraney Kathirithamby, copyrighted, see
www.tolweb.org/tree/). (b) A stylopized female P. dominulus.
The arrow shows the more evident cephalotheca of the
X. vesparum male puparium, and the more cryptic female
cephalothorax is encircled.
1038 Behavioral Ecology Vol. 15 No. 6
by guest on June 3, 2013http://beheco.oxfordjournals.org/Downloaded from

that is, early ‘‘worker phase’’ (Reeve, 1991). Low replication
per nest was owing to the difficulty of infecting brood, so we
could not assess the colony effect.
To imitate a natural situation, infected nests were placed in
small nest-box (simulating natural nesting inside cavities,
volume ¼500 cm
3
) and connected to large cages in which
food, water, and paper for cell construction was placed
(more than 12,000 cm
3
). In 2001 only the position of evi-
dently parasitized and control workers was recorded three
times/day (0730, 1130, 2330 h) for 3 days. In 2002 the behavior
(time inactive; time off the nest; cell checking, a reliable
indicator of work on the nest; and aggression received) of
infected and control workers was recorded for the first 10
days from emergence (10 min/day per wasp; total observation
time ¼43 h 40 min). Observations were blind to the state of
parasitism and conducted daily between 1400–1700 h.
We used the program MARK (www.cnr.colostate.edu/
;gwhite/mark/mark.htm) for mark-recapture data analysis
and SPSS for all other analyses. All tests are two tailed, and
means are presented 6SE. Where indicated, we performed
transformation to approximately normalize the data. Logistic
regressions were performed with Enter method.
RESULTS
Occurrence and composition of aggregations
A total of 91 aggregations (n¼1322 wasps, 99.99%
P. dominulus), were checked from July–October: 97% of occu-
pants were female, and henceforth only P. dominulus females
are discussed. The first aggregation was observed on 16 July,
the last on 10 October 2000. Before this, from 8 June–15 July,
17 stylopized were captured singly and inactive on the
underside of leaves (no uninfected wasps were found here).
Two individuals were captured before 0700 h, that is, before
foraging begins, indicating they had spent the night away
from the nest. A further two individuals were captured flying
around leaves, and again uninfected wasps were not found at
this locality. All 19 wasps were presumably nominal workers;
they had perfect wing condition and their tergites were not
discolored as those of overwintered females are (Reeve, 1991).
They contained five female adult X. vesparum that were not
releasing first instar larvae (unlike overwintered ones which
were observed) and 17 males at different stages of de-
velopment. Adult free-living male X. vesparum were present
on the 11 July, as evidenced by the capture of a wasp with two
empty puparia.
In Tuscany, summer aggregations were usually found on the
vegetation surrounding houses with high nest density. Wasps
typically clustered under and between leaves of a few branches
on prominent hedges and trees (Molbus,Ficus,Hedera,Vitis);
P. dominulus males intensively patrolled the more exposed
branches of the same trees in August and September. In area
1, many aggregations were observed on the lower branches of
two large mulberry trees which were the sites of a previous
long-term studies on Polistes male lek-territoriality at land-
marks (Beani and Turillazzi, 1988, 1990). After 26 August,
when the temperature abruptly decreased, aggregations were
mainly found in sheltered sites, for example, inside creepers
on buildings (Hedera,Vitis), eaves, interstices, and corners.
For July and August censuses, the mean size of 74
aggregations on vegetation was 12.93 61.07 wasps, and this
did not vary according to either area (seven areas, Kruskal-
Wallis test, H
6
¼3.48, p¼.74) or week (5 weeks, H
4
¼5.03,
p¼.28). The parasite prevalence, that is, the proportion of
stylopized females, was 0.98 60.01 regardless of area of
collection (logistic regression, Wald ¼1.45, df ¼6, p¼.96) or
week (Wald ¼0.21, df ¼1, p¼.88). Twenty-seven wasps from
aggregations appeared unparasitized but after dissection were
found to contain late endoparasitic stages of both sexes.
In area 1, where aggregations were continuously checked
until October, the number of wasps and aggregations reached
a peak at the beginning of August (Figure 2a). The mean
parasite prevalence decreased from 0.98 60.01 during the
first 5 weeks to 0.64 60.07 during 6–12 weeks. The likelihood
of being parasitized decreased significantly between weeks 5
and 6 (odds ratio ¼1.38, Wald ¼84.15, df ¼1, p,.001). This
corresponded with the abandonment of mulberry trees.
