Nesting

To study nest architecture, we excavated several nests at both sites in the Zonda Valley. The nests were always grouped near one another in horizontal areas that were frequently exposed to the sun. Distances between neighboring nest entrances were 20 cm or less. The nest entrance, a circular hole 5.0–5.2 mm in diameter, remained open and lacked a tumulus. The main tunnel descended more or less obliquely, running to a depth of about 15–20 cm, where 5–6 brood cells were found per nest. Cells were oriented with their long axis titled about 45° to the rear. It was not possible to observe lateral tunnels leading to cells, as they were apparently soil filled.

We collected a total of 137 cells. Of these, 97 contained eggs and larvae in various developmental stages. Twenty-two were filled with soil and likely represented chambers from which the brood had emerged. In the lab we verified that 51 cells contained Canephorula larvae and 46 held larvae of Melectoides Only 3 cells contained Canephorula imagos (1 female and 2 males). Theremaining 15 cells were maintained in a terrarium to follow developmental periodicity.

Cells were oval in shape. Their inside dimensions ranged from 11.0 to 14.0 mm in length and 6.5 to 7.5 mm in maximum diameter (N = 15). Cell walls were 0.6–1.0 mm thick, composed of soil more compact than the substrate, and cemented with a substance of unknown origin (fig. 5). The inner surface of cells, in contrast with the outer layer, was very compact and polished, although the soil particles that composed it were the same as those of the rest of the wall, and lined with a delicate, transparent pellicule (insoluble in ether), detachable from the substrate (fig. 5). In one freshly built cell the food consisted of a small orange (gruellike) mass of wet pollen (presumably incomplete), occupying about one-fifth of the cell volume (fig. 5). Eggs were on the top of provisions.

The cocoon of Canephorula (fig. 6), except for the closure end, conformed exactly to the shape of the cell, and the clear, colorless lining tended to adhere to specimens removed from the substrate. Except at the closure end, cocoons preserved in ethanol seemed to consist of two transparent layers of amber, cellophane-like material between which were sandwiched an opaque, rather even but thicker layer of feces (fig. 7). (On a dry specimen this layer was about 0.06 mm thick at the cell bottom.) The exterior transparent layer consisted of an outer colorless layer and an inner amber-colored layer that adhered completely to one another (although easily separated with fine forceps) and that, when viewed under high magnification, revealed traces of fibers even though the silk was sheetlike (i.e., without fenestrations). Similarly, the inner transparent layer consisted of two completely distinct adhering layers, the one next to the feces being amber, while the one next to the larva was colorless. Possibly the amber hue was derived from the feces. Fecal material was finely granular. When preserved in ethanol, it was medium brown, but, when dry, mottled dark brown and black when viewed from outside. Under high magnification, some pollen exines could be identified although most were crushed and distorted.

The closure end of the cocoon (figs. 6–8) had a very different appearance from the rest of the cocoon. Instead of being evenly curved like the bottom of the cell, it was more or less conical when seen from the side (fig. 6), angling abruptly where it met the rest of the cocoon. (This angle hereafter is called the periphery of the closure end.) It consisted of several fragile, transparent, cellophanelike amber layers, which were mostly, but not completely, separated from one another except at the periphery and at the apex of the cone (the center of the closure end). These layers showed some threadlike strands connecting them. The layers are almost certainly extensions of the silk sheets from the rest of the cocoon although there may have been additional sheets deposited as well. In addition to these thin sheets there was a much thicker, leathery inner layer (figs. 7, 8) consisting of a number of layers of transparent silk with smears of feces in between. Some places had been left uncoated and therefore remained transparent. The feces in dried cocoons appeared almost black when viewed from inside the cocoon, the inner surface of which was shiny. The dark-brown fecal material was semitransparent (not nearly opaque like elsewhere in the cocoon) and may have consisted mostly of metabolites rather than pollen exines. It is uncertain whether or not the fecal material was sandwiched between double layers of sheetlike silk as found elsewhere in the cocoon. The leathery inner layer had several distinctive features. At its center and at a right angle to the long axis of the cocoon, there was a flat, fibrous disk (figs. 7, 10, 11) (here termed “filter”), 1.5–2.0 mm in diameter (N = 9), that was pale, nonreflective, yellowish-tan rather than the essentially black, highly reflective surface of the surrounding inner surface of the closure end. In cross section (fig. 11), this disk was composed of numerous threadlike strands of silk and may function as a filter, permitting gas exchange while excluding parasites (Rozen and Buchmann, 1990).

