INTRODUCTION

We present here information on the nesting biology of two species belonging to separate genera of the Tapinotaspidini, Monoeca and Lanthanomelissa.5 Both genera and, indeed, all members of the tribe are thought to use floral and other plant oils in nest construction, nest provisioning, or both (Michener, 2000). The two species, Monoeca haemorrhoidalis (Smith) and Lanthanomelissa betinae Urban, nest in the ground. We found the nests of M. haemorrhoidalis attacked by an unnamed species of Protosiris, described and named P. gigas in the appendix. This is the first association of any species of Protosiris with a host. Nests of L. betinae were parasitized by Parepeolus minutus Roig-Alsina, mature larvae of which were recovered from the nests. This is the first confirmed host/parasite association of a member of Parepeolus, although Rozen (1984b) found an adult of Parepeolus niger Roig-Alsina in a nest of Tapinotaspoides serraticornis (Friese) (cited as Tapinotaspis tucumana (Vachal)), and Roig-Alsina (1989) tentatively associated Ecclitodes stuardi (Ruiz) with Chalepogenus caeruleus (Friese).

In addition to describing various aspects of the nesting biology of the two host species, we present biological information on their two cleptoparasites. The mode of cleptoparasitism of Protosiris gigas provides the first understanding of this matter for any member of the cleptoparasitic tribe Osirini. Because the immature stages of the Tapinotaspidini and Osirini are poorly known, we include an extensive treatment of eggs, larvae, and pupae of species whose nesting we investigated.

Data resulting in this report were accumulated over period of more than 3 years. The nesting biologies of Monoeca haemorrhoidalis and Protosiris gigas were studied at Mananciais da Serra, Piraquara, Paraná, Brazil. Melo (G.A.R.M.), Rozen (J.G.R.), and Aguiar (A.A.) gathered information for several days starting November 20, 2002, and G.A.R.M. and A.A. subsequently visited the nesting site on January 24, 2003. Starting on December 3, 2003, all authors, including Alves-dos-Santos (I.A.S.), returned to the same nesting site, which was again active, and examined it on and off over a 9-day study period. G.A.R.M. and A.A. carried out further observation on the site January 17, 2004, December 5, 2004, and January 30, 2005.

I.A.S. discovered one of the nesting sites of Lanthanomelissa betinae on the campus of Universidade do Extremo Sul Catarinense (UNESC), Criciúma, Santa Catarina, Brazil, in October 2002. All authors carried out excavations on it on November 15 and 16 of that year. Further studies were made there by I.A.S. in November 23–26, 2003. The other Lanthanomelissa site was on the campus of Universidade Federal do Paraná, Curitiba, Brazil. It was initially discovered by G.A.R.M. in late October 2001; J.G.R. excavated nests there in November 25, 2002.

Thus, all authors contributed to the biological information presented herein and to collecting and preserving immature stages. J.G.R. described the immature stages herein, and G.A.R.M. identified the adults and prepared the taxonomic description of the new species presented in the appendix.

METHODS AND TERMINOLOGY

Preserved larvae were first illustrated while in ethanol, resulting in pencil illustrations of the entire specimen and enlarged frontal and lateral views of the head. Heads were then separated from the bodies, and both were cleared in a heated aqueous solution of sodium hydroxide until all tissue was removed. If large, specimens were examined in ethanol, or if small (usually), they were placed in glycerin. Details of anatomy such as setae and sensilla were added to the illustrations, and mandibles and spiracles were dissected and illustrated.

In estimating the approximate length of a strongly curved larva, four or five relatively straight sections of it were measured in lateral view at the level of the spiracles; these measurements were then summed, as illustrated in figure 52.

The term mature larva refers to the fifth instar after it has finished feeding. We use the term postdefecating larva (often called the prepupa by others) for the mature larva after it voids its feces. The term predefecating larva refers to the mature larva before it defecates. These two stages, although of the same instar and therefore usually identical in such structures as mandibles and spiracles, are quite different in overall shape, integumental texture, and behavior. The predefecating form is more robust, its integument is thinner and far less wrinkled, often of a different hue, and it is active (i.e., its head and body can move and its mandibles open and close). The postdefecating larva is more slender, often with body tubercles more pronounced, its integument is thicker and more wrinkled, and, when in full diapause, the larva is completely inactive. For most univoltine species, the postdefecating larva is the stage in which the bee diapauses until the next nesting season.

Because postdefecating larvae are the forms most frequently collected, descriptions of bee larvae are usually based on that stage. Descriptions of the mature larvae presented here are based on postdefecating forms. However, in cases where we also have been able to study mature predefecating larvae, we described significant differences with respect to body tubercles and the shape of the terminal abdominal segments, following the description of the postdefecating form. Other differences, such as body color of predefecating forms, are not mentioned, and similarly the more extensive expression of internal head ridges of predefecating larvae are not described, except in the case of Lanthanomelissa betinae, which will serve as an example. It seems likely that the less prominent internal ridges in the head of postdefecating forms result from the thickening of the cuticle elsewhere as the larva enters diapause, so that there is less contrast between ridges and nonridged areas.

In the frontal-view diagrams of larval heads, sensilla and internal ridges are indicated on only their left side.

