Parasitoid increases survival of its pupae by inducing hosts to fight predators

Many true parasites and parasitoids modify the behaviour of their host, and these changes are thought to be to the benefit of the parasites. However, field tests of this hypothesis are scarce, and it is often unclear whether the host or the parasite profits from the behavioural changes, or even if parasitism is a cause or consequence of the behaviour. We show that braconid parasitoids (Glyptapanteles sp.) induce their caterpillar host (Thyrinteina leucocerae) to behave as a bodyguard of the parasitoid pupae. After parasitoid larvae exit from the host to pupate, the host stops feeding, remains close to the pupae, knocks off predators with violent head-swings, and dies before reaching adulthood. Unparasitized caterpillars do not show these behaviours. In the field, the presence of bodyguard hosts resulted in a two-fold reduction in mortality of parasitoid pupae. Hence, the behaviour appears to be parasitoid-induced and confers benefits exclusively to the parasitoid.     

Good housekeeping: why do shelter‐dwelling caterpillars fling their frass?

Although both feeding and elimination of waste are imperatives for all animals, ecologists and evolutionary biologists have devoted considerable attention to foraging, while largely ignoring defecation. Many organisms, however, exhibit defecation behaviours that appear to have been shaped by natural selection in a range of ecological contexts. Accumulation of waste may pose particular challenges for animals that exhibit high site fidelity or live within enclosed spaces. In a taxonomically widespread but largely unexamined behaviour, many caterpillars that construct and inhabit leaf shelters ballistically eject their individual faecal pellets (frass) great distances at great speeds. Here, I show that elimination of chemical cues for natural enemies is likely to have been a driving force behind the evolution of frass ejection behaviour in skipper caterpillars; hygiene and crowding are less important in this system.     

Shelter-building caterpillars in the cerrado: seasonal variation in relative abundance, parasitism, and the influence of extra-floral nectaries

Caterpillar shelters provide protection against desiccation and natural enemies, whereas extra-floral nectaries (EFNs) may be an anti-herbivore adaptation that reduces herbivore abundance by attracting predators and parasites. We used a large, long-term dataset for caterpillars found in the Brazilian cerrado to examine temporal variations in the relative abundance of shelter-building caterpillars and exposed caterpillars, and to determine how much variation depends on the season and the presence of EFNs on host plants. We also compared the patterns of parasitism between sheltered and exposed caterpillars, between seasons, and between different host plants. The cerrado has a marked dry season, and its vegetation is a mixture of mostly deciduous shrubs and trees. Leaf production occurs mainly during the rainy season, and many plant species bear EFNs. Our results show that 60?% of cerrado caterpillars build shelters. These caterpillars were found to be proportionally more abundant during the dry season and less parasitized than exposed ones. The proportion of caterpillars building shelters was highest on plants with new leaves (functional EFNs), and parasitism of caterpillars on these plants was higher. Even though our study includes a taxonomically diverse suite of caterpillars that build many different types of shelter and a diverse set of plants and EFN types, our results suggest that EFNs play an important role in structuring caterpillar assemblages in the cerrado, and that the prolific use of shelters by caterpillars may be a result of their effectiveness in protecting caterpillars from natural enemies and desiccation.     

A single lycaenid caterpillar gets an ant‐constructed shelter and uninterrupted ant attendance

