The interaction between hosts and parasites or parasitoids is one of the most important forces driving evolutionary and ecological processes [1]. In order to reduce the impact of parasitoids, host species may evolve mechanisms to detect their enemies and employ adequate counterstrategies [2–5]. Parasitoids, in turn, are selected to evolve mechanisms that reduce the probability of being detected by their hosts to circumvent such countermeasures. This sets the stage for repeated cycles of adaptations and counteradaptations ("evolutionary arms race" [6–10]) between hosts and parasites, especially if the parasitoid is highly specialized on a single host species and has a large impact on host fitness [2, 11, 12].

Progeny of brood caring bees and wasps are particularly susceptible to parasitism [13–15]. Females of these species store large amounts of valuable nutrients as larval provisions in brood cells. These valuable resources attract a variety of brood parasites, either cleptoparasites that reduce the amount of resources available to the host's progeny or parasitoids that obligatorily kill the host larvae. Mostly, females of these parasitic species have to enter the nest or the brood cell to deposit eggs or larvae. Thus, the traces that are left by female brood parasites might be detected by the host. As a result, the hosts might abandon the nests or remove or destroy eggs of brood parasites [5, 16]. Brood parasites might also be encountered in the nest by the host and might be driven away, injured, or even killed (E. Strohm, unpubl. observations). Since insects heavily rely on their chemical senses for any kind of recognition or localization process [1, 17, 18], concealment of a brood parasite's actual or previous presence will require chemical camouflage (compounds sequestered from the host or the host's nest) or chemical mimicry (compounds synthesized by the mimic, definitions sensu [19]). In this study, we investigated the interaction between a hunting wasp, the European beewolf, Philanthus triangulum (Hymenoptera, Crabronidae) and its highly specialized brood parasitoid Hedychrum rutilans (Hymenoptera, Chrysididae). We investigated the following questions: Are cuckoo wasps (H. rutilans) detected and recognized by beewolf females at all? Is there a difference in host response towards the cuckoos outside and inside the nest and do the cuticular hydrocarbons play a role for the detection of the cuckoo wasps inside the nest? Is the chemical composition of the cuticular hydrocarbons of H. rutilans females similar to their host or to their host's prey?

Females of the European beewolf hunt honeybee workers (Hymenoptera, Apidae) as food for their progeny. Several paralyzed bees are temporarily couched in the main burrow (up to 1 m long) of the underground nest (see [20] for details on nest architecture). Eventually, the female closes the nest entrance, excavates a side burrow and a terminal brood cell, brings in one to six paralyzed bees, and oviposits on one of the bees [21]. Thereupon, she carefully closes the side burrow and subsequently has no contact to her progeny.

The cuckoo wasp, H. rutilans, is a specialized brood parasitoid of the genus Philanthus [16, 22]. However, since in Central Europe only one member of the genus, P. triangulum, is fairly abundant, H. rutilans is effectively monospecific in this region. This considerable degree of specialization is expressed by the unique oviposition strategy of H. rutilans. Most chrysidid wasps oviposit into the brood cell of their hosts at a defined stage of the provisioning cycle or after the brood cell has been finally closed [22]. In beewolves, however, the brood cell is excavated only after the female has brought in the bees and the nest entrance has been closed. Thus, the nest is blocked up and the female is attendant until the brood cell is finally closed. This leaves little opportunity for a cuckoo wasp to deposit an egg in the brood cell. As a consequence, H. rutilans females pursue two alternative strategies. Either they rapidly pounce and oviposit on a paralyzed bee when the female alights with its prey and enters the burrow ([23], E. Strohm unpubl. observation), or H. rutilans females wait in front of the nest until the host female leaves to forage and then enter the burrow and oviposit on the paralyzed bees that are temporarily couched there [22]. Thus, H. rutilans use the paralyzed bees as a Trojan horse to bring the egg into the brood cell. The latter seems to be the much more frequent mechanism. The mobile larva of H. rutilans climbs onto the beewolf larva, kills it, and feeds on the host larva and the bees. Thus, infestation by H. rutilans inevitably leads to a fitness reduction of the host. H. rutilans is considered to be the most important brood parasite of P. triangulum. The rate of parasitism varies between 3% and more than 30% ([4, 24–26], E. Strohm, unpubl. data). H. rutilans might even drive local aggregations of P. triangulum to extinction [25].

In both oviposition strategies, detection of the cuckoo wasp female by the beewolf female might decrease the cuckoo wasp's success. First, when encountered in the nest, cuckoo wasps might be carried to the nest entrance by beewolf females and thrown out [27]. Mostly, cuckoo wasps are not severely harmed due to the solidity and strong sculpturing of their cuticle and their ability to adopt a rolled-up defensive posture that protects the most vulnerable parts of the body (legs, mouthparts, antennae [22, 28]). Nevertheless, the wings are rather unprotected and might be injured by a beewolf female. Second, if beewolves females detect signs of the presence of cuckoo wasps they might remove bees from the nest that have possibly been parasitized [25]. Thus, a cuckoo wasp should avoid detection to minimize wastage of time and investment. This means that cuckoo wasp females should avoid detection when they are encountered by a host female in the nest. However, it would probably be even more important for the cuckoo wasps not to leave any detectable traces of their presence when they had entered the nest and oviposited on a paralyzed honeybee.

