How can cleptoparasitic drosophilid flies emerge from the closed brood cells of the red Mason bee?

Abstract

Aculeate Hymenoptera provision their progeny with large amounts of food. To protect their investment against brood parasites, females of many bee and wasp species construct brood cells that are hard to penetrate when finally sealed. However, the sealed brood cells also pose a problem for parasites that oviposit in the brood cell during provisioning. Brood parasites are smaller than their host and may lack strong mandibles to break through the solid brood cell walls. Furthermore, in nests built in existing cavities, newly-eclosed brood parasites need information about the location of the nest entrance. In the present study, the mechanisms of emergence are investigated in Cacoxenus indagator Loew (Diptera, Drosophilidae), the major cleptoparasite in nests of the red Mason bee Osmia bicornis L. (Hymenoptera, Megachilidae). Larvae of C. indagator move to brood cells closer to the nest entrance and sometimes make small emergence holes in the final closure of the nest entrance. Nevertheless, approximately one-third of newly-eclosed flies orientate and break through at least one intact cell partition to emerge. Flies make most of their attempts to emerge at the correct side (i.e. the one pointing to the nest entrance, probably by using the shape of the cell partition as a cue). Newly-eclosed flies use their head blister (ptilinum) to exert hydraulic pressure on particles of the cell partitions and produce small holes. Thus, C. indagator exhibits a set of behavioural and physiological adaptations enabling them to successfully emerge even from closed brood cells of their host.

Introduction

Host parasite interactions are among the most important components of ecological and evolutionary processes (Godfray, 1994). Host species should evolve mechanisms to reduce the impact of parasites (Tengö & Bergström, 1977; Rosenheim, 1988; Quicke, 1997; Strohm et al., 2001). Parasites, in turn, should evolve countermeasures to circumvent the host's protective adaptations. Adaptations and counteradaptations should be most obvious if the parasite is specialized and has a strong impact on host fitness (Tengö & Bergström, 1977; Sick et al., 1994; Spencer, 1998).

Aculeate Hymenoptera are particularly prone to parasitism. Female bees and wasps provision brood cells with large amounts of nutrients as food for their progeny (Wilson, 1971). These valuable resources and the defenceless larvae attract a variety of cleptoparasites and parasitoids. The larvae of these brood parasites cause death or reduced body size of the host larvae (Westrich, 1989). One counterstrategy of bees and wasps is to construct more or less sophisticated nests (e.g. those that extend deep in the soil; Strohm & Linsenmair, 1994–1995) and to carefully seal the brood cell after provisioning and oviposition (Krombein, 1967). In species that nest in pre-existing cavities, females gather different kinds of material (mud, sand, debris, chewed plant material) to construct cell partitions. These walls cannot be penetrated by most potential brood parasites and, thus, provide an effective protection once the brood cell is finally sealed (Krombein, 1967; O’Neill, 2001).

Although bees and wasps minimize the time the brood cell is available for parasitism (Goodell, 2003; Seidelmann, 2006), many brood parasites deposit their eggs in the brood cell during provisioning. For these parasitic species, brood cell sealing may have a delayed secondary effect. After development, newly-eclosed adult parasites face the problem of emerging from the nest because they often lack strong mandibles to break through the walls of the brood cells. In nests of mud dauber wasps, for example, approximately 12% of the parasitic miltogrammine flies fail to escape from the hosts' brood cells (Downing, 1995). Thus, there may be substantial costs for a parasitic female ovipositing in nests of hosts that protect their brood cells by thick walls of compact material. Moreover, in nests constructed in holes, the eclosed brood parasites require information about the direction that leads to the nest entrance to enable emergence (Downing, 1995). If a species of brood parasite fails to evolve adaptations to solve such problems, it should avoid parasitizing the respective host species. Thus, the host would have gained protection from some potentially parasitic species.

The present study addresses how newly-eclosed adults of the cleptoparasitic drosophilid Cacoxenus indagator Loew emerge from sealed brood cells in nests of its host, the red Mason bee Osmia bicornis L. The cell partitions in O. bicornis nests completely seal the brood cell and are composed of walls of mud (2–6 mm thick), with a final nest closure of up to 10 mm in thickness (E. Strohm, unpublished observations). Once dried, the partitions become hard and sturdy.