Ant‐lycaenid associations range from mutualism to parasitism and the caterpillars of some species of lycaenids are reported to enter ant nests for shelter, diapause, or pupation. The present study aimed to examine the nature of the association between Euchrysops cnejus (Fabricius) (Lepidoptera: Lycaenidae) and Camponotus compressus (Fabricius) (Hymenoptera: Formicidae) worker ants on the extrafloral nectary‐bearing cowpea plant, Vigna unguiculata (L.) Walp. (Fabaceae). The abundance patterns of the ants and the lycaenid caterpillars together with the spatial patrolling patterns of the ants on the plants revealed that ant abundance increased with the occurrence of the lycaenid caterpillars and the ants preferred the lycaenids over the extrafloral nectar. Camponotus compressus worker ants constructed a shelter at the cowpea plant base after interaction with one or more lycaenid caterpillar(s) and tended the caterpillars and pupae till the emergence of the butterfly. The ant‐constructed shelters (ACSs) inhabited by the minor caste workers (13 ± 1.3 ants per ACS), were utilized by the caterpillars to undergo pupation. The ants confined their activities predominantly to tending the pod‐feeding caterpillars and the solitary pupa within each ACS. It appears that the behavior of the tending worker ants is modulated by the lycaenid vulnerable stages.     

Fitness costs and benefits of shelter building and leaf trenching behaviour in a pyralid caterpillar

1. Shelter building and petiole trenching in the Lepidoptera is a behaviour that mediates ecological pressures including those exerted by both food plants and natural enemies. 2. Fitness costs and benefits of trenching and shelter‐building behaviour related to predation and larval performance were investigated in a pyralid species that inhabits and feeds on leaf shelters. 3. Assays comparing the performance of caterpillars feeding on trenched versus non‐trenched foliage and fresh versus dry leaves were conducted. Whereas pupal weight was positively affected by petiole trenching, larval development was delayed when caterpillars fed on dry leaves. 4. A field experiment comparing predation on caterpillars inside and outside shelters demonstrated that predation was significantly higher for exposed caterpillars. 5. No physiological costs associated with shelter building were found given that caterpillars performed equally regardless of the number of shelters they built. 6. The effect sizes of top‐down and bottom‐up forces on pupal weight, development time, and predation risk indicated that the major effect of shelters is through the reduction of predation risk. The integration of experiments and natural history observations showed that fitness benefits provided by shelters change across ontogeny.     

Host‐size decisions of female parasitoid wasps seeking hidden hosts

1. Choice of host size may play a critical role in parasitoid success, a task that takes on added complications when dealing with concealed hosts, but most such studies of insect behaviour have only taken place in the laboratory. 2. This study investigates the success of a wasp (Alabagrus texanus: Braconidae) in finding host caterpillars Herpetogramma theseusalis (Crambidae) of the most effectively handled size hidden in shelters, in both the field and the laboratory. 3. First, the study tested wasp preference and success in parasitizing large, middle‐sized and small caterpillars (> 5, 3–5, < 3 mm) presented in the open, one at a time, in the laboratory. The wasps attacked (inserted or attempted to insert their ovipositor) a higher proportion of middle‐sized (3–5 mm) caterpillars compared with either small (< 3 mm) or large (> 5 mm) caterpillars. Naïve wasps attacked large caterpillars more often than did experienced wasps. Wasps responded to increasing caterpillar size by increasing the number of legs used to pin their prey rather than by increasing handling time. 4. The frequencies of visits to shelters in the field containing a majority of either large or middle‐sized caterpillars were then compared, followed by a test providing the wasps with similar choices under controlled laboratory conditions. Wasps most frequently visited shelters containing a majority of middle‐sized caterpillars both in the field and under controlled laboratory conditions. 5. The combined results confirmed that the wasps can size‐select their hosts both in the field and in laboratory tests.     

Temporal and spatial variation of Stenoma cathosiota Meyrick (Lepidoptera: Elachistidae) caterpillar abundance in the cerrado of Brasilia, Brazil

