INTRODUCTION

We present information about the immature stages and biology of the cleptoparasitic leafcutter bee Dioxys cincta (Jurine) resulting from two field trips that visited the same nesting site in northeastern Turkey in separate years. Discovery of some of the immature stages permits us to reevaluate and refine Micheli's (1936) and Rozen's (1967) interpretations of larval anatomy of the genus and to make comparisons with the immatures of the North American D. pomonae Cockerell (Rozen, 1967; Rozen and Favreau, 1967). Dioxys is Holarctic in distribution; all other genera of the Dioxyini are from the Old World (Michener, 2000).

The specimens for this study were excavated from a large, complex nesting site occupying a vertical bank at 22 km WSW of Oltu, Erzurum Province, Turkey, in June 2001 and again in July 2003 (fig. 1). The bank was part of an arroyo, which was nearly waterless at the time of our field studies. Numerous burrows of bees and wasps penetrated the face of the bank (fig. 2). The complexity of anastomosing fresh and old tunnels suggested that the embankment had served the nesting requirements of these animals over many years. Nonparasitic bee genera nesting there included, among others: Hylaeus (Colletidae); Osmia, Protosmia, Megachile (Megachilidae); and Xylocopa (Proxylocopa) and Anthophora (Apidae). Parasitic genera collected from the site included Coelioxys, Dioxys, and Stelis (Megachilidae: Megachilinae) and Melecta and Thyreus (Apidae: Melectini).

When Rozen (1967) described the immatures of Dioxys pomonae, he concluded that the species had four larval instars. He knew that an earlier cryptic stage was known for some bees and was termed the “first instar” by Hackwell and Stephen (1966). He looked for but did not detect such a stage in D. pomonae. Since then he has become aware that the first instar of many (but apparently not all) bees is inactive relative to subsequent instars and remains mostly surrounded by (i.e., pharate within) the egg chorion; in such cases the second instar is the first to become active (e.g., Hackwell and Stephen, 1966; Baker, 1971; Torchio et al., 1988: Torchio, 1989; Garófalo and Rozen, 2001). On reviewing pictures of the D. pomonae study, Rozen was convinced even before the present study that the apparent egg “just before” hatching in figure 5 of Rozen and Favreau (1967) was actually the first instar surrounded by the chorion and that the instars described by Rozen (1967) were actually the second, third, fourth, and fifth and not the first, second, third, and fourth. The original picture of the true first instar has been digitized, cropped, and reproduced here (fig. 4) to provide more details of the appearance of a pharate first instar. The picture of the egg from the former study (Rozen, 1967: fig. 4) is also shown (fig. 3) so that visual comparisons can be made between the two stages. In the following passages, we assume that all species of Dioxys have five instars, and we identify them accordingly.

EGG AND FIRST LARVAL INSTAR

Presumably entangled in the outer surface of the cocoon, a shed chorion of Dioxys cincta with its nodular dorsal chorion3 intact (see Rozen and Özbek [2003] for a description of the unusual egg/mature oocyte of this species) was preserved with a female pupa that had been accidentally killed during our excavation. When we examined the chorion in the laboratory, we discovered the cast skin of the first instar within it. The skin, though colorless and nearly featureless, could be identified by the crossed mandibles (fig. 5), which were slender, slightly curved, and tapered gradually to a single point, unlike the bidentate mandibular apices of the presumed first instars of Stelis chlorocyanea Cockerell (Rust and Thorp, 1973) and S. elongativentris Parker (Rozen, 1987). This discovery confirms the conviction (expressed above) that Dioxys has a pharate first instar within the egg chorion and that the second instar is the one that first becomes active. Other identifiable features of the exuviae, seen through a compound microscope, were the salivary opening below the mandibles and the spiracles. A linear string of granules (spicules?) ran between the spiracles on each side of the body, as was noted by Alves dos Santos et al. (2002) on the first instar of Tetrapedia diversipes Klug (Apidae: Apinae: Tetrapediini). Because such linear series of granules are found in separate families, they are probably a widespread phenomenon among first instars of bees. Their function is unknown but may be associated with the rupturing of the chorion during the early stages of eclosion, as the chorion of some bees is known to split along the spiracular line on each side (Rozen, 1964: fig. 5; Alves dos Santos et al., 2002).