After feeding on the provisions in a brood cell, the larvae of C. indagator might move forward to the outermost brood cell or vestibulum (i.e. the empty space behind the final nest closure) and make small holes in the nest closure that serve as exits after eclosion from the puparium (Juillard, 1947, 1948; Coutin & de Chenon, 1983). As a result of their sclerotized mouthhooks and their flexible body, fly maggots appear to be well equipped to perforate and pass through the cell partitions. Preliminary observations suggest, however, that a considerable proportion of maggots do not prepare a way through the brood cell partitions and do not move forward to the outermost brood cell. Teneral adult Cacoxenus flies do not appear to be well equipped to break through the cell partitions because they lack strong mandibles. The easiest way to leave the brood cell would be to wait for an emerging host to break down the cell partitions. This would also eliminate the orientation problem because the flies could quickly find the exit. Thus, flies should eclose at the end of the emergence period of their host, as reported by Coutin & de Chenon (1983). However, not all flies might be freed by their hosts. Sometimes, a whole nest is parasitized (E. Strohm, unpublished observations) or flies emerge in the innermost brood cell, or the bees in the more basal brood cells die. In these cases, no host bee will break down the cell partitions and prepare an exit. Thus, the first goal of the present study is to assess the proportion of flies that is not freed by their host.

Assuming that breaking through the cell walls is somehow costly or not possible at all, a strategy that might increase the probability of an easy emergence is to minimize the number of cell partitions that an offspring has to penetrate. Accordingly, parasites should preferably oviposit in brood cells that are close to the nest entrance, as is suggested for the study species by Seidelmann (1999). Thus, the second goal of the present study is to test the prediction that parasitism should increase from the innermost to the outermost brood cell.

To identify the direction of the nest entrance poses a problem for both developmental stages, larvae as well as newly-eclosed adult C. indagator. However, the same orientation problem exists for the host's own progeny. To direct the emergence of their offspring, information about the location of the nest entrance is provided by the mother in trap-nesting bees and wasps: the direction that leads to the nest entrance is indicated by the structure of the cell partition (Cooper, 1957; Krombein, 1967; Torchio, 1980). Females construct the cell partitions in a way that the front side of the brood cell, which points toward the nest entrance, is rough and convex, whereas the rear side is smooth and concave (Fig. 1). If cell partitions are experimentally reversed, the young bees and wasps will move forward towards the dead end of the nest and may even die. Krombein (1967) suggests that brood parasites might use the same orientation cues as their hosts. However, to the author's knowledge, this has not yet been tested. Alternatively, female brood parasites might provide their own cues for their progeny or there is no particular orientation mechanism and brood parasites simply use a ‘trial and error’ approach. The latter would certainly be more costly in terms of energy required and probability of failure. The third goal of the present study is to test whether newly-eclosed C. indagator are able to use the shape of the cell partitions for orientation as their hosts do.

Figure 1. Brood cell of Osmia bicornis with the pollen provisions, an egg and cell partitions. The nest entrance is located to the right. The front wall of the brood cell (to the right) as seen from inside is convex and rough, whereas the rear wall (to the left) is concave and comparatively smooth.

Possible mechanisms by which newly-eclosed flies break through cell partitions are rarely studied. Alcock (2000) speculates that newly-eclosed adults of a miltogrammine fly moisten the wall of their host's brood cell to penetrate it. Downing (1995) reports observing the head blister of teneral adults of another species of miltogramminae pulsing when they emerge from nests of mud dauber wasps. However, he is unsure about the significance of this observation and speculates that the flies use their mouthparts to break through the cell partition. Thus, the final objective of the present study is to investigate whether newly-eclosed C. indagator are able to break through intact cell walls of their host nests and to provide a description of the mechanism.

To study the questions raised above, the location and requirements for emergence of the fly larvae in trap nests of O. bicornis are determined and an association between the location of a brood cell within the nest is tested for, along with the probability of it being parasitized. The timings of emergence of O. bicornis and C. indagator are compared. In addition, orientation by newly-eclosed flies is examined in artificial brood cells that provide only a convex and a concave partition. Finally, close observation of flies during emergence reveals the possible mechanisms that might be used to break through cell partitions.