Thereafter, aggregations were found in more sheltered sites
on buildings. The mean size of these late aggregations did
not significantly differ from early ones (14.32 60.03 versus
13.87 60.02 wasps; t
54
¼0.01, p¼.92).
All aggregations in area 1 (Figure 2b), except week 12,
included wasps infected by parasites of both sexes, mainly one
per wasp (81%). The presence of adult male X. vesparum in
the environment is indicated by a peak of 50 empty and 163
closed puparia at the beginning of August; a lower spike in
September was owing to presence of empty puparia. The sex
Figure 2
A survey of aggregations from area 1 over the course of the season.
(a) The number of parasitized and uninfected female wasps from
aggregations according to the week of collection and the
prevalence, that is, the proportion of infected individuals. Number of
aggregations in parenthesis. The mean parasite prevalence was not
affected by week when only wasps from weeks 1–5 or 6–12 were
included in the model (Wald ¼0.02, df ¼1, p¼.88 and Wald ¼2.29,
df ¼1, p¼.14 respectively). (b) The number of X. vesparum
(including empty puparia) in aggregations according to week
and sex ratio (proportion of males, not including empty puparia).
Hughes et al. •Behavioral changes in Polistes after parasitism 1039
by guest on June 3, 2013http://beheco.oxfordjournals.org/Downloaded from

ratio of X. vesparum was initially male biased, with a definitive
shift toward females in late September–October; an increasing
female bias was expected owing to the short male lifetime and
the fact that only females overwinter. In the first week of
August, 67 wasps in three aggregations were collected from
the roof of an abandoned building in area 1, within interstices
later used during diapause (not included in Figure 2a,b).
These aggregations were characterized by a higher proportion
of female parasites compared with ones on vegetation from
the same week and area, (0.55 versus 0.16, odds ratio ¼0.14,
Wald ¼37.15, df ¼1, p,.001) and a lower parasite
prevalence (0.68 versus 0.98, odds ratio ¼0.07, Wald ¼51.72,
df ¼1, p,.001).
Behavior of wasps within aggregations
Stylopized wasps within aggregation did not, apparently,
return to their nests, although at least until mid-August
colonies were fully active. In area 1, none of 297 stylopized
wasps, which were marked and released from 16 July–7
August, were resighted on any of the 26 P. dominulus nests
(801 females) collected during the same period within 200 m
of the aggregation sites, that is, under the mean flight range
of workers (Ugolini and Cannicci, 1996). Behavioral observa-
tions in the field have thus focused on extranidal aggregations
on vegetation, in which 98% of wasps were parasitized. Wasps
alighting on a leaf moved directly to the aggregation, joined
it, and were subsequently inactive (mean time inactive/2
min ¼0.88 60.05). For individuals that had already
aggregated, the proportion of time spent inactive was 0.97 6
0.01. Typical nest behaviors, such as aggression and trophal-
laxis, were not observed among aggregating wasps. Although
within aggregations wasps were predominantly inactive, they
did move between aggregations.
Of the 402 stylopized wasps marked up to 31 August, only
50 were recaptured. For five cohorts examined in detail
(Table 1), the probability of recapture of a marked wasp on
subsequent censuses was very low, owing to spacing out of
recaptures (not on successive days) and extensive movement
among aggregations, rather than to removal of marked wasps
from the population owing to mortality. In fact, the
probability of survival until subsequent sampling was very
high (Table 1), and some individuals were recaptured after
a long time (29 days in three cases!). Although a high
turnover of individuals existed among aggregations, the same
leaves were occupied by aggregating wasps; in one case wasps
were present on a single leaf on a mulberry tree for 36 days.
Behavior of artificially infected workers
To observe the spatial behavior of wasps, we divided laboratory
cages into nesting and foraging areas. In the first experiment,
nine wasps (infected by six male, three female X. vesparum)
left the colony, grouped together, and remained inactive
inside the foraging area regardless of the sex of the parasite,
whereas nine controls did not (proportion of time off nest,
Mann-Whitney Utest, U¼45; n
1
¼9, n
2
¼9; p,.001).