On the outer surface of the leathery layer (fig. 8), a series of approximately 12 wrinkled ridges radiated from the level of the filter toward the periphery of the closure end, thus dividing the closure end into a number of shallowly concave facets. Although it is unclear how the ridges were created, they may be an artifact resulting from the application of the feces to the outside silk layer of the leathery layer. This application might have caused a shrinking and weighing down of the silk layer, which probably would have been domelike when first manufactured. Although the inside silk layer had indistinct facets, the fact that it did not show the fine wrinkles as did the outer layer seems to support this hypothesis. Examination of cocoons while being built would almost certainly elucidate the construction mechanism for this peculiar structure.

Miliczky (1985) and Rozen (1991a) have provided comparative descriptions of the cocoons of other genera of eucerine bees based on original observations and published accounts. The cocoon of Canephorula apiformis shows several unusual features when compared with those of other genera. Unlike in other genera, the feces of C. apiformis are incorporated in the entire cocoon, whereas in other known genera the feces are placed mostly outside the cocoon, just below the cell closure, so that they form a plug, although they sometimes extend partway down the cell wall, as in Eucera (Synhalonia) (called Tetralonia by Miliczky, 1985, and references therein). In Eucera (Synhalonia) defecation is completed before silk production commences, as indicated by the fecal placement exterior to the cocoon. In Idiomelissodes silk is incorporated in feces, although most of the cocoon is produced after defecation (Rozen, 1991a). The sandwiching of feces of Canephorula between two layers of silk implies that the larva first spins the outer layer(s), stops spinning, and defecates, and then completes the cocoon by spinning the inner silken layer.

Cocoons of all other eucerines are rather flexible and semitransparent, so that the larva or pupa can be seen vaguely through the cocoon wall. The fecal layer renders the Canephorula cocoon rigid and opaque from the outside.

The closure end of the inner, leathery layers of the Canephorula cocoon is unique in that radiating ridges extending from the level of the filter to the periphery have no counterpart among other eucerine genera whose cocoons have been studied. The small diameter of the flat fibrous disk of this genus is unusual but not unique; cocoons built by species of Florilegus sp., Idiomelissodes and Eucera (Synhalonia) have a small disk diameter (compared with periphery of the closure end), while cocoons of Thygater analis (Lepeletier), Eucera sociabilis (Smith), and Svastra obliqua obliqua (Say) have a filter that is the same diameter as the periphery of the closure end (all cocoons preserved in the AMNH).

The cocoon of Melectoides bellus was brown and substantially different in shape (fig. 9) from that of the host. It was shorter compared with its maximum diameter, and its closure end consisted of a low, lidlike dome. Externally, the dome (figs. 12, 14) was composed of coarse silk fibers and was roughly 1 mm thick at the periphery. In cross section (fig. 13), the dome was composed of numerous, essentially parallel sheets of woven silk. The intermediate sheets could be pulled apart by forceps to reveal silk fibers finer than those of the external surface. The innermost layer was finely textured, without distinct threads of silk but with fenestrations, particularly around the periphery. Presumably these fenestrations and coarser cocoon fabrics permit gas exchange while excluding parasites.

Fecal material adhered to the external surface of the dome (fig. 14) and was also found between its layers of silk, indicating that the cocoon spinning and defecation overlapped in time. It also suggests that the dome is spun before the rest of the cocoon, since feces are not incorporated into the cocoon wall elsewhere.

The rest of the cocoon conformed to the shape of the lower part of the cell. The cocoon wall was externally smooth, much thinner than the closure end, leathery-tough (difficult to tear with forceps), and, on a dried specimen, single layered. On material examined in alcohol, several layers of nearly clear fabric with identifiable silk threads could be separated from the leathery exterior, suggesting that the cocoon wall was composed of numerous sheets of silk. The exterior layers were fused to one another and lacked silk threads, while the interior layers exhibited silk threads. The lower part of the cocoon was slightly transparent, revealing the vague shape of the larva.