The caterpillars of Stenoma cathosiota Meyrick feed on Roupala montana Aubl. (Proteaceae) in the cerrado of the Distrito Federal, Brazil. They construct shelters by joining leaves of the plant where they feed and pupate. The caterpillars are parasitized by a wasp (Hymenoptera: Brachonidae), which emerges from the pupae. Caterpillar abundance and parasitism frequency were associated in an area of frequently burned cerrado (biennial fire) and in another area that burns sporadically (1987 and 1994). For S. cathosiota, the variation among years in a single area, with sporadic fires, was greater than the variation among areas with different fire regimes. Caterpillar abundance among years was significantly different in the area that burns sporadically (chi2 = 24.06; df. = 1; P = 0.000). However, there were no significant differences on caterpillar abundance between areas for the same period (chi2 = 3.45; df. = 1; P = 0.063). Parasitism frequency was high, reaching 29% of the collected caterpillars, and did not differ among areas. The great temporal variation in abundance of lepidopteran caterpillars in the cerrado makes it difficult to determine the effects that fire exerts over this fauna.     

To mate or not to mate,and subsequent host search by a haplodiploid female parasitoid wasp

Mated females of haplodiploid species can vary the sex ratio of their offspring, but virgin or sperm‐depleted females can produce only males. Depending on the costs and benefits, the theory of constrained sex allocation states that female haplodiploids may vary in their propensity to mate, with important implications for the populations’ sex ratio. Unmated female parasitoid wasps Alabagrus texanus (Braconidae) have been observed to reject matings with males, even under highly confined spatial conditions. We performed field trials to determine whether unmated female A. texanus would mate. We then compared the preferences of the resulting unmated (constrained) and mated (unconstrained) female wasps for shelters constructed and occupied by their hosts, fern moth caterpillars Herpetogramma theseusalis (Crambidae) on fern fronds. We presented the wasps with pairs of shelters collected from the field that differed in size, as well as freshness, fern species, and presence or absence of caterpillars. Unconstrained females searched small shelters more frequently than did constrained females and tended to favor senesced (old) shelters over fresher ones. They did not differ in choice between shelters on marsh and sensitive ferns or between shelters containing caterpillars and empty shelters. The preference for small shelters by the unconstrained wasps may enhance their rate of contacting favorable hosts, but the foraging regime of the constrained females should expose them to caterpillars at sites infrequently searched by the unconstrained females.     

Parasitism by Coptera haywardi and Diachasmimorpha longicaudata on Anastrepha flies with different fruits under laboratory and field cage conditions

Larvae of Anastrepha ludens and A. serpentina that developed in mango and sapodilla fruits, respectively, were exposed to Diachasmimorpha longicaudata (Ashmead) (Hymenoptera: Braconidae) and sequentially exposed as pupae to Coptera haywardi (Oglobin) (Hymenoptera: Diapriidae). Sequential exposure to both parasitoid species contributed to a decrease in fruit fly emergence due to higher levels of parasitism, which varied according to fruit type. In creole mango, D. longicaudata represented the highest percentage of parasitism. C. haywardi parasitism was greater in pupae from Ataulfo mangos and sapodilla, where the pulp size and volume may have acted as a refuge, allowing fly larvae to escape and leaving a greater number of unparasitised pupae available to C. haywardi. Similar results were obtained under field cage conditions, but the level of parasitism by C. haywardi was lower, suggesting that its effectiveness has some limitations under natural conditions. Our results suggest that both species can exert complementary parasitism, which represents an alternative worth to investigate under open field conditions.     

Elevated volatile concentrations in high‐nutrient plants: do insect herbivores pay a high price for good food?

1. A tritrophic perspective is fundamental for understanding the drivers of insect–plant interactions. While host plant traits can directly affect insect herbivore performance by either inhibiting or altering the nutritional benefits of consumption, they can also have an indirect effect on herbivores by influencing rates of predation or parasitism. 2. Enhancing soil nutrients available to trees of the genus Eucalyptus consistently modifies plant traits, typically improving the nutritional quality of the foliage for insect herbivores. We hypothesised that resulting increases in volatile essential oils could have an indirect negative effect on eucalypt‐feeding herbivores by providing their natural enemies with stronger host/prey location cues. 3. Eucalyptus tereticornis Smith seedlings were grown under low‐ and high‐nutrient conditions and the consequences for the release of volatile cues from damaged plants were examined. The influence of 1,8‐cineole (the major volatile terpene in many Eucalyptus species) on rates of predation on model caterpillars in the field was then examined. 4. It was found that the emission of cineole increased significantly after damage (artificial or herbivore), but continued only when damage was sustained by herbivore feeding. Importantly, more cineole was emitted from high‐ than low‐nutrient seedlings given an equivalent amount of damage. In the field, predation was significantly greater on model caterpillars baited with cineole than on unbaited models. 5. These findings are consistent with the hypothesis that any performance benefits insect herbivores derive from feeding on high‐nutrient eucalypt foliage could be at least partially offset by an increased risk of predation or parasitism via increased emission of attractive volatiles.     