Materials and methods

Insects

Osmia bicornis is probably the most abundant solitary bee in Central Europe (Westrich, 1989). The flight period usually starts in April and lasts until June. Females use different kinds of pre-existing cavities to construct their nests. They provision brood cells with pollen and nectar from a large number of plant species. Eggs are glued onto the provisions. Larvae feed on the provisions, spin a cocoon, pupate and eclose within the cocoon approximately 14 weeks after the egg was laid. Adults hibernate in the cocoons and emerge in late March or April.

In the study area, C. indagator is largely specialized on O. bicornis. Other Osmia species and species of the genus Megachile are less often parasitized. The flight period of C. indagator lasts from approximately April to June (Westrich, 1989). Females of the species perch in the vicinity of the nest entrances of their host and inspect nest tubes from time to time. They deposit two to eight eggs (Westrich, 1989) in a brood cell when it is being provisioned. Larvae feed on the provisions and, depending on the number of flies present in a brood cell, can reduce the size of the bee or even cause the death of the host larva. Larvae hibernate, pupate in mid-March and emerge in April shortly after the hosts (Coutin & de Chenon, 1983; Krunićet al., 2005). Up to 28% of the brood cells of O. bicornis may be parasitized by C. indagator (Raw, 1972) and often these flies are the most important cause of larval mortality in O. bicornis (Strohm et al., 2002, Krunićet al., 2005).

Examination of nests

Trap nests were placed in a known nesting habitat of O. bicornis near the Biocenter of the University of Würzburg, Germany. Trap nests consisted of two pieces of wood (20 × 2 × 1 cm). These two parts were screwed together (resulting in a piece of wood measuring 20 × 2 × 2 cm) and drilled lengthwise to a depth of 15 cm with a diameter of 5, 7 or 10 mm. These trap nests could be opened by unscrewing the two parts. Nest contents could thus be inspected without harming the brood cells. Osmia bicornis females readily accepted these trap nests for reproduction. After hibernation in the field, trap nests were brought to the laboratory and analyzed. Nests contained a mean ± SD of 5.5 ± 3.3 brood cells and 1.2 ± 0.5 empty chambers between the provisioned brood cells and the nest closure (such an empty chamber is called a vestibulum). The cell partitions were 2–6 mm in thickness, made of a mixture of loam and sand and always sealed a brood cell completely. The final nest closure was constructed from the same material and was approximately two to three-fold thicker than a cell partition. At the time of nest examination, brood cells contained cocoons with adult bees and/or fly puparia or the content of the brood cell was overgrown by fungus.

Parasitized brood cells could be easily identified by the filamentous faeces of the C. indagator larvae (Juillard, 1947, 1948). The distribution of parasites in nests was determined by counting the number of fly larvae in each brood cell at the end of February before emergence of the host bees and C. indagator flies. If brood cells contained fly maggots but no faeces, it was assumed that these individuals had moved there from a neighbouring brood cell. Using this information, it was possible to determine the minimum number of brood cell partitions that these maggots had penetrated on their way to their actual location.

Timing of emergence of O. bicornis and C. indagator

Unmanipulated nests containing O. bicornis cocoons and C. indagator puparia were exposed to natural environmental conditions in a small hut at the study site. From the beginning of March, nests were checked each morning and newly-eclosed male and female bees as well as flies were recorded. These observations were accomplished using trap nests with a transparent foil covering the nest tube (Strohm et al., 2002).

Orientation during emergence

Artificial brood cells were constructed in glass tubes (inner diameter 5 mm). Thus, chemical markings of the host or the fly or other cues possibly present in a natural brood cell were excluded. The artificial cell partitions closely resembled the original with respect to material (i.e. the same mud as used by the bees) and degree of convexity (i.e. it was not possible to use original cell partitions because they could not be fitted tightly into the glass tubes without damaging their shape). The artificial cell walls were created by shaping cylinders (diameter 5 mm) of mud with a custom made pistil with either a convex or concave head side, resulting in a concave or convex surface and press fitting these cylinders into the glass tubes. The surface of these artificial partitions was much smoother than in original brood cells. Each artificial brood cell had one convex (angle between side walls and cell partition of approximately 70°) and one concave (angle between side walls and cell partition of approximately 110°) side approximately 1 cm apart, as in natural brood cells. The orientation of the artificial brood cells in the room was altered randomly. An individual fly puparium was transferred from an O. bicornis nest to the artificial brood cell. The orientation of the puparium was chosen randomly. The behaviour of the newly-eclosed fly was videotaped under infrared illumination (Sylvania Mini-Kat Infrared Illuminator, 880 nm, completely dark for the human observer) using a monochrome CCD camera [Panasonic WV-CD20 (Panasonic Corp., Japan) equipped with a Cine-Tele-Xenar 1 : 2, 8/75 mm c-mount lens (Schneider-Kreuznach, Germany) and a Panasonic NV DV2000 digital video recorder]. The frequency of attempts to break through the convex and concave side of the brood cell was recorded for 1 h after eclosion of a fly. An attempt was defined as an approach to one side of the brood cell followed by behaviours with the possible function of breaking through the cell partition (such as holding, pushing, nibbling and scratching). If a fly left one side of the cell (no contact to the wall), approached it again and attempted to pass through the cell partition once more, this was counted as a new attempt.