In order to assess whether beewolf females respond to the presence of H. rutilans females at all we observed the behavioral interactions outside of the nest. To test whether H. rutilans females employ chemical camouflage inside the nest we conducted to sets of behavioral experiments. First, in observation cages we recorded the interaction of the cuckoo wasps with beewolf females inside the nest burrow. Second, we conducted a recognition bioassay by assessing the response of beewolf females towards filter discs treated with different extracts: solvent only, cuticular extracts of another chrysidid, Chrysis viridula, and cuticular extracts of H. rutilans. We predicted that beewolf females should ignore the discs treated with solvent only (negative control), they should respond to the discs treated with C. viridula extract (positive control) and they should not (or only weakly) respond to H. rutilans extracts.

There are two evolutionary options for H. rutilans femalesto avoid olfactory detection by beewolves. First, cuckoo wasps could mimic the odor of the honeybees that are temporarily couched in the main burrow. Second, H. rutilans females might mimic their beewolf host. We consider the imitation of the cuticular compounds of the beewolf host the better alternative, since the host's hydrocarbon profile can be found all over the nest walls due to the digging activity (Kroiss and Strohm unpubl. data) and also on the honeybees. This is because in order to prevent the paralyzed bees from molding they are treated by the beewolf females with a secretion from the postpharyngeal gland that is identical to the beewolves' cuticular hydrocarbons [29–31].

To assess both alternatives, we analyzed the composition of cuticular compounds of beewolf females, cuckoo wasps, and honeybees. Furthermore, we included beewolf males as the a priori most similar group to beewolf females and, thus, a crucial comparison for assessment of this hypothesis. To control for the possibility that chrysidids and crabronids have similar patterns of cuticular hydrocarbons by chance we also analyzed closely related species: the chrysidid wasp Hedychrum nobile and its crabronid host Cerceris arenaria (subfamily Philanthinae, a weevil hunting wasp). A reasonable null hypothesis for the resemblance among the species under study might be based on their phylogenetic relationship. Crabronids and apids are closely related and constitute the superfamily Apoidea, whereas chrysidids branch off very early [32, 33]. Thus, the null hypothesis would predict that the cuticular profiles of beewolf females should be most similar to conspecifics males and Cerceris females, fairly similar to honeybees, and least similar to cuckoo wasps. Accordingly from a phylogenetic point of view, the chemical profiles of the two congeneric chrysidid species should be most similar to each other.

Beewolf females often attacked and evicted H. rutilans when they encountered them in front of their nest. This seems to be the rule for interactions between hosts and chrysidids although Linsenmaier [37] reported that there are also cases where chrysidids do not elicit antagonistic behavior by their hosts. Prolonged stays at hosts' nests as observed in H. rutilans, have also been reported for other chrysidids [22, 37, 38]. Staying in vicinity of the host nest might allow the chrysidids to adjust the timing of oviposition to the most suitable stage of the provisioning cycle or to enter the nest in the absence of the host female. That H. rutilans females placed themselves under some cover (e. g. leaves) during prolonged stays might, besides the reduction of water loss, represent an attempt to hide themselves from the host females. Other chrysidid species also seem to hide near the entrance of a host nest and inspect the nest or brood cell after the host female has deposited provisions and departed for a new foraging flight [37]. This suggests that, similar to beewolves, most host species might recognize cuckoo wasps visually outside the nest. Since most chrysidids are brightly colored (see e.g. drawings in [37]) this is not surprising.

However, there was no evidence that H. rutilans females were recognized by beewolf females in their nests although the nest owners approached the chrysidids several times. In contrast to Olberg's few reported cases [27], we could never observe that chrysidids were thrown out of the nest by beewolf females neither in the study population in the field, nor in observation cages in the laboratory, nor during prolonged observations of beewolf nest aggregations as part of another study [4]. Possibly, different populations show varying local adaptations in the defense of the parasitoid or this behavior is extremely rare. However, we can not exclude that the chrysidids that Olberg [27] saw were not H. rutilans. Anyhow, our observations suggest that the H. rutilans females are often not detected in the nests. This finding is supported by the recognition bioassay. Filter paper discs treated with an extract of H. rutilans were recognized significantly less frequently than paper discs treated with an extract of the "alien" cuckoo wasp C. viridula. This suggests that the properties of the cuticular hydrocarbons of H. rutilans allow the cuckoo wasp to avoid recognition either when encountered inside the nest or after visiting the nest for inspection or oviposition. Our observations furthermore suggest that in the vast majority of cases H. rutilans females enter the nest when the female is out foraging and that they are only rarely directly encountered by the nest owner. Thus, it seems most important not to leave traces in the burrow during inspection of the nest or oviposition.