Larvae and eggs were examined with a Hitachi S-5700 scanning electron microscope (SEM) in the Microscopy and Imaging Facility of the American Museum of Natural History after they were critical-point dried and coated with gold/palladium. Mature larvae so examined unexpectedly revealed elongate, setiform papillae associated with the salivary openings (figs. 60, 68, 96, 99). These structures are termed salivary papillae rather than spicules because they appear to be nonsclerotized. Their patterning differed from one species to another but suggests that they may have some function associated with the application of the silk in cocoon construction, such as serving like the end of a paint brush in distributing liquid over a broad surface. The papillae were scarcely visible when viewed with a light microscope and thus may have been overlooked in previous studies of other taxa. The extent to which other cocoon-spinning mature larvae have them should be investigated.

The egg index, referred to in the descriptions of eggs and in the section on Ovarian Statistics, is a method 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). It is calculated by dividing the length of the egg or mature oocyte by the distance between the outer rims of the tegulae. We gathered these data two ways. For the section on Ovarian Statistics, we measured the length of the longest mature oocyte in a female and divided that figure by her intertegular distance. For the descriptions of eggs of the taxa, we averaged the lengths of eggs recovered from nests and then divided the resulting figure by the average intertegular distance of a sampling of females. Iwata and Sakagami (1966) proposed a five-category classification of bee eggs (mature oocytes) based on their indices; that is, length of the egg (oocyte) (E) divided by the distance between the outer rims of the tegulae, or metasomal width (M):

The abbreviations used for adult structures are: T for metasomal tergum and S for metasomal sternum, followed by an Arabic number denoting the segment (e.g., T6 refers to sixth metasomal tergum).

BIOLOGY OF MONOECA HAEMORRHOIDALIS (SMITH)

Several nesting sites of Monoeca haemorrhoidalis were discovered in 2002, 2003 and 2004 along a seldom-traversed, unpaved road extending more than 3 km through the heavily forested mountainous area. The largest site (fig. 1) was excavated during these 3 years to provide the information presented here. It occupied about a 12-m-long, horizontal section of the roadway. Being about 3 m wide at either end, its midsection widened to about 6 m and contained the most nests. The site was shielded marginally by the forest canopy so that much of it received considerable sunlight on clear days. Ground cover, mostly grasses, obscured much of the surface, so that many nest entrances were partly or completely hidden (fig. 2). In 2002, several smaller nesting sites along the forested roadway were discovered, but none of these showed activity of Protosiris gigas. However, in 2003 we discovered this cleptoparasite at a less populous site about 1 km from the study site.

In early December 2003 the main site was extremely active midmorning to late afternoon, with flying Monoeca haemorrhoidalis and Protosiris gigas producing a constant hum. Both species tended to fly no more than 0.5 m above the ground, although occasionally individuals briefly landed a meter or more above the surface on surrounding vegetation. Because of the dense, low groundcover, nest density could not be measured. Rough estimates of the number of individuals of both species combined ranged from approximately 50 per m² on December 4, 2003, during a very active period, to 25 per m² the following day, a reduction presumably resulting from different weather conditions. At these times, most of the Protosiris were males, whereas the Monoeca population was of both sexes, though males seemed more abundant. Although hosts and cleptoparasites were generally scattered over the entire nesting area, we frequently saw small flying aggregations of mostly male Protosiris, suggesting that they may have been pursuing a receptive female.

In December 2003, females of Monoeca haemorrhoidalis were seen near the nest aggregation, visiting flowers and flower buds of Tetrapterys guilleminiana A. Juss. (Malpighiaceae), a tree-climbing woody vine that presents yellow flowers arranged in large inflorescences. Most plants were in the tops of trees, but some branches had fallen onto the forest edge. One female collected on these Tetrapterys flowers carried only oil in the scopae of the hindlegs. On one plant of T. guilleminiana with nonglandular flowers, we saw one Monoeca female collecting pollen and another female with barren scopae approaching three flowers but leaving without landing. The scopal loads of five females carrying pollen consisted of Tetrapterys pollen. Also, all samples of pollen from seven broad cells contained Tetrapterys pollen, with most of them composed solely of Tetrapterys pollen. Only two of the cells analyzed revealed a half-and-half mixture of Tetrapterys pollen and unidentified pollen grains (triangular pollen grains with three furrows, each one crossed by a transverse furrow).

Adult activity at the site was greatest on December 4, 2003, and declined gradually afterward although both parasites and hosts were still moderately abundant when we last visited the site on December 11, 2003, at the end of the joint study period. When G.A.R.M. and A.A. revisited the site on January 17, 2004, they discovered many flying males and nesting females of Monoeca haemorrhoidalis, which seemed about as abundant as in early December, but adults of Protosiris gigas were completely absent.

The start of the nesting activities of the next generation in 2004 seemed to have been delayed at the site. The adult activity on December 5, 2004, was only about one-fifth to one-sixth of that observed in December 2003, despite the favorable weather conditions. Also, no fresh cocoons or full-grown larvae were found.

Copulating Monoeca haemorrhoidalis were occasionally observed on the low vegetation covering the nesting site, with one or more other males often attempting to intercede with the copulating pairs. Females of Monoeca haemorrhoidalis presumably spent the night in nest burrows. Sleeping places of males were not found.