Breaking through cell partitions

To detect possible mechanisms that might allow the flies to penetrate the brood cell walls, the behaviour of newly-eclosed flies in natural brood cells was observed under a stereo microscope. For documentation, the behaviour of the flies was video taped using a digital video camera (Panasonic NV-DA1 equipped with a 100 mm close-up lens).

Statistical analysis

Data are represented as the mean ± SD if the sample was normally distributed or as first, median and third quartiles if the distribution was not normal and could not be satisfactorily normalized by transformations. As a result of small sample sizes, nonparametric tests were used throughout. Calculations of statistical tests were conducted using spss, version 16.0 (SPSS Inc., Chicago, Illinois).

Results

Occurrence and location of C. indagator larvae

A total of 34 nests with 230 brood cells were investigated. Thirty-two nests contained one vestibulum, and nine nests also had a second vestibulum. The 189 cells that had been provisioned contained 104 bee cocoons and 311 larvae of C. indagator. In total, 60 brood cells were parasitized (32%) and 25 brood cells (13%) were destroyed by fungi or other unknown causes. In 28 nests (82% of all nests), at least one brood cell was parasitized. Individual brood cells from which no larvae had emigrated (see below) contained one to 11 fly larvae (first, median, third quartile: 2, 5, 6, n = 31). In most cases, infestation with C. indagator resulted in the death of the host larva. Both a bee cocoon and fly faeces were found in only ten brood cells; however, the number of fly larvae was low in these brood cells (1, 1, 2). The mean parasitism did not increase from the innermost to the outermost brood cell, the correlation was even negative, although not significantly so (Fig. 2, Spearman rank correlation: r_s = −0.33, n = 11, P = 0.325). In approximately one-third of the nests (11 out of the 34), the innermost brood cell was parasitized by C. indagator.

Figure 2. Proportion of brood cells parasitized as a function of brood cell number (1 = the innermost brood cell of a nest; values above data points indicate the number of the respective brood cells in the sample).

The location of the fly larvae relative to the parasitized brood cells in a nest (indicated by the filamentous faeces of the fly larvae) revealed that some fly larvae had moved from their feeding brood cell to cells closer to the nest entrance. Maggots had come forward to the entrance from at least eleven brood cells (18%) and, on average, had passed two cell partitions (first, median, third quartile: 1, 2, 2). No larva had moved towards the back of the nest. Thus, there is a significant difference in the number of brood cells from which larvae forwarded to the nest entrance as opposed to the rear of the nest (binomial test: P = 0.001). Small holes could be seen at the bottom of the cell partitions that maggots obviously had produced to pass through. Only 12 maggots (4%) from three nests had forwarded to the empty outermost brood cell and, in two of these nests, maggots had perforated the final closure. All these movements took place in early spring and maggots did not change their position again before pupation.

In 14 brood cells, a total of 98 fly larvae (31.5%) had no chance of being freed by a bee because there was no bee cocoon present in the same or a more basal brood cell. Moreover, no holes made by larvae were visible in the cell partitions. On average, these flies had to pass through 1.6 ± 0.85 (first, median, third quartile: 1, 1, 2, n = 14 brood cells) cell partitions to emerge from the nest or to reach a brood cell with partitions broken through by an emerging bee. Thus, approximately one-third of adult flies had to pass through at least one cell partition.