Does a polyphagous caterpillar have the same gut microbiota when feeding on different species of food plants?

We used classical culture techniques to explore gut bacteria and changes associated with dietary change in the highly polyphagous, tropical caterpillar Automeris zugana (Saturniidae). Fifty-five third instar wild-caught sibs feeding on Annona purpurea (Annonaceae) in the Area de Conservación Guanacaste (ACG) in northwestern Costa Rica were divided into eight groups. Each of seven groups was reared to the ultimate instar on another species of food plant normally used by A. zugana. Some pupae were also analyzed for the presence of bacteria. Aerobic bacterial cultures were obtained from all 33 caterpillar guts and the eight pupae inventoried. There was no clear pattern in species composition of cultivated bacteria among the eight diets, and each caterpillar on a given food plant carried only a small fraction of the total set of species isolated from the set of caterpillars feeding on that food plant. Taken as a whole, the larvae and pupae contained 22 species of cultivable bacteria in 12 genera. Enterobacter, present in 81.8% of the samples, was the genus most frequently isolated from the caterpillars, followed by Micrococcus and Bacillus. Bacillus thuringiensis was isolated from 30.3% of the dissected caterpillars, but found in caterpillars feeding on only half of the species of food plants.     

Delayed density-dependent parasitism of eggs and pupae as a contributor to the cyclic population dynamics of the autumnal moth

Many populations of forest Lepidoptera exhibit 10-year cycles in densities, with impressive outbreaks across large regions. Delayed density-dependent interactions with natural enemies are recognized as key factors driving these cyclic population dynamics, but emphasis has typically been on the larval stages. Eggs, pupae and adults also suffer mortality from predators, parasitoids and pathogens, but little is known about possible density relationships between mortality factors and these non-feeding life stages. In a long-term field study, we experimentally deployed autumnal moth (Epirrita autumnata) eggs and pupae to their natural enemies yearly throughout the 10-year population cycle in northern Norway. The abundance of another geometrid, the winter moth (Operophtera brumata), increased in the study area, permitting comparisons between the two moth species in predation and parasitism. Survival of autumnal moth eggs and pupae was related to the moth abundance in an inverse and delayed manner. Egg and pupal parasitoids dominated as density-dependent mortality factors and predicted the subsequent growth rate of the host population size. In contrast, effects of egg and pupal predators were weakly density dependent, and generally predation remained low. Parasitism rates did not differ between the autumnal and winter moth pupae, whereas predators preferred winter moth pupae over those of the autumnal moth. We conclude that parasitism of the autumnal moth by egg and pupal parasitoids can be related to the changes of the moth density in a delayed density-dependent manner. Furthermore, egg and pupal parasitoids cannot be overlooked as causal factors for the population cycles of forest Lepidoptera in general.     