Only a single female occupied each nest. Nest entrances of Monoeca haemorrhoidalis were surrounded by abundant moist tumuli of excavated soil, and they lacked distinct turrets, although the worked wall of the main burrow extend through the tumulus at least on some occasions. Entrances were randomly scattered, with some being 5 cm apart and others more distant. Many could be seen at one time over a broad area, but most were partly to nearly completely hidden by ground cover (fig. 2). Main burrows descended vertically with limited turning through the moist, nearly rock-free substrate and were open their entire length. Because of on-going rainy spells, the substrate was always moist, but heavy downpours did not flood our large study excavations, attesting to the drainage capabilities of the nesting substrate. Burrow diameters ranged from 7.0 to 9.0 mm, and burrow walls were smooth and moderately shiny (fig. 3). At least in one area, the burrow walls were lined with extremely fine, reflective, claylike material about 1 mm thick that was red and contrasted with the browner surrounding substrate. Their color matched that of the lower substrate and no doubt resulted from the lower substrate material being brought to the surface during nest building. However, whether the tunnel linings were an accidental byproduct of nest excavation or a special construction by the female M. haemorrhoidalis is unknown. The surface of the burrow lining showed distinct, obviously repetitive, tamping impressions only in some areas (fig. 3). These impressions are thought to have been from the female using her pygidial plate, an activity known in other bees. A water droplet applied to the wall remained beaded on the surface for at least a minute. The surface was not coated with the same waterproof material used on the cell walls, so that its shiny, hydrophobic surface may have resulted from the mechanical compression of soil particles through the tamping process. Water-retardant burrow walls are an uncommon feature among solitary, ground-nesting bees. Perhaps water leaking through the walls of deep burrows with large diameters is hazardous in wet environments.

Overall nest configuration of this species could not be determined, although nests were obviously deep and almost certainly consisted of numerous cells. Too many main tunnels from the current and previous generations penetrated the soil to permit any one to be followed to reveal the branching pattern, and too many cells, both current and old, were encountered to assign them to a single main burrow. In one area fresh cells were first encountered at a depth of about 70 cm, and others may have extended below the 1 m level. In other areas a few cells were found as shallowly as 30 cm. All cells tended to be vertical but were tipped slightly, although we noticed one inclined as much as 45°.

Cells were bilaterally symmetrical around their long axes. One side was more outcurved, and the opposite side was more or less incurved above and straighter below, so that the long axis of the cell tended to be curved (figs. 7, 9). This feature is uncommon in bees, but interestingly it is characteristic of the distantly related Diphaglossinae (Rozen, 1984a), where it is often more accentuated. (Andrenidae and Halictidae commonly have one side more outcurved and the opposite side nearly flat, but the opposite side is not incurved as in the Diphaglossinae and Monoeca haemorrhoidalis.)

The cell lining was reflective, waterproof, and faintly milky. It not only coated the cell wall but also extended partly into the connecting lateral. When examined with a stereomicroscope, the material appeared brittle, fracturing rather than bending under stress. Its surface was patterned with faint curved parallel lines, one patch adjoining another, as if swept by a series of brushstrokes. A sweeping by the female's much enlarged pygidial plate (see fig. A7) will likely be found to account for these marks as she applies the coating to the cell surface. When cells were allowed to dry over time, their lining tended to fracture and pull away from the soil, so that pieces could be removed from the cell wall with forceps. When we placed a flake on a microscope slide that was then heated on a hotplate, the flake melted into a nearly clear liquid, which solidified when cooled to room temperature. The cooled material became less transparent, and, when scraped with forceps, the scrape mark was shiny. Thus, the material had the physical characteristics of wax. We would be surprised if this substance did not come from the large, pale dorsal wax gland on the anterior part of the female's sixth metasomal tergum (T6) that is normally hidden under the previous tergum (see also appendix, figs. A7–A8).

Santos et al. (2004) studied the histology of integumental glands in the metasoma of Monoeca xanthopyga Harter-Marques, Cunha, and Moure. The large gland present in the female's T6 was found to be a class I gland that histochemically produced lipids. We assume that the gland in M. haemorrhoidalis is homologous to that present in M. xanthopyga. We examined additional species of Tapinotaspidini and found that the external evidence of the gland (a transverse depression at the base of the female's T6) was present in M. schrottkyi (Friese) and M. pluricincta (Vachal) but absent from Monoeca lanei (Moure), Monoeca sp., Trigonopedia glaberrima (Friese), T. ferruginea (Friese), Arhysoceble dichroopoda Moure, A. huberi (Ducke), Tapinotaspoides serraticornis (Friese), Paratetrapedia amplipennis (Smith), P. punctifrons (Smith), P. larocai (Moure), and Chalepogenus sp.

The inner surface of the cell closure was a deeply concave spiral of 4–6 coils (fig. 4), which lacked a waterproof lining. It absorbed a water droplet immediately when tested.

Cell lengths measured from the apex of the concave spiral closure to the cell bottom were 18.0–26.0 mm (x̄ = 23.2 mm, N = 15). The great range in lengths may have resulted from variation as to where the female placed the cell closure; for example, the shortest cell (18.0 mm) had the largest diameter at the closure (8.0 mm). Maximum cell diameters were 8.5–10.0 mm (x̄ = 9.3 mm, N = 10). Cell diameters at the closure ranged from 5.5 to 8.0 mm (x̄ = 6.3 mm, N = 9), with all closure diameters (except for the two extremes) between 6.0 and 6.5 mm. In general, cells were elongate relative to their maximum diameters, with their lengths being approximately 2.5 times their maximum diameters. We wonder if the unusual elongation of the cells of this species might have been driven by natural selection: host eggs farther away from the cell closure are distanced from the stings of the cleptoparasite (see Biology of Protosiris gigas, below).