All three cocoons of Melectoides bellus available for study displayed an elongate, scarlike deformity, about 1.5 mm long, on the cocoon wall (fig. 9). The scar was raised externally and also internally, and the cocoon fabric in the vicinity of the deformity was somewhat darker and thicker. A cross section of one scar revealed that it consisted of layers of silk. These facts suggest that the larva may have applied an abundance of silk to an elongate pit in the cell wall. This feature was consistent in all three cocoons; we hypothesize that this might be an egg-insertion pit (as found in most Nomadinae) of the cleptoparasite. Further investigation might illuminate why the pit is so thickly sealed from the lumen. At this time, we do not know how females of the Isepeolini introduce their eggs into host cells, nor have any of the eggs been described.

Michener (1957) accurately described, but did not illustrate, the cocoon of Isepeolus viperinus (Holmberg), representing the only other genus in the Isepeolini. Because he preserved a specimen, additional details can now be added and comparisons made with the cocoon of Melectoides bellus The cocoon of I. viperinus is an oblong ovoid, conforming to the shape of cell of its host (Colletes). Hence, its shape is quite different from the truncated hemi-ovoid of the cocoon of M. bellus Another remarkable difference between the two cocoons is in their external structure. The dome of M. bellus is constructed with a thick, opaque fabric of partly fused, coarse silk strands, while the remainder of the cocoon is smooth and semitransparent. In contrast, the entire external surface of the cocoon of I. viperinus is covered by a thick, opaque fabric of partly fused coarse silk (fig. 15). However, these external fabrics of the two genera appear to be identical when viewed as SEM micrographs (figs. 12, 15). Both (figs. 13, 16) are composed of numerous layers of silk; an external layer is constructed of course fibers, while fine silk is used to spin more internal layers. The innermost layer of I. viperinus is shinier than that of M. bellus However, as noted by Michener, the lining at the closure end of the Isepeolus cocoon (fig. 16) is more fibrous, supporting the idea that gas exchange may take place in this region. Michener noted fecal material on the outside of the cocoon of I. viperinus; as on M. bellus fecal material was also dispersed among the silken sheets of the cocoon of I. viperinus indicating that both bees defecate while cocoon-spinning.

The cocoons of these two related cleptoparasitic genera display great differences in shape and macrostructure, yet share tremendous similarities in microstructure.

Life Cycles

A total of 275 adults of Canephorula apiformis were collected: 257 (211 males and 46 females) in flowers and 18 (15 females and 3 males) near nest entrances or in nests. During the first survey period (1993–1994), the bees were active from November to February (fig. 17); however, in additional samplings (1996), 1 male and 1 female were recorded in early March. In 1996–1997 theactivity of the bees in the study site was restricted to November and December (fig. 17). The absence of activity in January and February (1997) was probably related to disturbance caused by caterpillar tractor activities on the nest aggregations.

The observations made at nest aggregations during 1993–1994, 1995–1996, and 1996–1997 revealed that the bees emerged in mid-November until early December; in December females were seen provisioning the nests. In January the activities practically ceased, with only a few bees flying over the nests. In the beginning of February, activity became intense again, and young adults were seen leaving the nests. In early March, activity completely ceased. This pattern, with two activity peaks, one in November–December and the other in February, with bees newly emerged in the beginning of each, indicates that Canephorula apiformis may be a bivoltine species, as the diagram of flower-visit frequencies (fig. 17) also suggests. On the other hand, the stage of development of the immatures obtained from the nest excavations is more indicative of only one generation per cycle. In the excavations at the end of November when many newly emerged bees were seen, only three cells were found that contained pupae from the emerging population. In early December only eggs and small larvae were found. In mid-December there were large larvae. In January, no excavation was made, but only mature larvae were found in excavations made in early to mid-February. The mature larvae observed in February entered diapause, remained in this state during fall, winter, and part of spring, and emerged in November–December. In February 1995, closed cells (N = 15) were collected and maintained in a terrarium. The individuals in these cells remained in diapause until the end of November, when the first adults began to emerge in synchrony with the emergence of adults at the nesting sites. Of the 15 cells, 8 individuals (5 males and 3 females) completed development and emerged.