Timing of emergence of O. bicornis and C. indagator

Emergence of bees started on 12 March and ended on 5 May (Fig. 3). Male bees emerged earlier than females. Emergence of C. indagator started on 13 April and ended on 4 May. Flies emerged shortly after the peak emergence period of the host females. Thus, the emergence of the flies was delayed in comparison with their hosts and the partitions of 68.5% of brood cells that contained fly pupae were penetrated by emerging bees at that time.

Figure 3. Number of emerged male and female Osmia bicornis and individuals of Cacoxenus indagator as a function of calendar date.

Orientation during emergence

Newly-eclosed flies in the artificial brood cells appeared to ‘inspect’ both sides of the brood cell soon after eclosion. From 20 flies, 19 made most of their attempts (see below) at the convex cell partition (binomial test: P < 0.001). In natural nests, this is the side of the entrance. Flies made significantly more attempts at the convex side of the cell partition (39.6 ± 13.8 per h) than at the concave side (8.1 ± 5.7 per h; Wilcoxon matched pairs test: Z = 3.87, n = 20, P < 0.001). However, in this artificial situation, the cell partitions were too dense and the walls were probably too smooth to allow an individual fly to make its way through the barrier.

Breaking through cell partitions

Newly-eclosed flies in unmanipulated brood cells from trap nests (equipped with a transparent cover) approached both sides of the brood cell and located small crevices in the upper part of the cell partitions. Then, they pressed their body against the cell partition and their head into a crevice. With their bodies fixed in this position, they abruptly expanded their head by protruding their head blister (ptilinum; Fig. 4, see Supporting Information, Video S1) and broke away small pieces of the cell partition through this hydraulic pressure. Any given attempt lasted for approximately 0.5–30 s. Notably, the attempts were targeted at the upper side of the convex side of the brood cell wall. If several flies were present in one brood cell, they often ‘worked’ jointly at the same spot of the cell partition, probably increasing the probability that their combined attempts were successful. In six (out of 11) brood cells that were closely observed, flies were successful in piercing the cell partition. Passing through the small holes is accomplished by peristaltic compression of the still soft body (see Supporting Information, Video S1). Indeed, in unmanipulated trap nests with original cell partitions, all 98 flies located in brood cells that were not opened by a bee emerged successfully.

Figure 4. A newly-eclosed Cacoxenus indagator during an attempt to break through the brood cell wall (still image captured from a video). The fly presses its head into a crevice (A) and protrudes the ptilinum (B) to exert a hydraulic pressure to break away small particles and prepare a hole for emergence.

Discussion

Cacoxenus indagator flies have evolved a set of behavioural and physiological adaptations that enables them to emerge from O. bicornis nests. Some maggots penetrate cell partitions. Newly-eclosed flies try to break through the cell partitions by use of their ptilinum. Both larvae and flies are able to correctly orientate towards the entrance of the nest.

To reduce the probability that their progeny will become imprisoned in the innermost brood cells, brood parasites might be selected to preferentially oviposit in brood cells closer to the entrance of a nest, thereby making it easier for their progeny to emerge. Contradicting this assumption, there is no correlation between the location of a brood cell in a nest and the respective rate of parasitism by C. indagator in the present study. By contrast, Seidelmann (1999) provides evidence for an increase in the proportion of parasitism by C. indagator of 5.5% in the innermost to 10.4% in the outermost brood cells. Such an effect would reduce the problem but not entirely circumvent it. Moreover, a possible disadvantage of parasitizing innermost brood cells might be balanced by the larger amount of provisions available there because these mostly contain female progeny that receive more food (Krombein, 1967; Seidelmann, 1995). Furthermore, if egg availability is high and does not limit reproductive success in the parasitic flies, the loss of a certain fraction of eggs might not reduce fitness greatly (see discussions of egg versus time limitation in parasitoid wasps; Rosenheim, 1996, 1999; Heimpel et al., 1998).

Cacoxenus indagator larvae are obviously able to determine the side of the brood cell that brings them closer to the entrance. Newly-eclosed flies also direct their attempts to break through the cell partitions at the correct side of the brood cell, namely the convex side. The experimental evidence strongly suggests that at least adult flies (and probably also larvae) determine the correct side of the cell partitions similar to their host, as is suggested by Krombein (1967). Assuming that the shape of the cell partition is a signal that a mother bee passes to its progeny, the use of this information by C. indagator would represent an unwarranted exploitation of this signal. The orientation mechanism of the flies must then be hereditary and might be the result of a coevolutionary adaptation of the brood parasite to the cues available in its host's nest.