Laterals appeared to approach cells from variable directions, sometimes rising shortly before bending downward to connect to a cell, and other times descending straight downward to connect to the cell. All were soil filled after cell closure, so that we were unable to trace them.

Pale cream-colored provisions when first brought into a cell were placed in the bottom and had a somewhat concave top surface (fig. 8), much as in Lanthanomelissa betinae. The full complement of provisions in a cell required a number of foraging trips, so that the basal portion with its concave upper surface appeared at more than one level among cells being provisioned. Completed provisions consisted of a homogeneous mass of moist pollen 4.5–7.0 mm deep (N = 4) from which arose a central mound, skewed away from the outcurved side of the cell (fig. 7). The height of the mound measured from the bottom of the cell was 8.5–9.0 mm (N = 5). The surface of the provisions was irregular. The female deposited her strongly curved egg on the less vertical side of the mass just below the summit, so that the posterior end of the egg pointed radially outward (fig. 7). Generally the anterior and posterior ends of the egg contacted the surface of the provisions, while its midsection was elevated above the surface.

The nature of our investigation did not permit us to measure developmental rates of eggs, larval instars, and pupae. However, casual observations suggested that the lengths of these stadia were not unusual for bees. Embryos of Monoeca haemorrhoidalis at first developed with the ventral surface dorsal within the chorion, and then they rotated 180° before hatching, so that the first instar's ventral surface rested next to the provisions, as mentioned, for example, by Torchio et al. (1988) for many other bees.

During eclosion, the embryonic fluid was ingested, so that the chorion adhered to the now defined, first-instar head capsule, and the trachea filled with air. The chorion split along the spiracular line on each side of the body (as was determined by examining preserved first instars for which the chorion had been partly removed by the wash of the preservative), while the ventral surface of the head was buried in the provisions. The rest of the body did not adhere closely to the food. By placing a droplet of a 1% aqueous solution of Fast Green on the provisions immediately in front of the head, we saw the green being ingested into the alimentary tract, an indication that the first instar was ingesting external fluid, not just embryonic fluid. During this time, the larva remained motionless except for faint contractions of the body surface presumably associated with fluid ingestion and tracheal ventilation. The body remained rigid and did not adhere closely to the provisions. Only when the second instar shed the chorion simultaneously with the first-instar integument did the ventral surface of the body come into complete contact with the provisions. At this time the mandibles started opening and closing, commencing pollen ingestion, and the second instar crawled away from the previous exuviae.

To investigate how the first instar is able to ingest liquid from the provisions, we examined a preserved first instar, still surrounded by chorion. It revealed a large, circular, bowel-shaped invagination of the chorion in the vicinity of the larva's mouth (i.e., below the labrum and between the mandibles), possible external evidence of the passageway of the fluid into the foregut. From another preserved first instar, we were able to remove the chorion covering the head and examine it with a compound microscope. Although we expected to see an opening in the chorion on the innermost surface of the invagination, there was no visible rent. However, the micropylar array was there or near there, and this discovery raised the question of whether the micropyle pores might be the passageway for the liquid to be sucked into the foregut. We were able to examine yet another pharate first instar, this time with an SEM. Although the front of the chorion partly came away from the larva as it was being mounted on the SEM stub, the resulting images (figs. 11, 12) showed the inner surface of the chorion with the micropyle fully displayed, just below the larval head. This appears to be strong evidence that the micropyle is indeed the passageway that allows the first instar to ingest fluid from outside the chorion.

On this same specimen, we examined with the SEM the line of minute granules that extends longitudinally just above the spiracles (see description of first instar Monoeca haemorrhoidalis). Under very high magnification (figs. 14, 15), these granules are definitely spicules in that they do not articulate with the cuticle. They are stout at the base and most, if not all, are sharply pointed. We suggest that their position and shape may serve as the tearing mechanism that causes the chorion to split along the spiracular line as the body of the embryo/first instar swells from ingesting first embryonic fluid and then perhaps external fluid during eclosion. This hypothesis and the hypothesized ingestion of external fluid through the micropyle need further investigation, but they may well shed light not only about eclosion in Monoeca haemorrhoidalis but more broadly about eclosion in many other groups of bees, since the lateral splitting of the chorion has been observed in other bees (Alves-dos-Santos et al., 2002; Rozen, 1964, 1969; Rozen and Buchmann, 1990; Torchio, 1989a, 1989b; Torchio and Trostle, 1986).

While feeding, the second instar roamed actively over the surface of the provisions. It was remarkably agile, being able to raise the anterior part of its body into the air while only the terminal 2–4 body segments remained in contact with the provisions. Thus, it was capable of leaving the provisions to climb the cell wall and, when being reared in an artificial container, to leave the cell altogether. Later-stage larvae lay sideways on the provisions, with their dorsal surface against the cell wall as they fed and moved forward. As a consequence, the provisions developed a more pointed conical central peak surrounded by a trough containing the larva (fig. 9). As larvae became more enlarged, they looped the midsection of their body upward, so that the head could feed on the provisions on one side of the cell and the apex of the abdomen touched the provisions on the opposite side. While in this position, the larva's feeding caused the central peak of the provisions to become facetted, that is, flattened on one side rather than evenly cone-shaped.