In the second experiment, both the position and behavior
of 12 wasps (infected by nine male, three female X. vesparum)
was compared with 14 controls. From the first day of
adulthood until day 10, infected wasps were significantly
more inactive than were controls (repeated-measures ANOVA,
p,.01) (Figure 3a) and spent significantly more time off the
nest than controls (p,.001) (Figure 3b). Infected wasps were
never observed feeding brood, building cells, or foraging, that
is, working, and checked cells less than did controls (p,.001)
(Figure 3c). Infected wasps did not receive more aggression
than did controls (p¼.09, post hoc power test, b¼0.63,
effect size ¼0.8) (Figure 3d); in fact, the trend was for
controls to receive more aggression.
All parasitized wasps deserted the nest and remained within
the attached net. The mean date of departure, that is, the first
day an individual spent 100% of its time off the nest, was 5.25
60.93 days posteclosion, which was before the extrusion of
the parasite through the cuticle of the host (8.7 60.60 days).
Controls only left the nest to forage. Nest departure before
the extrusion of the parasite through the cuticle was also
confirmed by the presence of 27 evidently nonparasitized
wasps in aggregations that were dissected and found to
contain endoparasitic final instar X. vesparum larvae with
sclerotized mandibles (to aid in their extrusion through the
cuticle): some were observed in the act of extruding during
capture. Further, based upon nest collections we noted that
very few X. vesparum extruded while their hosts were on the
nest. Of the 894 adults dissected from 21 P. dominulus nests
(area 1, 20/21 nests contained parasitized adults), 61 wasps
were infected with 69 X. vesparum, and of those, only eight had
extruded through the host cuticle (Fisher Exact test, p,
.0001). Thus, less than 1% of adults on nests from a highly
infected area were visibly parasitized (i.e., had extruded
cephalothorax/cephalothecae of X. vesparum).
DISCUSSION
Colony desertion and subsequent extranidal aggregation
formation by stylopized Polistes females has here been
extensively documented in the field and induced in the
laboratory after the artificial infection of a social insect with
a macroparasite. The occurrence of aggregations from seven
sites over the first 5 weeks in which 98% of female P. dominulus
were stylopized is a priori evidence of parasite-mediated
behavioral change (Figure 2a). Although the phenomenon is
common (aggregations were noted also in 2001, 2002, and
2003 at our study sites, data not shown), this is the first
example, to our knowledge, of a behavioral change induced
by a parasite in Polistes wasps. In other social insects, colony
desertion, but not aggregation outside the nest, may follow
parasitism by nematodes, entomopathogenic fungi, tremat-
odes, and parasitoids (Moore, 2002; Schmid-Hempel, 1998).
The lack of attention hitherto given to strepsipteran
parasitism of Polistes, despite its high prevalence in paper
wasp colonies (Hughes et al., 2003), is owing to their cryptic
nature (Figure 1b) and the absence of evidently parasitized
adults on both field and laboratory nests: colony desertion
occurred early in adult host life (mean ¼5.25 days), before
the parasite extruded through the cuticle. The subtle
influence of this parasite has obvious implications for studies
on the social behavior of Polistes; 31 other species have been
recorded as hosts of Strepsiptera (Hughes, 2003).
Table 1
Mark-recapture analysis from area 1: probability of survival and
recapture for wasps marked on a certain date (i.e., cohort)
Cohort
date (N)
Proportion
recaptured
Probability
of survival
Probability
of recapture
16 July (21) 0.38 0.97 60.02 0.07 60.04
24 July (55) 0.20 0.91 60.03 0.07 60.02
25 July (31) 0.10 0.88 60.07 0.06 60.05
28 July (31) 0.12 0.86 60.09 0.06 60.06
1 August (61) 0.28 0.93 60.03 0.07 60.02
1040 Behavioral Ecology Vol. 15 No. 6
by guest on June 3, 2013http://beheco.oxfordjournals.org/Downloaded from

Aberrant aggregations
Stylopized wasps deserting the field nests early in the season,
as well as in our laboratory experiments, were nominal
workers; that is, in the absence of parasitism they would have
engaged in colony tasks. Nevertheless, determining the caste
for stylopized members of this primitively eusocial wasp is
difficult because cues of caste—such as size, fat levels, ovarian
development, and behavior with siblings—are all likely to be
affected by parasitism. At the end of August (week 6), when
the temperature notably decreased, the location of aggrega-
tions changed from exposed branches on trees to low
sheltered vegetation and buildings, which were later adopted
as hibernation sites. There was a significant decrease in
parasite prevalence among aggregation occupants in area 1
(Figure 2a). This might be explained by the entry of healthy
future queens, as these leave the nest to mate and aggregate
outside, and may represent the transition from worker phase
to prediapause aggregations. In addition, the shift from male-
to female-biased sex ratio of the parasite (Figure 2b) suggests
that only wasps infected by female X. vesparum moved to
hibernacula, whereas wasps containing empty puparia pre-
sumably died after male X. vesparum emergence, owing to
their absence in extensive collections of overwintering healthy
and infected wasps (data not shown).