A total of 180 adults (171 males and 9 females) of Melectoides bellus was collected on flowers (fig. 17) and 2 additional females were captured around the nests of Canephorula. Melectoides was active from November to the beginning of February, overlapping with the entire activity period of its host (fig. 17), and appears to have one generation per year. December was the peak of activity, with 133 cleptoparasites collected.

Floral Preference and Phylogeny

Figure 18 shows the flowering periods and flower visitation frequency of Canephorula apiformis The plant species visited were: Prosopis strombulifera, Atamisquea emarginata, Tessaria absinthioides, Baccharis salicifolia and Larrea divaricata. Tessaria absinthioides was visited by the greatest number of individuals, with 83 males and 35 females collected. In descending order of visitation frequency were A. emarginata with 67 males and 9 females; B. salicifolia with 35 males and 2 females; L. divaricata with 23 males; and P. strombulifera with 3 males (fig. 18). The individuals of Canephorula apiformis were active from 8:00 hr to 20:00 hr, mostly between 10:00 hr and 14:00 hr (fig. 20).

The majority of the females (35 of the 46 collected on flowers) were collected on Tessaria absinthioides However, the pollen load carried by 22 of the 46 females was exclusively composed of pollen from Atamisquea emarginata In 6, the load was a mixture of pollen predominantly from A. emarginata and in lesser quantity from T. absinthioides One female had pollen only from Baccharis salicifolia Of all the females that visited T. absinthioides only 4 had a few grains of pollen from this species. Since the amount of pollen from T. absinthioides was small, we consider it a contaminant.

Therefore, at least during November and December, our data indicate that the main source of pollen at the study sites was Atamisquea This floral species blooms only until the beginning of January; during January and February another plant should substitute for it. Likely candidates are Baccharis salicifolia which blooms until the end of the season, or Larrea whose flowering season overlaps that of Baccharis during February and March (fig. 18). During this period only 8 females were collected (6 at flowers of Tessaria and 2 at flowers of Baccharis), but they were not carrying pollen loads except for 1 female with some pollen grains of Baccharis and 4 with some pollen grains of Tessaria

Jensen-Haarup (1908a, 1908b) and Cocucci et al. (1992) provided records of Canephorula visiting some caesalps in Argentina. According to Simpson and Neff (1985), Canephorula should be restricted to caesalpinoid types of Fabaceae (e.g., Cercidium, Caesalpinia and Zuccagnia). Although Cercidium praecox (Ruiz and Pav.) Harms and Zuccagnia punctata were present in the sampled areas and other species of bees were observed at flowers of C. praecox during this study, no males or females of Canephorula were found visiting them.

The full pollen load of Canephorula apiformis is typically large, covering the entire anterior surface of the hind tibia and basitarsus.

The flowers most commonly visited by Melectoides bellus were Tessaria absinthioides (129 individuals), Prosopis strombulifera (43 individuals), and Atamisquea emarginata (8 individuals) (fig. 19). This bee was active mostly between 8:00 and 18:00 hr (fig. 21).

Postdefecating Larva of CANEPHORULA APIFORMIS (Friese) Figures 22–28

Diagnosis: Mature larvae of Eucerini are quite similar (for references to descriptions, see McGinley, 1989; Rozen, 1991a), and the larva of Canephorula apiformis conforms well to the tribal description of larvae presented by Rozen (1965). The significant tribal features are in boldface below. Characters in italics may prove useful for distinguishing the larva of this genus from those of other eucerine genera.

Description: Length (if straight) approximately 20 mm.

Head (figs. 23, 24): Integument unpigmented except for following areas: hypostomal ridges, anterior mandibular articulations, apices and extreme base of mandibles, base of cardines, stipites toward distal ends, and articulating arms of stipes; cranium with widely scattered, small setiform sensilla; labral tubercle, apex of labrum, maxillae (particularly apically), and apex of labrum with larger setiform sensilla; following areas with conspicuous elongate setiform spicules: surface of labrum between tubercles (fig. 24), epipharynx on either side of midline (above similar patches on closed mandibles), dorsal surface of mandibles (fig. 25).