It seems likely that the changes in body spiculation from one instar to the next (treated in the descriptions of the larval stages) relate to their mode of activity. Unfortunately, we were unaware of these changes when we were in the field where we could have studied them in greater detail. Nonetheless, changes in spiculation of the early instars seem clearly related to larval activity. The first instar does not crawl and has no body spicules (except for the minute ones above the spiracular line); the second and third instars, which actively crawl, have conspicuously spiculate venters. For the fourth instar, spicules over all surfaces of its body probably relate to the fact that it feeds on its side as it circles the food mass. However, the nearly complete loss of body spicules of the fifth instar is difficult to understand.

Toward the end of the feeding phase and before spinning their cocoons, larvae start biting and ingesting the cell lining, presumably at first to eat the few pollen grains still adhered to the wall, but later clearly consuming the lining. Inspection of cell walls and direct observations of live specimens collected in January 2005 indicate that the amount of lining removed by the larvae varies among cells. In some cases, only a few portions of the lining showed bite marks, while in others, the entire lining covering the bottom three-fourths of the cell was removed.

After provisions were entirely consumed, larvae of Monoeca haemorrhoidalis started to spin their cocoons, all of which occupied the lower (rear) end of the cells (fig. 10). Completed cocoons had the following dimensions: length measured from middle of the truncated upper end to the bottom (rear), 14.0–18.0 mm (x̄ = 15.7 mm, N = 8); maximum diameter, 9.0–9.5 mm (x̄ = 9.1 mm, N = 8); diameter of top, 7.0–7.5 mm (x̄ = 7.1 mm, N = 6). Thus, considerable empty space (figs. 10, 23) remained between the top of the cocoon and the cell closure, a distance that measured 7.0–10.0 mm (N = 4). In most cells, mold hyphae grew in this space after cocoon construction (fig. 23). These brown hyphae grew from fine, whitish hyphae covering the bottom part of the cocoon, apparently developing on the bee feces. They grow over the cell wall, extending through the cell plug all the way to the lateral tunnel, where they produce dark spherical bodies (spore-producing bodies?). Similar mold hyphae were found in association with cocoons of Tapinotaspoides serraticornis by Rozen (1984b). The nearly ubiquitous presence of the hyphae in Monoeca cocoons and the position of the putative fruiting bodies, facilitating the contact of the emerging bees with the spores, suggest a strong, possibly mutualistic association.

Cocoons were elongate, occupying and conforming to the shape of the lower end of their cells. In side view, one side curved outward and the opposite side was nearly straight since it conformed only to the lower part of the cell (figs. 10. 17). The top of the cocoon was planar, with the perimeter usually slightly raised as a soft rim. The plane of the top angled downward so that the surface slanted toward the straighter side of the cocoon.

The entire cocoon, except for the top, consisted of an outer layer of fine, brown, silk fibers and an inner layer of thin, fenestrated sheets of fibrous silk intermixed with feces. The inside of the inner layers was coated with feces. On older cocoons, the inner surface was often blackened, perhaps resulting from bacterial action on the feces. When moist (i.e., when first excavated), the cocoon fabric had a soft, feltlike texture, which hardened on drying. The fabric of the cocoon was moderately thin in contrast to the thicker fabric of the Protosiris cocoon (see description of Protosiris cocoon, below).

The top of the cocoon (fig. 18) consisted of several fenestrated layers of silk well separated from one another but connected by a loose network of strands. Feces were not incorporated in this part of the cocoon, and the fabric was sufficiently open so that light was transmitted through the top end when it was held between a light source and the observer. Apparently, variation exists in the nature of the innermost layer of silk on the top. In one specimen this layer was a fine, transparent sheet of silk. However, in other specimens the two-dimensional, sheetlike material was absent, and a multidirectional network of individual strands existed. The latter appeared to be the more usual construction. Feces, while not occurring within the fabric at the top end, are dabbed on the margins to the inner surface of this end, though not usually on the center.

On eclosion, Monoeca haemorrhoidalis adults destroy the tops of the cocoons altogether so that only the sidewalls and bottoms of the cocoons remain (fig. 21). This contrasts with egress of adults of Protosiris gigas, a more slender bee, in which the outer rim of the top of the cocoon remains mostly intact while the center of the top is destroyed (fig. 22).

Monoeca haemorrhoidalis has a single generation per year. A sampling of the cocoons still containing live Monoeca offspring from 2002 was made at the nesting site on December 11, 2003. These cocoons contained 19 males (12 live adults, 7 live pupae), 11 live females (1 live adult, 10 live pupae), and 10 live postdefecating larvae, the sexes of which could not be determined. Obviously, the males were further advanced in their development than were the females, the pupae of which tended to be not fully pigmented. The adults and pupae would have emerged later in the season, but we cannot be certain if the postdefecating larvae would have. If not, then, this would seem to be a case of parsivoltinism (Torchio and Tepedino, 1982).

Aguiar et al. (2004) presented a synopsis of the nesting biology of tapinotaspidine bees, including the biology of M. haemorrhoidalis presented here. Interestingly, the physical properties of the cell lining of M. lanei were as described above for M. haemorrhoidalis. As with that species, a piece of the lining from the previous investigation (Rozen, 1984), preserved in the collections of the American Museum of Natural History for the last 30 years, was melted on a glass slide placed on a warm hotplate. Despite the similarities in the cell linings, females of M. lanei do not have an indication of an epidermal gland on T6, as noted above.

BIOLOGY OF PROTOSIRIS GIGAS MELO

Protosiris gigas is a cleptoparasite of Monoeca haemorrhoidalis. Males and females were in abundance at the nesting site, immatures (eggs, larvae, and pupae) were recovered from host cells, and cleptoparasitic females were occasionally encountered in the host nests.