Previous extranidal collections of female stylopized Polistes
(Hubbard, 1892; Pierce, 1909, 1918; Wheeler, 1910) do not
mention aggregations. It is probable that high host density at
our study sites promoted encounters among deserting wasps.
Aggregation sites overlap spatially and temporally with either
lek-sites (Beani, 1996) or prehibernation sites, which may
signal some nonrandom spatial preference by infected and
uninfected individuals. Aggregation formation follows from
simple rules such as ‘‘stop when you encounter another
individual’’ (Deneubourg et al., 2002) and could explain their
patch distribution (only some branches of a tree or a portion
of a roof eave had been selected as an aggregation point).
Aggregations were characterized by extreme inactivity of wasps
(97% time inactive), high turnover (Table 1), and strong site
attachment (in area 1 a particular leaf was occupied for
Figure 3
The behavior of artificially infected female wasps (filled circles) versus controls over the first 10 days of adult life. Parasitized wasps spent
significantly more time inactive (repeated-measure ANOVA, F
1,24
¼7.57, p,.05) (a); away from the nest (F
1,24
¼29.78, p,.001) (b);
checked cells significantly less (F
1,24
¼29.64, p,.001) (d); and did not receive more aggression than controls (F
1,24
¼3.23, p¼.09) (d).
Hughes et al. •Behavioral changes in Polistes after parasitism 1041
by guest on June 3, 2013http://beheco.oxfordjournals.org/Downloaded from

36 days). (A high turnover and inactivity are not mutually
exclusive as behavioral observations focused on within-
aggregation behavior rather than a daily time budget, which
would have included interaggregation movement.)The de-
fensive capabilities of Polistes, combined with their aposematic
coloration, should promote aggregations (Guilford, 1990),
which are common in paper wasps at the end of the
reproductive phase (Reeve, 1991; Turillazzi, 1980; West-
Eberhard, 1969). To date, aggregations of stylopized social
insects are known only for the genus Polistes. These ‘‘aestiva-
tion/hibernation gatherings,’’ first noted by W.D. Hamilton
(e-mail to Laura Beani, 21-3-1999, 11:20 hrs; both e-mails
stored at W.D. Hamilton Archive, in preparation at the British
Library, material available by appointment. See Summers A,
Leighton John J, 2001. The W.D. Hamilton Archive at the
British Library. Ethology, Ecology and Evolution 13:373–384)
may be an example of exploitation of the preexisting
gregarious behavior and a way to reduce extrinsic mortality
outside the nest during final parasite development.
Why do stylopized wasps leave the nest?
Although aggregations of stylopized females are surprising,
the salient point is that parasitized females deserted the nest
both early in their adult life and in the season. For Polistes,
nest departure normally occurs for sexuals (males and future
queens) to secure matings while workers remain on the nest
to gain indirect fitness benefits through work (but for
desertion of healthy workers for direct fitness, see Reeve et
al., 1998). Stylopized females cannot achieve direct fitness as
they are physiologically castrated (Strambi and Strambi,
1973), so may be expected to remain on the nest and gain
indirect fitness benefits through work. In the present study,
infected individuals were not observed to work (Figure 3c),
and this corresponds to the general lethargy reported for
infected Vespa (unpublished data cited in Matsuura and
Yamane, 1990). Therefore, why do infected wasps leave the
nest? Desertion is undoubtedly associated with the presence
of Strepsiptera and may benefit the parasite or the host or
simply be a ‘‘boring by-product’’ of infection (Dawkins, 1990;
Moore, 2002; Poulin et al., 1994). Within this framework, we
explore some explanations for the nest desertion by
P. dominulus females after infection by Strepsiptera.
Desertion owing to eviction by siblings of lazy and costly
workers was not observed, and the trend was for stylopized
individuals to receive less aggression than did controls (Figure
3d). The time spent on the nest, by stylopized individuals, was
characterized by extreme inactivity (Figure 3a). Desertion to
satisfy the nutritional needs of the developing parasite, or
stressed host, appears unlikely as the nest is a rich source of
food, whereas lek-sites only occasionally have resources
(Beani, 1996). Stylopized wasps were observed maxillating
prey items and imbibing larval secretions; the latter are 50
times more rich in amino acids than is flower nectar (Hunt et
al., 1982). In fact, desertion occurred soon before the
extrusion of the final instar and after a period of maximal
growth. Desertion as an altruistic act to reduce infection to kin
would appear a good general strategy for infected social
insects but is untenable in this case because a female X.