Head size moderately small compared with body; head capsule wider than long in frontal view. Tentorium complete (including dorsal arms); anterior arms robust; rest of tentorium moderately developed; anterior tentorial pit much closer to anterior mandibular articulation than to antenna; posterior tentorial pit conspicuous, well impressed, and found at junction of hypostomal ridge and postoccipital ridge. Median longitudinal thickening of head capsule well developed to level of antenna. Postoccipital ridge moderately well developed; hypostomal ridge, pleurostomal ridge, and lateral arms of epistomal ridge well developed; median section of epistomal ridge faintly developed, far less so than lateral arms. Parietal band evident. Antennal prominence scarcely developed; antennal disk differentiated from papilla; antennal papilla moderately projecting (height less than diameter), with approximately three sensilla. Front of head capsule as seen in lateral view (fig. 23) sloping normally so that labrum extending well beyond clypeus and clypeus beyond frons. Labrum broad as seen in frontal view, apically trilobed, with pair of low tubercles apicolaterally; labral sclerite (as found in Megachilidae, McGinley and Rozen, 1987: fig. 27) not evident; epipharynx apparently a simple, slightly curved surface.

Mandible (figs. 25–27) robust, moderately long so that, in repose, mandibular apices meet; outer surface with several elongate setae on small tubercles; mandibular apex bidentate with dorsal tooth longer and broader; both teeth acutely pointed but their extreme apices rather blunt; dorsal adoral surface of mandibles separated from apical concavity by distinct ridge that more or less parallels dorsal apical edge; dorsal adoral surface with irregularities toward base but not distinctly denticulate as in Idiomelissodes duplocincta (Cockerell) (Rozen, 1991a), smooth toward apex; cusp and apical concavity well developed with irregular surface near cusp. Labiomaxillary region strongly projecting. Maxillary apex distinct, bearing elongate palpus apically, inner apical surface not strongly produced and with integument irregular; galea represented by very low mound bearing cluster of approximately three elongate setae between palpal base and inner apex; cardo and stipes well developed; articulating arm of stipes evident. Labium divided into prementum and postmentum; premental sclerite not developed; apex of labium with surface irregular; labial palpus elongate, nearly as large as maxillary palpus. Salivary opening a transverse slit on projecting lips, the opening as in most other eucerines extremely broad, as wide as distance between labial palpi; hypopharyngeal groove not incised and therefore not evident. Hypopharynx a flat continuation of dorsum of labrum to buccal cavity.

Body (fig. 22): Integument without setae except for very fine setiform sensilla on abdominal segment X, mostly below anus; integument very finely, evenly spiculate in many areas; integument without spines or sclerotized tubercles. Body form robust; intersegmental lines moderately defined; intrasegmental lines evident on most body segments although sometimes partly obscured by bending of body; paired dorsal tubercles present on caudal annulet of each thoracic segment; these tubercles transverse, only briefly separated dorsally on each segment; abdominal segments without dorsal tubercles (however, indefinite paired dorsal tubercles on anterior abdominal segments appearing on immature larvae); lateral tubercles (below level of spiracles) absent; abdominal segment X in lateral view (fig. 22) attached dorsally to segment IX; anus dorsal in position on X as seen in lateral view; perianal area with strong transverse ridges above anus. Spiracles (figs. 22, 28) moderately large, subequal in size; division between atrium and subatrium uncertain because primary spiracular opening apparently without collar and because of presence of irregularly formed chamber intermediate in size between surface chamber and subatrium chambers (this intermediate chamber possibly subdivision of atrium; see Remarks below); atrium projecting slightly beyond body wall, with rim; subatrium short, with approximately 5–6 chambers; atrial and subatrial walls without denticles. Female with pair of imaginal disks ventrally on each of segments VII, VIII, and IX, those of IX being contiguous and preceding ones successively farther apart; cuticle associated with each disk slightly impressed. Male sex characters unknown.

Material Studied: Two postdefecating larvae, Zonda, San Juan Province, Argentina, December 17, 1995 (E. Michelette).