The similar appearance of males and females of Protosiris gigas in flight at the nesting site and the great abundance of males made it impossible to observe the female host-nest searching behavior. Males of Protosiris gigas were found in the late afternoon sleeping while clinging to leaves and stems of bushes adjacent to the site. With wings plated over their bodies, they clung to the vegetation holding on only with their mandibles (fig. 24). Females, few in number during our observations, were not observed sleeping, but they too probably sleep holding onto the vegetation with their mandibles. The abundance of males of this species at the site left little doubt that males were searching for females and that copulation occurred there.

We noticed no interaction between adults of Monoeca haemorrhoidalis and Protosiris gigas at the nest site; they seemed oblivious of one another.

Invariably, when we encountered parasitized cells that retained the cell closure, we found that the Protosiris female had created a large hole in the center of the spiral closure. Thus, this cleptoparasite attacks cells after they have been sealed by the host female. After the female Protosiris gigas oviposited, she plugged the hole with seemingly loose soil (figs. 5, 6) in each case. Three of these holes, roughly circular, measured 3.3 × 4.0, 3.5 × 3.75, and 2.5 × 3.5 mm. Host immatures (eggs or early instars) were found dead in cells containing unhatched Protosiris eggs on seven occasions. It therefore seems clear that the female cleptoparasite is primarily responsible for killing the host immature. The behavior and anatomy of the first-instar Protosiris indicate that it too can kill the host immature, should it survive the attack by the adult cleptoparasite.

Although not fully understood, eclosion of Protosiris gigas is different from that of the host, described above. Instead of being pharate, the first-instar Protosiris is active and crawls away from the chorion. A preserved first-instar specimen (fig. 28), studied with an SEM, showed it emerging from an opening at the anterior end of the egg while the longitudinally shriveled chorion with posterior end intact was left behind. A jagged rent (indicated by arrows in fig. 28) seemed to occur along one side of the chorion and may be comparable to the lateral longitudinal splitting of the Monoeca chorion along the spiracular line. A similar tearing of the chorion on the opposite side could not be certainly detected but may have been there. We were unable to identify a linear series of small spines or the spiracles on the specimen studied with the SEM, and a cleared specimen in glycerin examined with a compound microscope showed no such linear series even though the spiracles were clearly visible. Thus, it seems unlikely that the linear series of spines exists on this species.

The question remains how does the female Protosiris kills the immature Monoeca. Two ways have been documented by which a female cleptoparasitic bee eliminates the host offspring: (1) The cleptoparasite female opens the cell and kills the host immature with her mandibles (Bennett, 1966, for Hoplostelis bilineolata (Spinola); Bennett, 1972, for Exaerete dentata (Linnaeus)); and (2) the cleptoparasite female makes a small opening in the side of the host cells using her mandibles, reverses her position, and kills the host egg with her metasomal apex (Garófalo and Rozen, 2001, for Exaerete smaragdina (Guérin-Méneville)). All other known adult cleptoparasitic Apidae that introduce their eggs into sealed host cells do not kill the host offspring, but rather they have larvae that are hospicidal at one stage or another (Rozen, 2003: table 1 and references therein).

In the case of Protosiris gigas, indirect evidence supports the hypothesis that the female uses her metasomal apex, and perhaps the sting, to kill the Monoeca egg or early instar. A total of 5 eggs (and one shed chorion) of Protosiris gigas were found attached to cell walls approximately halfway between the cell closure and the top of the provisions (figs. 7, 25, 27), proof that the female must be able to extend her metasoma through the hole in the cell closure. In all of these instances, the host immatures (4 eggs, 1 second instar) were clearly dead, showing loss of turgor or even being misshapen, but none was removed from its normal position on the food mass, as might have been the case if it had been killed by the mandibles of the Protosiris female. Also, none was missing, as if it might have been eaten by the cleptoparasite. Because we found another 5 live Protosiris eggs at various locations on the provisions (e.g., on the summit, fig. 26; on the periphery), egg placement by Protosiris gigas may be variable, or the eggs may have originally been attached to the wall and the attachments failed, sending them onto the surface of the provisions.

On two other occasions a seemingly live host egg with an apparent fine puncture exuding a droplet of yellowish fluid on its anterior dorsal surface was noticed in a cell with a live Protosiris egg (fig. 27). This may have resulted from a puncture created by the parasite female's sting. We perhaps encountered these cells shortly after they had been attacked, so that the host eggs had not yet lost their turgor.

We reject the hypothesis that the female Protosiris used her mandibles to kill the host immature, both because there is no evidence that the Monoeca immatures were dislodged or missing from their normal positions and because it is doubtful that the opening of the cell closure made by the female would have been sufficiently large to allow her to reach into the cell. Furthermore, it seems unlikely that the Protosiris female would then reposition herself to attach her egg to the cell wall.

However, is it reasonable to think that the sting (and metasomal apex) of the Protosiris female can reach the host egg, that is, to the level of the top of the central mound of the provisions? In the laboratory, we measured the maximum metasomal diameter and the length of the female metasoma from the petiole to the apex of the sting on two dead females with the metasomas fully extended. In both cases, the maximum metasomal diameter was 3.0 mm, and the distance from petiole to sting apex was 15 mm. The size of the actual holes in the cell closures (given above) would certainly permit the metasomas to extend into the lumen of the cell. Three measurements of the distance between the cell closure and the top of the provisions were 12, 14, and 15 mm. We conclude partly on the basis of these distances that the female Protosiris probably inserts her entire metasoma through the hole in the cell closure so that the sting can damage the host egg or larva. At this time the cleptoparasite presumably places her egg on the cell wall (or drops it onto the provisions). This conclusion is also partly based on the fact that the evidence does not point to any other explanation.