The hatched egg (fig. 6) was 1.63 mm long and 0.88 mm in maximum width. The eclosion hole was 0.38 mm from the anterior end, with the latter being identified by the micropyle occurring on a shred of the thin ventral chorion. The hole (fig. 6) was off-center, more-or-less transverse, and slightly askew. Its maximum width and height were 0.25 and 0.11 mm, respectively. Its borders were somewhat irregular and the chorion around it was milky (not visible in fig. 6 because of backlighting), contrasting with the clearer chorion elsewhere. Possibly the milky area was created by an eclosion fluid that modified the texture of the chorion. How the second instar penetrates the thick chorion is unknown, but the irregular nature of the opening suggests that the second instar might chew its way out. If this is true, then this is an unusual form of eclosion for a cleptoparasitic bee larva; most have chorions that split along defined lines (e.g., Rozen et al., 1978; Rozen, 1994, 2003).

By dividing the length (1.63 mm) of the egg by the maximum distance (3.60 mm) between the outer rims of the pupal tegulae, we calculated an egg index of 0.45. The index is a system of defining the size of a bee's egg relative to the size of the female's body, as developed by Iwata and Sakagami (1966). The egg index in this case is somewhat greater than 0.39, which Rozen and Özbek (2003) calculated for the same species using mature oocytes, but both values fall within the “dwarf” category defined by Iwata and Sakagami (1966: table 2).

FOURTH LARVAL INSTAR

The recovery of a fourth larval instar of Dioxys cincta makes possible its description and a comparison with the cast skin of the last (fifth) larval instar. Its assignment to the fourth larval stage is based on its large size and on the fact that the relatively short antenna of the last instar and its bidentate mandibles were visible within the cleared head capsule.

The single specimen was first studied and drawn before the head capsule was cleared in an aqueous solution of sodium hydroxide. The capsule and the sclerotized portions of the mouthparts were darkly pigmented in certain areas. After clearing, much of the pigmentation was lost, suggesting that dark pigments are subcuticular. Although sclerotized areas tend to be faintly darker than nonsclerotized areas on the cleared head, labral and maxillary sclerites, which were darkly pigmented on uncleared specimens, were rendered nearly colorless on cleared preparations. Cast larval skins are shed with the dark pigments in place, as evidenced by examination of two cast fifth-instar exuviae of this species and a single cast skin of a third instar of Dioxys pomonae from an earlier study by Rozen (1967). Thus, in the following description of the head, descriptions of dark pigmentation refer to the uncleared specimen. The dark pigment, while presumably noncuticular, indicates areas of sclerotization.

The specimen described was nearing ecdysis when preserved. The unexpected loss of pigmentation during the clearing treatment, the obfuscation caused by the developing fifth-instar cuticle within the cleared head, and our lack of making detailed sketches of pigmentation and sclerites before treatment combine to make certain features in the following description and accompanying illustrations uncertain, as indicated.

FIFTH LARVAL INSTAR

The last larval instar of Dioxys cincta was described and illustrated by Micheli (1936). Recovery of the cast larval skins associated with the pupae allows us to (1) add details to augment his description of the head, (2) compare it with the fourth instar, described above, and (3) compare it with the fifth instar of other species of the genus. Although description of the pigmentation is based on that of the untreated cast skin, one of the cast skins was boiled in an aqueous solution of sodium hydroxide. Unpredictably, no diminution of pigmentation was detected in contrast to the treatment of the head capsule of the fourth instar. The labral sclerite, maxillary sclerites, and internal head ridge remained dark.

COCOON

We retrieved two cocoons of this species from different nests. In neither case was the host identified. One cocoon contained a pupa, which was accidentally killed during our excavation; the other held a postdefecating larva that pupated the day after discovery. This pupa was studied and illustrated while alive, and the species identification was verified after the adult emerged. The cast larval skins associated with both pupae are described above; the pupa is described below.