Alternatively, flies might simply make more attempts at the side providing an acute angle between cell partition and the side wall of the nest because only this side provides some opportunity to press the head into a crevice. However, although the angle between cell partition and side walls is less than 90° (approximately 70°), it is not sufficiently small to allow for a proper employment of the head blister. Thus, it is more likely that it is the shape of the cell partition and not the opportunity for employing the head blister that directs the flies' attempts to emerge. Regardless of the mechanism, the flies actually focus their attempts to the correct side of the brood cell.

Breaking through solid and rather thick cell partitions might appear to be difficult for small insects that lack strong mandibles. However, specialized brood parasites such as the bombyliid fly Anthrax anthrax produce pupae equipped with a crest on the head that enables them to open the tough host cocoon and break down the cell partitions in Osmia nests (Krunićet al., 2005). Fly maggots with their piercing mouthparts and flexible body can perforate even solid cell partitions. Maggots of parasitic miltogrammine flies might use this ability to break down cell partitions and destroy the content of other brood cells (Krombein, 1967; Cross et al., 1975). Indeed, a number of C. indagator maggots move to the entrance in some nests. Such behaviour is reported for the study species (Juillard, 1947, 1948), as well as for miltogrammine flies (Krombein, 1967). Coutin & de Chenon (1983) state that most larvae of C. indagator move forward to the outermost brood cell after the feeding period in July. By contrast, in the present study, most larvae have not entered the outermost brood cell at the end of February. Moreover, 31.5% are located in brood cells that would not be opened by emerging bees and that do not show emergence holes made by maggots. Perhaps the existence of holes in the cell partitions also has negative consequences for the flies. Larval or pupal parasitoids of C. indagator might also gain access to the nest and parasitize the flies. In any event, in the present study, approximately one-third of flies have to pass at least one cell partition after eclosion from the puparium.

In other Cyclorrhapha that pupate in the soil or leaf litter (e.g. Calliphora spp., Lucilia spp.), teneral adults use their ptilinum to make their way through the substrate during emergence (Cottrell, 1962; Reynolds, 1980). In these species, however, the substrate that teneral adults have to pass through is softer and more porous, and thus much easier to penetrate. Despite the firmness of cell partitions, adult C. indagator are able to break through these barriers by making use of their ptilinum, which exerts hydraulic pressure on particles of the brood cell wall to produce a small hole. Possibly, several individuals have to cooperate for a successful escape. Notably, the flies try to pass through at the upper part of the brood cell wall. This indicates that they are not primarily using holes that maggots might have prepared in the lower part of the cell partitions. That the flies have to pass through very small holes by peristaltic movements of their bodies shows that the cuticle has to remain soft until final emergence. Thus, mouthparts that might be considered to aid in the preparation of holes (Downing, 1995) are also not yet hardened yet are unsuitable for chewing the solid loam walls of O. bicornis nests.

Miltogrammine flies might also emerge from brood cells of mud daubers by the action of the ptilinum (Downing, 1995), possibly facilitated by moistening the cell walls (Alcock, 2000). Nevertheless, 12% of the parasitic miltogrammine flies are not able to make their way through the cell walls of their mud dauber hosts (Downing, 1995). Such a failure to emerge from brood cells of a certain a host species might be one reason for host specialization of some parasitic species. Indeed, C. indagator primarily parasitizes O. bicornis and the closely-related O. cornuta that share major characteristics of nest structure (Westrich, 1989).

In conclusion, cleptoparasitic C. indagator flies use a set of behavioural and physiological adaptations to emerge from the sealed brood cells of their host. Some of these adaptations already have an adaptive function in the predecessors of these flies (e.g. the use of the ptilinum to open the puparium). Others are probably novel adaptations to the specific problems of escaping from sealed brood cells (e.g. determining the direction of the entrance). These adaptations result in a considerable degree of specialization in the cleptoparasite C. indagator and enable them to successfully parasitize and emerge from closed brood cells in nests of the red Mason bee O. bicornis.

Acknowledgements

This study was partly supported by the Deutsche Forschungsgemeinschaft (DFG, SFB 554, TP B3).