Ecological consequences of shelter sharing by leaf-tying caterpillars

A wide variety of insect herbivores construct and inhabit leaf shelters (ties, rolls, folds, and webs). Shelter construction can lead to a high rate of secondary occupation by other arthropods, including other species of constructors. The consequences for the inhabitants of secondarily occupying these shelters are currently unknown. In this study, we conducted field experiments to examine the fitness consequences (survival and attack by natural enemies) for caterpillars that (i) occupy a shelter with conspecifics vs. occur singly; and (ii) establish a new shelter vs. colonize a pre‐existing one. In addition, we conducted factorial laboratory experiments to test the hypothesis that caterpillars sharing shelters with conspecifics might have reduced construction costs (a potential benefit of shelter‐sharing or secondary occupation). Larvae of Psilocorsis quercicella Clemens (Lepidoptera: Oecophoridae) placed in white oak [Quercus alba L. (Fagaceae)] leaf ties alone or in groups of three had equal likelihood of survival from natural enemies. This same caterpillar species, however, had a higher disappearance rate when placed in pre‐existing leaf ties than when placed in newly formed ones, suggesting a potential cost of secondary colonization. A similar experiment with a closely related species [Psilocorsis cryptolechiella (Chambers)], however, failed to detect a cost of secondarily occupying shelters made on beech, Fagus grandifolia Ehrh. (Fagaceae). In the laboratory experiment, we found no evidence of shelter‐sharing benefits; rather larvae reared in shelters in groups of three had lower pupal mass (and thus lower potential fecundity) than larvae reared singly, suggesting a cost of shelter sharing. Moreover, groups of larvae forced to repeatedly construct new shelters tended to have reduced survival relative to the other treatment, suggesting that energetic constraints are more likely to reduce fitness when larvae cohabit shelters. Taken together, these results indicate that the common phenomenon of shelter sharing by leaf‐tying caterpillars has either neutral or negative effects for the occupants. The fact that these leaf‐tying caterpillars actually share shelters may simply reflect limited availability of oviposition sites.     

Development of a solitary koinobiont hyperparasitoid in different instars of its primary and secondary hosts

Parasitoid wasps are excellent organisms for studying the allocation of host resources to different fitness functions such as adult body mass and development time. Koinobiont parasitoids attack hosts that continue feeding and growing during parasitism, whereas idiobiont parasitoids attack non-growing host stages or paralyzed hosts. Many adult female koinobionts attack a broad range of host stages and are therefore faced with a different set of dynamic challenges compared with idiobionts, where host resources are largely static. Thus far studies on solitary koinobionts have been almost exclusively based on primary parasitoids, yet it is known that many of these are in turn attacked by both koinobiont and idiobiont hyperparasitoids. Here we compare parasitism and development of a primary koinobiont hyperparasitoid, Mesochorus gemellus (Hymenoptera: Ichneumonidae) in larvae of the gregarious primary koinobiont parasitoid, Cotesia glomerata (Hymenoptera: Braconidae) developing in the secondary herbivore host, Pieris brassicae (Lepidoptera: Pieridae). As far as we know this is the first study to examine development of a solitary primary hyperparasitoid in different stages of its secondary herbivore host. Pieris brassicae caterpillars were parasitized as L1 by C. glomerata and then these parasitized caterpillars were presented in separate cohorts to M. gemellus as L3, L4 or L5 instar P. brassicae. Different instars of the secondary hosts were used as proxies for different developmental stages of the primary host, C. glomerata. Larvae of C. glomerata in L5 P. brassicae were significantly longer than those in L3 and L4 caterpillars. Irrespective of secondary host instar, every parasitoid cluster was hyperparasitized by M. gemellus but all only produced male progeny. Male development time decreased with host stage attacked, whereas adult male body mass did not, which shows that M. gemellus is able to optimally exploit older host larvae in terms of adult size despite their decreasing mass during the pupal stage. Across a range of cocoon masses, hyperparasitoid adult male body mass was approximately 84% as large as primary parasitoids, revealing that M. gemellus is almost as efficient at exploiting host resources as secondary (pupal) hyperparasitoids.     