Remarks: Spiracular morphology of mature eucerine larvae does not conform to that of most bee larvae, in which the spiracular apparatus consists of a large atrium subtended by a narrow, multichambered subatrium (Michener, 1953: fig. 15). Instead, larval representatives of Svastra, Xenoglossa, Melissodes, Peponapis, Thygater and apparently Eucera (Synhalonia) (Miliczky, 1985; Packer, 1987; Rozen, 1965) appear to have the subatrium divided into two parts, termed (in boldface) by Rozen (1965: fig. 5) as the more transparent and wider outer subatrium and the optically denser and narrower inner subatrium. Rozen (1991a) considered the subatrium of Idiomelissodes to consist of only the three chambers, but reexamination of the specimen reveals a short, optically dense area immediately below the atrium. The spiracle of this genus needs further study to determine if the denser area may be homologous to the inner subatrium of the other genera. The spiracle of Canephorula (fig. 28), described above, is clearly a departure from the spiracles of the other genera in that it lacks an optically dense inner subatrium. However, the intermediate chamber between the surface chamber (fig. 28) and the narrower inner subatrium is ambiguous in that it can be interpreted either as a single-chambered outer subatrium (with the inner subatrium being the remaining narrow chambers) or a subdivision of the atrium.

Postdefecating Larva of MELECTOIDES BELLUS (Jörgensen)Figures 29–34

The Isepeolini are a small group of South American cleptoparasitic bees comprising two genera (Isepeolus and Melectoides) and 21 species according to Roig-Alsina (1991). Immatures of only the following two species have been described previously: Isepeolus luctuosus (Spinola) (Claude-Joseph, 1926 [mature larva]) and I. viperinus (Holmberg) (Michener, 1957 [first instar, pupa]; Lucas de Oliveira, 1966 [first and last larval instars]; Rozen, 1966 [last larval instar]; Rozen, 1991b [first instar]).

Diagnosis: The mature larvae of Melectoides bellus, Isepeolus luctuosus and I. viperinus can easily be separated from other bee larvae by their labrum bearing a single median tubercle and by their more-or-less evident lateral body tubercles in addition to the paired dorsal tubercles on most body segments. Melectoides bellus and I. viperinus can be separated from one another by the features presented in italics below. (These features are unknown for I. luctuosus) Characters shared by M. bellus and I. viperinus are printed in boldface; presumably they are the main diagnostic features of mature larvae of the Isepeolini.

Description: Length approximately 10 mm.

Head (figs. 28, 29): Integument very faintly pigmented except darkly pigmented in following areas: tentorium in vicinity of anterior and posterior tentorial pits, hypostomal, pleurostomal, and lateral segments of epistomal ridges, lateral basal angles of labrum, cardo, stipes, base and apex of mandible, premental sclerite, and apices of salivary lips; in addition, palpi, antennae and dorsomesal streak from each anterior tentorial pit faintly pigmented; cranium, apices of labium and maxillae with fine setiform sensilla; labrum with all surfaces nonspiculate; hypopharynx with spicules.

Head size small compared with body; head capsule wider than long in frontal view. Tentorium complete (including dorsal arms), moderately developed but not robust; anterior tentorial pit moderately small, well separated from anterior mandibular articulation, about midway between it and antenna; posterior tentorial pit conspicuous, well impressed, and found well in front of posterior boundary of head capsule, as seen in lateral view (fig. 29). Median longitudinal thickening of head capsule scarcely developed. Postoccipital ridge weak dorsally, becoming well developed below where it gradually curves forward to become linear extension of hypostomal ridge at posterior tentorial pit (thus contrasting with lower part of postoccipital ridge of Isepeolus, which angles upward immediately behind posterior tentorial pit); hypostomal and pleurostomal ridges moderately well developed; lateral arms of epistomal ridge less well developed; median section of epistomal ridge apparently represented by darkly pigmented, slightly thickened cuticle running dorsomesally from anterior tentorial pit; this thickening ending well laterad of midline of head. Parietal band represented by short oblique line (fig. 28) above antenna. Antennal prominence moderately weakly developed; antennal disk differentiated from papilla; antennal papilla moderately projecting (height about one-half diameter) with approximately six sensilla. Front of head capsule as seen in lateral view (fig. 31) sloping normally with labrum extending well beyond clypeus and clypeus beyond frons; boundary between clypeus and labrum apparently an elongate membranous area (see Remarks below). Labrum small with single median tuberclelike projection bearing cluster of sensilla; labral sclerite (as found in Megachilidae, McGinley and Rozen, 1987: fig. 27) not evident; epipharynx a simple curved surface bulging beneath labrum.