vesparum is infective only if inseminated and copulation
appears not to occur on the nest (see below); moreover, wasps
parasitized by both sexes of the parasite desert the nest. In
bumblebees infected by conopid flies, nest desertion at night
retards parasite development (the nest is thermoregulated;
Muller and Schmid-Hempel, 1993). Polistes nests, which are
not theromoregulated, did not show a similar pattern in
desertion times. In the later stages of conopid infection, the
host deserts the nest completely and, through aberrant
‘‘digging’’ behavior, promotes overwintering survival of the
conopid pupa (Muller, 1994). In both host groups, obvious
parallels exist with the final desertion of the colony by the
infected individual occurring to promote parasite life-cycle
completion.
In line with Hamilton’s attention to parasites, we here
hypothesize that nest desertion and aggregation by stylop-
ized wasps is an example of adaptive parasite manipulation
of host behavior in order to facilitate parasite mating.
Copulation is unlikely to occur on the nest as free-living
males are vigorously attacked by occupants (when stylopized
wasps are constrained to remain close to the nest until male
emergence; Hubbard, 1892; Hughes D, personal observa-
tion). Moreover, the short adult stage (less than 5 h) of
male X. vesparum means that males would have to emerge at
the exact time a neotenic female, parasitic within a host on
a nest, is both present and ready to be inseminated. Mate
encounter is enhanced away from the nest, at aggregations.
The peak of the mating period, as indicated by the
maximum number of empty puparia, was in week 3 (Figure
2b), which corresponded with the peak in both aggregation
and stylopized wasp number (Figure 2a). Although male
Strepsiptera are generally considered extremely rare, they
can be found close to aggregations. We observed five volant
X. vesparum males and one mating on a leaf 20 cm from an
aggregation. We also attracted seven males to a caged
receptive female (data not shown). In addition, assuming
copulation could take place on the nest, the fecundity of
female X. vesparum is more than 3000 first instars per female
(data not shown), so although remaining on the nest affords
vertical transmission opportunities, the high probability of
injurious superparasitism (at any one time there are less
than 50 larval wasps/nest) has probably selected for nest
dispersal and increased horizontal transmission.
Anecdotal evidence suggests that stylopized workers of Vespa
are inactive in the colony (Matsuura and Yamane, 1990) and
that stylopized ants display positive phototropism and eleva-
tion seeking on vegetation outside the nest (Cook, 1996;
Ogloblin, 1939). The date of departure from the nest by
infected Polistes (5.25 days) probably reflects a balance
between staying on the nest to gain food for parasite
development and leaving to reduce the risk of attacks towards
lazy and costly nest mates, or some hitherto unknown cost
related to the extrusion of the parasite through the host
cuticle. Thus, we suggest that nest desertion occurs to facilitate
parasite reproduction.
In conclusion, infection of Polistes females by Strepsiptera
results in a parasite-mediated behavioral change that we
suggest is adaptive to the parasite for life-cycle completion,
although specific tests to evaluate the costs and benefits of
nest desertion need to be carried out. Parasite-induced
changes in host behavior range from the relatively subtle
(Poulin and Latham, 2002) to the incredibly complex
(Eberhard, 2000). In the case of stylopized wasps, their
behavior is aberrant because it expressed out of context, both
spatially and temporally: colony desertion by workers and
summer aggregations of castrated wasps at leks. Finally, our
data has pertinence for the debate concerning the role that
parasites play in the evolution and maintenance of social
behavior. It has been strongly claimed (O’Donnell, 1997) that
parasitic castrators, particularly Strepsiptera in Polistes, reduce
intracolony conflict over reproduction and may promote
social behavior, sensu lato the ‘‘subfertility hypothesis’’ (West-
Eberhard, 1975). Here we demonstrate that stylopization,
rather than promote sociality, actually results in early nest
desertion by infected individuals.