Interspecific competition between Xanthopimpla stemmator Thunberg and Dentichasmias busseolae Heinrich (Hymenoptera: Ichneumonidae), pupal parasitoids attacking Chilo partellus (Lepidoptera: Crambidae) in East Africa

Interspecific competition between Xanthopimpla stemmator and Dentichasmias busseolae was studied using pupae of the invasive crambid stemborer Chilo partellus as the host. While X. stemmator is an old association, D. busseolae formed a relatively new association with C. partellus in East Africa. Two different time intervals between parasitism (0 and 48 h) and two parasitoid sequences [i.e., X. stemmator before D. busseolae (Xs–Db) and D. busseolae before X. stemmator (Db–Xs)] were chosen. In addition, the parasitoids’ performance on pupae in maize stems and ears was assessed. For both X. stemmator and D. busseolae, there was no difference in foraging time between unparasitized pupae and pupae previously parasitized by the other species, indicating that the two species were not capable of interspecific host discrimination. In the Xs–Db sequence, the time interval between parasitism did not have an influence on the percentage of pupae producing either parasitoid species. By contrast, in the Db–Xs sequence, the percentage of pupae producing X. stemmator was almost 8 times higher in the 0-h than the 48-h interval, while for D. busseolae it was the reverse. In the 0-h interval, X. stemmator outcompeted D. busseolae irrespective of whether it parasitized first or second, while in the 48-h interval, the parasitoid parasitizing first won. While D. busseolae successfully searched for and parasitized pupae in both stems and ears, parasitism of pupae in ears by X. stemmator was negligible. It was concluded that the two species could co-exist because they partly exploit different ecological niches.     

Ant-related oviposition is not associated to low parasitism of the myrmecophilous butterfly Allosmaitia strophius in an extrafloral nectaried shrub

In Lycaenidae-ant mutualisms, ovipositing females select plants based on the presence and/or species of ants in order to maximize survival rates of immatures. The ants are supposed to protect the immatures from parasitoids, but there is large variation in the protection provided. Here, we experimentally investigated whether the occurrence of the facultative myrmecophilous Allosmaitia strophius (the dominant species in our study system) was ant-related. The parasitism rates of immatures collected in the field and reared in the laboratory were also investigated. Stems of the extrafloral nectaried shrub Peixotoa tomentosa were designated as either ant-present (control) or absent (treated). The occurrence of A. strophius on ant-present stems was five times greater than on treated stems. Most eggs and larvae were associated with Camponotus blandus and Ectatomma tuberculatum, two aggressive ant species in the Brazilian savanna. Egg parasitism rate was 9%, and all the parasitized eggs were on ant-present stems. Pupal parasitism on ant-present and ant-absent stems was 25.6% and 7%, respectively. The higher parasitism rate in the presence of ants might also have been density-dependent, because caterpillars were more abundant in ant-present stems. Tropical lycaenids are frequently found in association with patrolling ants. Nevertheless, there is growing evidence that parasitism is higher in the presence of ants, owing to caterpillar's density-dependent effects in plants with ants, and/or to the weak lycaenid-ant associations. This indicates that the offspring of myrmecophilous lycaenids may not benefit, at least in terms of lower parasitism, by living with ants.     

Predatory wasps learn to overcome the shelter defences of their larval prey

Larvae of Epargyreus clarus (Hesperiidae), the silver-spotted skipper, inhabit leaf-and-silk shelters that they construct on their leguminous host plants. In the field, Polistes spp. (Vespidae) wasps land on the shelters, quickly extracting and killing the larvae within. In marked contrast, wasps that emerge from field-collected colonies maintained in the laboratory visit and examine leaflets bearing sheltered caterpillars, but only rarely do they extract and kill the sheltered larvae. To examine whether learning is involved in the development of the ability of Polistes wasps to forage successfully on sheltered E. clarus larvae, we tested the responses of P. fuscatus and P. dominulus wasps to sheltered E. clarus larvae before and after their exposure to unsheltered larvae that were visible either on an opened host-leaf shelter (P. fuscatus and P. dominulus) or on a nonhost leaf in the absence of a shelter (P. fuscatus). After killing and processing an unsheltered larva that was visible on an opened leaf shelter, a majority of foragers subsequently extracted and killed larvae from closed shelters. Wasps that killed and processed an unsheltered larva on a nonhost leaf, on the other hand, generally did not later open shelters. Thus, it seems that experience with an exposed larva in the context of its shelter is necessary for a wasp to be able to prey on sheltered larvae. We conclude that the wasps must learn to associate the taste of the larva with shelter-related cues, such as presence of leaf damage and silk. In nature, this initial exposure may occur when the larva is visible in or near its shelter, perhaps when feeding or constructing a new shelter. Learning opportunities will thus depend on larval density. Our results show that invertebrate predators can learn to overcome their prey's defences, and are therefore able to make use of previously inaccessible prey.     