The two cocoons were of pale cream color, contrasting with the dark brown fecal pellets that adhered loosely to their fuzzy exterior surface. The cocoons did not cling closely to the cell walls, as they could easily be removed from the substrate. No fecal material was incorporated in the cocoon fabric itself, and no feces occurred in the interior of the cocoon, indicating that defecation takes place prior to spinning. The cocoon fabric was semitransparent and soft (nonrigid) with numerous silk threads visible on the outer surface, which therefore appeared dull. The inner surface was somewhat shinier because the silk strands fused with one another there. One cocoon measured 11 mm long and 8.5 mm in maximum diameter. The front end was somewhat more pointed than the rear but was not as acutely pointed as the cocoon pictured by Micheli (1936: fig. 6). The fabric was thicker and denser at the pointed end than elsewhere, an indication that the pointed end was a weak nipple, a specialized feature seen on other megachilid cocoons. The variation in the distinctiveness of the nipple may depend on the size of the cell in which the cocoon is spun.

The cocoon of the North American Dioxys pomonae described by Rozen and Favreau (1967) contrasted with that of D. cincta described here. Not only did it have a more distinct nipple anteriorly, but it also was dark in tone, rigid, and consisted of three layers: an outer more fibrous one, a middle layer of dark feces (accounting for its overall dark appearance and rigidity), and an inner one that was “loose and light brown”, forming “a cellophane-like coating even though some individual silk strands were detected.” Thus the differences in the cocoons of the two species can be explained by the behavioral trait of whether the larva defecates before cocoon spinning (D. cincta) or whether feces are sandwiched between two layers of silk (D. pomonae).

PUPA

The pupa of Dioxys cincta (fig. 15) is nearly identical to that of D. pomonae (Rozen, 1967) except for size. The median metanotal tubercle on pupae of these two species, accommodating the development of the same feature on adults, permits easy identification of the pupae to genus. The pupa of D. cincta will readily key correctly to genus in Rozen (2000). A formal description is unnecessary here because the pupal description of D. pomonae given by Rozen (1967) applies equally to that of D. cincta except that setae are found on the vertex of the latter (fig. 15).

DISCUSSION

We interpret both the swollen parietals and the strong constriction of the posterior part of the head capsule of the fourth larval instar to be adaptations providing a stable platform for the strong mandibular musculature, with the latter being necessary for attacking host immatures and those of other cleptoparasites. The enlargement and fusion of the clearly defined cardines with the parietals are similarly interpreted as a ventromesal extension that adds firmness to the head capsule. This combined enlargement of the cardines and their fusion to the head capsule is an unusual feature in bee larvae. It is quite visible in the cast head capsule of a third instar of D. pomonae that was attached to the fourth instar described (as a third instar) by Rozen (1967). The fourth instar from that study was cleared with an aqueous solution of sodium hydroxide so that its pigmentation is no longer visible, but the cast skin of the third instar remained untreated in the collection of the American Museum of Natural History. Its large fused cardo is similar to that described here for the fourth instar of D. cincta, an indication that third instars of at least these two species are equipped to attack.

It seems likely that in all species of Dioxys, the second, third, and fourth instars will exhibit this fusion since mandibles of these instars of D. pomonae in Rozen (1967: figs. 21–23, 27, 28) and the fourth instar of D. cincta pictured here (figs. 10, 11) are massive, elongate, strongly curved, and sharply pointed, and are obviously weapons of attack against host or competing cleptoparasite immatures. The fact that the cardo of the fifth instar of D. cincta, as indicated above, articulates with the rear of the hypostomal ridge suggests that the last larval instar is not hospicidal, as do its apically bidentate mandibles, with the upper tooth only slightly shorter than the rather blunt ventral one.

It is instructive to compare these larval adaptations for cleptoparasitism with those of other cleptoparasitic Megachilidae. Larvae of other Dioxyini, all of which are parasitic, are unknown. Distantly related parasitic clades in the Megachilinae whose larvae have been described show certain similar (but nonhomologous) adaptations. Thus, the second and third instars of Coelioxys have strongly sclerotized head capsules that are bridged ventrally by a nearly complete sclerotization of all but the apex of the labiomaxillary region. This construction provides a firm platform for the massive mandibular musculature required for the huge mandibles (Baker, 1971). The ventral sclerotization of the labiomaxillary region can also be interpreted as armament, shielding this part of the head from attack. The foramina magna of Coelioxys are constricted, further strengthening the head. Baker pointed out that the third instar is the stage that usually kills the host egg or young larvae. Subsequent instars have a more normal anatomy and are presumably nonhospicidal.