The sting apparatus of Protosiris gigas is elongate relative to body size, presumably an adaptation for reaching the host egg and depositing its own, with the sting shaft, gonostyli, first and second rami, and furcula being longer than those of most of other bees. However, these structures are not as long as those of the related Osiris (Packer, 2003).

Certain matters, however, remain unclear. Might the posterior part of a female's mesosoma also be inserted through the cell closure, enhancing her ability to reach the host immature? How does the female manage to penetrate the thick soil-filled closure, or does she only reach the closure prior to the lateral being filled? How does she position her legs when she extends her metasoma?

A host cell containing a dead Monoeca second instar and a live Protosiris egg (fig. 25) indicated that a parasite female can gain access to a cell that has been closed for some time after cell closure. Whether she is able to penetrate a soil-filled lateral is unknown. The presence of a live Protosiris first instar with a dead Protosiris egg in another cell indicated either that a cell can be attacked sequentially by two parasite females or (less likely) that one female parasite may places two eggs in a cell at a time.

In two cells in which the Protosiris egg was attached to the wall, the attachment was clearly by its posterior end (one egg so attached projected into the cell lumen at nearly a right angle to the cell wall; fig. 27). In all other cases cleptoparasite eggs seemed attached by their full lengths to the walls. SEM examination of an egg (fig. 76) did not reveal any special modification of the sculpturing of either the posterior end or other surfaces that might function to hold the egg to the wall. In the distantly related Melectini, which attach their eggs to the cell closures or upper cell walls by their posterior ends, the surface sculpturing of this end is quite different from the rest of the chorion (Rozen and Özbek, 2003: figs. 63, 65; 2005: fig. 11), being possible adaptations to enhance attachments to cell surfaces.

The chorion left behind by a newly emerged Protosiris larva was carefully examined for remnants of a cast first-instar exoskeleton. Since none was detected, we concluded that the active larva next to the chorion was the first instar (the five larval instars of this species are differentiated and described below). All first instars found were active and agile. One was observed crawling over the surface of the provisions and returning several times to chew at the remnant of the host egg. This larva was observed clinging to the substrate with the venter of the last three abdominal segments and raising the anterior part of its body until its body looped backward so that its head was posterior to its abdominal apex. The narrow, transverse anterior folds on the venter of abdominal segments 3–7 (see fig. 78) may assist the larva in this maneuver by providing extra ventral integument to elongate the larva's backward reach. With a microscope, we observed a larva crawling upside down on a glass surface. We saw that these folds also permitted the larva to crawl forward; as the abdominal apex pushed against the substrate, the larva's preceding segments stretched, one after another, so that the anterior end of the larva moved forward. The larva evinced no special structure (pygopod) at its abdominal apex to accomplish this, although the ventral surface of the larva including that of the abdominal apex was spiculate.

Although the female Protosiris presumably usually kills the host egg (or larva), the first-instar Protosiris seems perfectly capable of killing a host immature. We observed a first instar repeatedly attacking a dead host immature. Another attempted to bite the tip of a pin when teased, and it repeatedly opened and closed its mandibles as it searched the air for the pin. The anatomy of the head with its sharply pointed, strongly curved mandibles (figs. 77, 79–81) also indicates that the larva is capable of killing the host egg or larva. It is likely that the evolutionary driving force for these behavioral and anatomical adaptations is the need to be able to battle other cleptoparasites in the same cell rather than for host killing. In the case where a dead Protosiris egg and a live first-instar Protosiris were found together in a cell, we cannot be certain whether the egg was killed by the first instar or by the second Protosiris female to find the cell. Both scenarios are possible. We also observed a third-instar Protosiris responding to our probing forceps by turning its head in the direction of the attack and repeatedly opening and closing its sharply pointed mandibles (fig. 83) (see Other Larval Instars of Protosiris, below). Thus, this aggressive behavior appears to continue beyond the first stadium and indirectly implies that the successful ownership of the provisions may go undecided for the duration of several stadia. Interestingly, Garófalo and Rozen (2001) reported that the female of the distantly related cleptoparasite Exaerete smaragdina also normally destroys the host egg and that her offspring (but the second and not first instar) also are capable of battling with other cleptoparasites or with a host immature that might have survived.

Cocoons of Protosiris gigas (fig. 19) resemble closely those of the host externally. They are tan and composed of fine silk strands, and they have the same shape and adhered closely to the lower part of the cell. They exhibited the same planar, truncated top surface of the Monoeca cocoon. They have the following dimensions (N = 5): length, 14.5–18 mm (x̄ = 16.3 mm); maximum diameter, 8.0–9.0 mm (x̄ = 8.6 mm); diameter of top, 6.0–8.0 mm (x̄ = 6.9 mm).