There are numerous reports of parasites of social species that gain access to their hosts' nests and protection from attacks mainly by chemical camouflage and more rarely chemical mimicry [19, 39–45] see also [46, 47] for examples of chemical mimicry of sex pheromones in sexually deceptive orchids). Hydrocarbons are considered to represent the principal cues for nestmate recognition in social bees and wasps [41] and are most probably also involved in nest identification and species recognition in solitary species [48]. In our analysis, the GC-MS profiles of the cuticular hydrocarbons of H. rutilans and beewolf females show considerable similarity. At least, the null hypothesis based on the phylogenetic relationship of a closer resemblance between the two Hedychrum species as well as the two philanthine species was clearly contradicted. Our preliminary analysis of other species of chrysidids apart from H. nobile showed also distinct differences to the profile of H. rutilans (J. Kroiss, T. Schmitt, P. Schreier, E. Strohm, unpubl. data). Furthermore, other aculeate Hymenoptera show considerably different compositions of cuticular compounds [49–52]. This clearly contradicts a general similarity among all Hymenoptera or all aculeates or between chrysidids and crabronids. The profiles of the H. rutilans females were as close as or even closer to beewolf females than the profiles of beewolf males. Together with the behavioral tests, this provides strong evidence that H. rutilans females are chemically cloaked. This will help to reduce the probability of detection during or -more importantly- after their presence in the nest.

A priori, chemical camouflage and mimicry seem unlikely to evolve in a chrysidid wasp that attacks a solitary host. Chemical camouflage, i.e. the acquisition of mimetic compounds from a solitary host by a parasitic species, might be problematic since, in contrast to parasites of social hosts, there is little opportunity to sequester cloaking chemicals. Social host species possess large nests and a large number of colony members that might serve as sources for the relevant compounds. Brood parasites of solitary brood caring Hymenoptera have rarely been studied in detail [53]. The only example of chemical camouflage in a brood parasite of a solitary species comes from Nomada bees. In some species of this genus, females have been reported to acquire mimetic odors by being perfumed by males during mating. Females of these species seemed not to elicit aggressive responses when encountered by host females of the genus Andrena [2]. Chemical cloaking in chrysidid wasps has not yet been reported.

With chemical camouflage being an unlikely option for a brood parasite of solitary species, chemical cloaking might evolve by synthesis of the compounds, i.e., chemical mimicry. For most chrysidids this is also unlikely since they attack a large number of different host species [22] with a varying composition of cuticular chemicals that is not compatible with an efficient chemical cloaking. However, H. rutilans is de facto monospecific in the study area and is, thus, predestined to evolve chemical mimicry. The behavioral observations show that cuckoo wasps do not regularly stay in nests for long periods. This makes a sequestration of host chemicals that are only available in relatively small amounts at the walls of the burrow rather unlikely. The assumption that the cloaking compounds are synthesized by H. rutilans females themselves is supported by details of the composition of chemicals on their cuticle. The occurrence of considerably proportion of the (Z)-7 isomers of the respective major components of the beewolf cuticle ((Z)-9-C25:1 or (Z)-9-C27:1) in the brood parasitoid but the lack thereof (at least in comparable proportions) in beewolf females contradicts an acquisition of the chemicals from their host. Thus, most probably cuckoo wasps produce at least some of the compounds on their cuticle by themselves.

In addition to the qualitative imitation of the hydrocarbons of its host, the cuckoo wasps might also employ a quantitative strategy to evade detection by beewolf females. Preliminary data suggest that H. rutilans (as well as several other chrysidids studied by us) have an overall low level of cuticular substances (Kroiss, Schmitt, Strohm, unpubl. data). Such a reduction in the amount of cuticular hydrocarbons as a means to avoid detection by the host has been suggested by Jeral et al. [54] in thievery ants (see also [41]). In fact, the background against which the cuckoo wasp is perceived (or rather not perceived) by the beewolf female is the wall of the nest burrow. This is contaminated with compounds from the cuticle of the nest owner due to the contact with its mandibles during excavation and with the tarsi and abdomen during processing of the excavated sand and movement inside the nest. Preliminary analyses suggest that the typical beewolf cuticular hydrocarbons can be sampled from the nest walls (Kroiss and Strohm, unpubl. data). Thus, the chemical traces or the presence of a cuckoo wasp whose hydrocarbon profile is sufficiently similar to the background might not be recognized by a beewolf female.

H. rutilans might employ a combination of strategies to evade detection. If encountered in the nest, they run away or remain motionless. They possibly leave only very small amounts of decisive and traitorous substances in the nest. Most notably, the composition of their cuticular hydrocarbons is very similar to that of their host. Thus, H. rutilans females seem to be able to avoid detection when directly encountered by a beewolf female in the nest. Much more important, however, is the reduction of the conspicuousness of scent marks left in the nest burrow or on the bee during oviposition. This is to our knowledge the first reported evidence for chemical mimicry (sensu [19]) in a parasitoid of a solitary wasp.