Torchio (1989) published an excellent account of the development of Stelis montana Cresson that clearly showed that the species has five larval instars, with the first instar being pharate within the egg chorion and with the fifth instar alone being hospicidal. Many earlier published observations regarding Stelis must be cautiously interpreted because it has not been generally recognized that first instars of many bees are inactive and pharate within their egg chorions. None of these earlier accounts recorded the total number of instars, and thus the true first instars may have been overlooked.

Michener (1955, and references therein) concluded several species of Stelis may battle immatures of host or other cleptoparasites when the Stelis are partly grown. He observed the mature larva of S. lateralis Cresson killing the host. He reported that the mandibles of the species are “acutely pointed … throughout larval life”, but it is unclear that he actually observed the true first instar. Rust and Thorp (1973) concluded that the supposed first instar of S. chlorocyanea Cockerell “usually does not destroy the host, but later forms do.” Although they term the mandible of this instar as bidentate, their illustrations (1973: figs. 10, 11) depicted it as sharply pointed with one tooth much shorter than the other, and that they observed one “first instar” feeding on the egg of the host (p. 554). They reported that the “second instar” has the upper tooth reduced to an obtuse angle and subsequent instars have elongate, curved mandibles with acute apices. We think that they may have actually seen the true first-instar as they watched late embryogenesis and may have overlooked the simultaneous shedding of the chorion and first-instar cuticle, just as Rozen and Favreau (1967) had overlooked this phenomenon in Dioxys pomonae.

Rozen (1987) found that the supposed first instar of Stelis elongativentris Parker, “still partly in its chorion”, had bidentate mandibles that probably would be incapable of killing the host egg. Although this larva may have been a true first instar, it was also reported to have consumed some pollen, which has not been an activity of known pharate first instars. The mandible of the last instar is also bidentate, but the dorsal tooth is small and far removed from the sharply pointed apex of the robust ventral tooth. This feature and the enlarged, heavily sclerotized parietals strongly suggest that the last instar is equipped to battle host larvae and perhaps competing larval cleptoparasites.

In contrast to these megachilid cleptoparasites, the last instar (as well as the supposed first instar) of Hoplostelis bilineolata (Spinola) has bidentate, apically broad mandibles, and the parietals are not swollen and heavily sclerotized (Rozen, 1966). Also in contrast to the other megachilid cleptoparasites, the adult female H. bilineolata is known to kill the host egg (or young larva) (Bennett, 1966). As implied by Michener (2000), the differences in the mode of parasitism between Hoplostelis and Stelis suggest that these genera did not have a common parasitic ancestor.

An interesting feature of many cleptoparasitic megachilids whose larvae are hospicidal is that a sequential series of instars (not just a single instar) may be capable of dealing with host immatures or competing parasitic larvae, although Stelis montana (Torchio, 1989) and perhaps other Stelis species are exceptions. On the basis of limited data, we venture the following preliminary conclusions: the second, third, and fourth instars of Dioxys exhibit hospicidal adaptations; the second and third instars of Coelioxys are hospicidal, but with Stelis, depending on the species, all larval instars may be hospicidal (although we strongly suspect that the first instar needs reexamination) or only one, the last instar, is hospicidal. These observations contrast with what we know about the cleptoparasitic lineages of the Apidae. Only one instar, the first, is hospicidal in the Nomadinae, Protepeolini, Isepeolini, Melectini, Tetrapediini (Coelioxoides), Rhathymini, and Ericrocidini (Rozen, 1991; Alves dos Santos et al., 2002). In the case of Exaerete, only one instar, the second, may be hospicidal, although the modifications of this instar may have been driven by competition with other cleptoparasite immatures since the cleptoparasite mother attempts to eliminate the host eggs when she oviposits (Garófalo and Rozen, 2001, and references therein). It will be interesting to see if future studies substantiate this difference between cleptoparasitic Megachilidae and Apidae.