1042 Behavioral Ecology Vol. 15 No. 6
by guest on June 3, 2013http://beheco.oxfordjournals.org/Downloaded from

We would like to thank Emily Caruso, Anna Seward, and members of
the Florence Group for the Study of Social Wasps for their assistance.
We thank the referee for valuable suggestions and Mark Brown, Alex
Kacelnik, and Alan Grafen for earlier comments that improved our
manuscript. L.B. and S.T. were supported by MURST grants, J.K. by
a Royal Society travel grant, and D.H. by the Hope Studentship in
Entomology, Oxford University and Jesus College, Oxford. We
dedicate this work to the lasting memory of Bill Hamilton.
REFERENCES
Baudoin M, 1975. Host castration as a parasitic strategy. Evolution 29:
335–52.
Beani L, 1996. Leklike courtship in paper wasps: ‘‘a prolonged,
delicate, and troublesome affair.’’ In: The natural history and
evolution of paper-wasps (Turillazzi S, West-Eberhard MJ, eds).
Oxford: Oxford University Press; 113–125.
Beani L, Turillazzi S, 1988. Alternative mating tactics in males of
Polistes dominulus (Hymenoptera, Vespidae). Behav Ecol Sociobiol
22:257–264.
Beani L, Turillazzi S, 1990. Male swarms at landmarks and scramble
competition polygyny in Polistes gallicus (Hymenoptera, Vespidae).
J Insect Behav 3:545–556.
Combes C, 2001. Parasitism: the ecology and evolution of intimate
interactions. Chicago: University of Chicago Press.
Cook JL, 1996. A study of the relationship between Caenocholax fenyesi
Pierce (Strepsiptera: Myrmecolacidae) and the red imported fire
ant Solenopsis invicta (Hymenoptera: Formicidae) (PhD disserta-
tion). College Station, Texas: Texas A&M University.
Dawkins R, 1990. Parasites, desiderata lists and the paradox of the
organism. Parasitology 100:S63–S73.
Deneubourg JL, Lioni A, Detrain C, 2002. Dynamics of aggregation
and emergence of cooperation. Biol Bull 202:262–267.
Eberhard WG, 2000. Spider manipulation by a wasp larva. Nature 406:
255–256.
Fitzgerald DV, 1938. Social wasps (Hym., Vespidae) from Trinidad
with a note on the genus Trypoxulon Latreille. Trans R Entomol Soc
Lond 87:181–91.
Guilford T, 1990. The evolution of aposematism. In: Insect defenses:
adaptive mechanisms and strategies of prey and predators. (Evans
DL, Schmidt JO, eds). Albany, New York: State University of New
York Press; 23–62.
Hamilton WD, 1964. The genetical evolution of social behaviour, II.
J Theor Biol 7:17–52.
Hubbard HG, 1892. The life history of Xenos. Can Entomol 24:257–
261.
Hughes DP, 2002. The value of a broad mind: some natural history
meanderings of Bill Hamilton. Ethol Ecol Evol 14:83–89.
Hughes DP, 2003. The behavioural ecology of strepsipteran parasites
of Polistes wasps (DPhil dissertation). Oxford: Oxford University.
Hughes DP, Beani L, Turillazzi S, Kathirithamby J, 2003. Prevalence of
the parasite Strepsiptera in Polistes as detected by dissection of
immatures. Insectes Sociaux 50:62–69.
Hunt JH, Baker I, Baker HG, 1982. Similarity of amino acids in nectar
and larval saliva: the nutritional basis for trophallaxis in social
wasps. Evolution 36:1318–1322.
Hunt JH, Brodie RJ, Carithers TP, Goldstein PZ, Janzen DH, 1999. Dry
season migration by Costa Rican lowland paper wasps to high
elevation cold dormancy sites. Biotropica 31:192–196.
Kathirithamby J, 1989. Review of the order Strepsiptera. Syst Entomol
14:41–92.
Kathirithamby J, 2000. Morphology of the female Myrmecolacidae
(Strepsiptera) including the apron, and an associated structure
analogous to the peritrophic matrix. Zool J Linn Soc 128:269–287.
Lebreton JD, Burnham KP, Clobert J, Anderson DR, 1992. Modeling
survival and testing biological hypotheses using marked animals:
a unified approach with case studies. Ecol Monogr 62:67–118.