Assessment of Ceratitis capitata (Diptera,Tephritidae) pupae killed by heat or cold as hosts for rearing Spalangia cameroni (Hymenoptera: Pteromalidae)

In this work, we study the suitability of using dead medfly Ceratitis capitata pupae, killed by heat- or cold-shock, for the mass rearing of Spalangia cameroni, a pupal parasitoid of key pests. 100% mortality of medfly pupae could be accomplished with cold-shock at –20°C for 60 min or with heat-shock at 55°C for 30 min. Neither parasitism percentage nor sex ratio of the offspring differed significantly among heat-shocked, cold-shocked and untreated pupae. In addition, there was no significant difference in the percentage of parasitoids that aborted (♂♂ or ♀♀) among pupal treatments. Some of the pupae were covered with peat because the third larval instar of the medfly buries itself before pupation. However, the buried pupae were not parasitised at a greater or lesser rate than those not covered with peat. The percentage of parasitism was also unaffected by whether the pupae had been killed recently or had been stored at between 4°C and 6°C over 15 or 30 days. The use of dead hosts and later storage permitted the following: (a) the use of hosts over long periods of time; (b) a rapid increase in parasitoid numbers and (c) the availability of pupae killed at the most suitable postpupation times for the production of parasitoids. Furthermore, in biological control projects, the use of dead parasitised pupae in the field avoids the risk of enhancing the pest and allows an increase in parasitism in the field through the use of pupae treated with cold- or heat-shock.     

Does Manipulation by the Parasitoid Wasp Cotesia glomerata (L.) Cause Attachment Behaviour of Host Caterpillars on Cocoon Clusters?

Some parasitoid wasps appear to control the behaviour of their hosts. However, altered behaviours of parasitised hosts are not necessarily caused by parasitoids but are sometimes the result of traumatic side effects of parasitism. However, it was difficult for us to discriminate the cause of host's behaviours between manipulation by parasitoids and traumatic side effects. Larvae of the parasitoid wasp Cotesia glomerata form cocoon clusters after egression from the parasitised host caterpillar Pieris brassicae . Following parasitoid egression, host caterpillars survive for several days and remain near the cocoon clusters. These caterpillars may repel solitary pteromalid hyperparasitoid wasps, Trichomalopsis apanteloctena , that attempt to parasitise fresh C. glomerata pupae. We allowed hyperparasitoids to attack cocoon clusters in the field and laboratory and then assessed the costs and benefits to C. glomerata of attachment by the parasitised caterpillars. The eclosion success of C. glomerata in cocoon clusters with attached caterpillars was higher than that in clusters without attached caterpillars in both field and laboratory experiments. This difference was attributed to shorter hyperparasitoid visits to cocoon clusters with attached host caterpillars. However, large cluster size was potentially costly for host attachment, because the duration of host caterpillar attachment decreased with increasing numbers of C. glomerata per caterpillar. This trade-off may be related to shortages of fat body resources, which are shared between the development of wasp larvae and the survival of host caterpillars. Therefore, we concluded that caterpillar attachment satisfied some requirements of host manipulation by C. glomerata .