Mandible (figs. 32, 33) moderately robust, short, curving to a rounded apex as seen in aboral and adoral view; apex with small indistinct point; upper and lower apical edges with a few small indistinct teeth; outer surface with numerous small sensilla; cusp not produced, nondentate. Labiomaxillary region strongly projecting, not greatly fused. Maxillary apex distinct, bearing elongate palpus apically; galea not evident; cardo and stipes well developed; articulating arm of stipes evident, connecting with premental sclerite (as in Paratetrapedia swainsonae [Cockerell], Rozen and Michener, 1988: fig. 12). Labium divided into prementum and postmentum; premental sclerite evident and pigmented; apex of labium finely wrinkled suggesting that labium beyond premental sclerite can be strongly projected during cocoon spinning; labial palpus elongate, nearly as large as maxillary palpus. Salivary opening a transverse slit on projecting lips; hypopharyngeal groove not incised and therefore not evident. Hypopharynx recessed well behind labral apex, projecting upward, and flat or nearly so in frontal view (fig. 30).

Body (fig. 29): Integument without obvious setae although that of perianal area with scattered, very fine setiform sensilla; integument appearing nonspiculate under normal stereoscopic examination but under high magnification patches of very fine, evenly spaced spicules visible on dorsal areas of thorax; integument without spines or sclerotized tubercles. Body form moderately robust; intersegmental lines moderately defined; intrasegmental lines not evident; paired dorsal tubercles present on each thoracic segment and on abdominal segmentsI–VII; these tubercles slightly transverse in that, viewed dorsally, each with anterior-posterior length less than lateral width; those of midbody segments separated dorsally on each segment by distance of approximate width of one tubercle; lateral conical tubercles (below level of spiracles) present on most body segments, evident on mesothorax, metathorax, and abdominal segmentsI–VIII (these tubercles not visible in lateral view, but their position indicated in fig. 27); these tubercles far less pronounced than those of Isepeolus viperinus (Rozen, 1966: fig. 2; Lucas de Oliveira, 1966: fig. 2); abdominal segment IX with venter not produce, so that segment X appearing to be attached to approximate middle of IX in lateral view; anus slightly dorsal in position because venter of X somewhat more produced than dorsum; perianal area without transverse ridges but with numerous fine integumental irregularities surrounding anus. Spiracles (figs. 29, 34) moderately small, subequal in size; atrium moderately small compared with subatrium; atrial wall with indistinct concentric ridges; atrium projecting beyond body wall but without rim; narrow peritreme present; primary tracheal opening with collar; subatrium short, with approximately 5–6 chambers. Sex characters unknown.

Material Studied: Three postdefecating larvae, Zonda, San Juan Province, Argentina, November 5, 1995 (E. Michelette).

Remarks: An unusual feature of this larva is the apparent elongate membranous area separating the labrum from the clypeus. This area lacks the conspicuous sensilla and smooth, shiny cuticle found on both the clypeus and the labrum; instead the cuticle of this area is finely wrinkled. It appears to allow for a flexible labrum. This possibility is also supported by the presence of a darkly pigmented spot on each side of the base of the labrum that suggests a muscle attachment point. The morphologically similar larva of Isepeolus viperinus was examined for homologous structures but was inconclusive in confirming the anatomy of the clypeal/labral boundary.

Previous investigations on the phylogenetic relationships of the Isepeolini to the other cleptoparasitic Apidae have been reviewed by Roig-Alsina (1991), and more recently Roig-Alsina and Michener (1993) have dealt with the phylogeny of all long-tongued bees, including this tribe, through a series of cladistic analyses. While the mature larvae of Melectoides bellus and Isepeolus viperinus share numerous unusual features that attest to the monophyly of the Isepeolini, the anatomy of M. bellus does not further illuminate the relationship of the tribe to other parasitic and nonparasitic groups.