Matsuura M, Yamane S, 1990. Biology of the vespine wasps. Tokyo:
Springer-Verlag.
Moore J, 2002. Parasites and the behavior of animals. Oxford: Oxford
University Press.
Muller CB, 1994. Parasitoid induced digging behavior in bumblebee
workers. Anim Behav 48:961–966.
Muller CB, Schmid-Hempel P, 1993. Exploitation of cold temperature
as defense against parasitoids in bumblebees. Nature 363:65–67.
Nelson JM, 1968. Parasites and symbionts of nests of Polistes wasps.
Ann Entomol Soc Am 61:1528–39.
O’Donnell S, 1997. How parasites can promote the expression of
social behaviour in their hosts. Proc R Soc Lond B 264:689–694.
Ogloblin AA, 1939. The Strepsiptera parasites of ants. In: Interna-
tional Congress of Entomology, Berlin 1938; 1277–1284.
Pardi L, 1942. Ricerche sui Polistini V. La poliginia iniziale in Polistes
gallicus (L.). Bollettino dell’instituto di Entomologia dell’Universita`
di Bologna 14:1–106.
Pardi L, 1948. Dominance order in Polistes wasps. Physiol Zool 21:
1–13.
Pierce DW, 1909. A monographic revision of the twisted winged
insects comprising the order Strepsiptera Kirby. Bull US Natl
Museum 66:1–232.
Pierce DW, 1918. The comparative morphology of the order
Strepsiptera together with records and descriptions of insects. Proc
US Natl Museum 54:391–501.
Poulin R, Brodeur J, Moore J, 1994. Parasite manipulation of host
behavior: should hosts always lose. Oikos 70:479–484.
Poulin R, Latham DM, 2002. Parasitism and the burrowing depth of
the beach hopper Talorchestia quoyana (Amphipoda: Talitridae).
Anim Behav 63:269–275.
Queller DC, Zacchi F, Cervo R, Turillazzi S, Henshaw MT, Santorelli
LA, Strassmann JE, 2000. Unrelated helpers in a social insect.
Nature 405:784–787.
Reeve HK, 1991. Polistes. In: The Social Biology of the Wasps (Ross
KG, Matthews RW, eds). Ithaca,New York: Cornell University Press;
99–148.
Reeve HK, Peters JM, Nonacs P, Starks PT, 1998. Dispersal of first
‘‘workers’’ in social wasps: causes and implications of an alternative
reproductive strategy. Proc Natl Aca Sci USA 95:13737–13742.
Schmid-Hempel P, 1998. Parasites in social insects. Princeton, New
Jersey: Princeton University Press.
Strambi A, Strambi C, 1973. Influence du de´veloppement du paraiste
Xenos vesparum Rossi (Insecte, Strepsitere) sur le syste´me neuro-
endocrinien des femelles de Polistes (Hyme´nopte` re, Vepside) au
debut de leur vie imaginale. Archives D’Anatomie Microscopique et
de Morphologie Experimentale 62:39–54.
Turillazzi S, 1980. Seasonal variations in the size and anatomy of
Polistes gallicus (L.) (Hymenoptera, Vespidae). Monitore Zoologica
Italiano 14:63–75.
Turillazzi S, West-Eberhard MJ, 1996. Natural history and evolution of
paper-wasps. Oxford: Oxford University Press.
Ugolini A, Cannicci S, 1996. Homing in paper-wasps. In: Natural
history and evolution of paper-wasps (Turillazzi S, West-Eberhard
MJ, eds).Oxford: Oxford University Press; 126–143.
West-Eberhard MJ, 1969. The social biology of Polistine wasps. Misc
Pub Museum Zool Univ Mich 140:1–101.
West-Eberhard MJ, 1975. The evolution of social behavior by kin
selection. Q Rev Biol 50:1–33.
Wheeler WM, 1910. The effects of parasitic and other kinds of
castration in insects. J Exp Zool 8:377–438.
Yamane S, 1996. Ecological factors affecting the colony cycle of Polistes
wasps. In: Natural history and evolution of paper-wasps (Turillazzi S,
West-Eberhard MJ, eds). Oxford: Oxford University Press; 75–98.
Hughes et al. •Behavioral changes in Polistes after parasitism 1043
by guest on June 3, 2013http://beheco.oxfordjournals.org/Downloaded from