The cocoon wall (i.e., except for the top) in cross section appeared thicker (at least on fresh, moist specimens) than that of Monoeca haemorrhoidalis. This was probably due to the fact that the cocoon consisted of three layers: an outer silken layer, an inner silken layer, and in between a layer composed of numerous fine sheets of silk sandwiching the entire meconial mass. Consequently, the inner surface of the cocoon exhibited no feces, contrasting with the cocoon of the host. Another noticeable difference between the host and parasite was in the tops of their cocoons. The cocoon top of Protosiris consisted of numerous densely layered fibrous sheets of silk (fig. 20), so that when held in front of a light source, no light penetrated through the top. Thus, the top is thicker, denser, and stronger than that of Monoeca. The open, meshlike outer surface of the Monoeca cocoon that loosely attaches to the inner surface is replaced in Protosiris with a sturdier fabric, the top surface of which is not easily compressed. On drying, the cocoons of both species shrink; the strong, dense disc of the top of the Protosiris cocoon shrinks less, giving the upper part of the cocoon a slight hourglass shape. On eclosion, the Protosiris adult does not demolish the top (front) end of the cocoons, as is characteristic of Monoeca; instead, the emerging adult chews its way out of the cocoon by burrowing through the middle of the top, leaving the periphery in place (fig. 22).

Although the ratio of Monoeca haemorrhoidalis to Protosiris gigas flying over the nesting site on December 4, 2003, appeared to be roughly 1:1, G.A.R.M. and A.A. inventoried the cells on January 24, 2003, when the site was quiescent. They recovered cells of 87 Monoeca, 8 Protosiris, and 28 containing Tetraonyx (Paratetraonyx) distincticollis Pic (Meloidae) (adults of which were tentatively identified by Dr. John Pinto).6 When G.A.R.M. and A.A. sampled the nesting site on January 17, 2004, 96 cells contained Monoeca (36 eggs and active immatures + 60 postdefecating larvae in cocoons), 8 Protosiris postdefecating larvae in cocoons, and 25 Tetraonyx distincticollis and yielded a host/cleptoparasite ratio of approximately 11:1. The discrepancy between this ratio and the estimated ratio of 1:1 based on flying adults on December 4, 2003, was caused by the earlier emergence of the cleptoparasite compared with that of the host, by a briefer period of emergence of the parasite compared with a longer emergence period of the host, by host females being away foraging while the parasites were concentrated at the nesting site, or by some combination of these phenomena. There is a selective advantage for a host to prolong its emergence so that it outlasts the seasonal activity of the cleptoparasite. Data obtained in January 30, 2005, revealed an even lower host/cleptoparasite ratio: 63 cells contained Monoeca (17 predefecating larvae + 46 postdefecating larvae in cocoons, 6 of them dead), 2 Protosiris (1 dead larva and 1 dead male imago in their cocoons, both apparently from the previous season), and 2 T. distincticollis. The few cells containing young immatures had no indication of parasitism by Protosiris (only one cell had a young meloid larva).

BIOLOGY OF LANTHANOMELISSA BETINAE URBAN

The nesting biology of this species was original described by Sakagami and Laroca (1988) under the name of Lanthanomelissa goeldiana (Friese) from a nesting site in Castro, Paraná, Brazil. Urban (1995) assigned two females from that nesting site as paratypes of L. betinae when she described the species.

Our observations are from notes we gathered primarily on the campus of the Universidade do Extremo Sul Catarinense, Criciúma, in 2002 and 2003, although information observed at the Curitiba nesting site was similar. At both localities, burrow entrances were discovered on sloping surfaces (figs. 29, 30) close to grassy areas where the floral oil/pollen plant Sisyrinchium micranthum Cav. (Iridaceae) grew. Although both sites were ecologically degraded because of campus landscaping and lawn maintenance, adult bees also occurred at other localities with relatively unaltered landscapes. The presence of this bee on these campuses was possible because of the persistence of Sisyrinchium under such situations.

Early in the 2002 season in Curitiba, we observed males of Lanthanomelissa betinae patrolling flowers of Sisyrinchium and Cuphea gracilis Koehne (Lythraceae) and sleeping in aggregations when females were not yet present (thus indicating protandrous emergence). Fifty-eight sleeping male aggregations of L. betinae were sighted between October 13 and December 18, 2002, all of them along the bank where we found the nests (fig. 30). Most sleeping male aggregations were 10–80 cm above the ground, on the top leaves of four herbaceous plant species: Chromolaena hirsuta (Hooker and Arnott) R.M. King and H. Robinson (N = 27), Eupatorium laevigatum Lamarck (N = 3), Chromolaena pedunculosa (Hooker and Arnott) R.M. King and H. Robinson (N = 18), and Solidago microglossa de Candolle (N = 4) (Asteraceae). Few other males were found between the inflorescences of Cuphea gracilis (N = 3) and Taraxacum officinale G.H. Weber ex Wiggers (Asteraceae) (N = 2). The largest number of the males was observed on October 26, 2002, when 166 males were counted, distributed in 24 aggregations (μ = 6.91 ± 13.2). The largest sleeping aggregations were observed in Chromolaena hirsuta (N = 3; maximum numbers, 53, 39, 39) and Solidago microglossa (N = 1; maximum number, 18). Male activity started to drop by early November and ceased after mid-December. Females presumably slept in their nests.

Males were also previously found by I.A.S. in Criciúma sleeping on flowers of Asteraceae and in flowers of Petunia integrifolia (Hook.) Schinz and Thellung (Solanaceae). On Asteraceae, they were just resting on the flowers with their head partly inserted into the flower. In Rio Grande do Sul, she once observed five males sleeping end-to-end around the ovary at the bottom of the corolla of P. integrifolia.

Nests, each occupied by a single female, were compact (maximum lateral spread of cells at most ca. 10 cm) and shallow (maximum depth ca. 15 cm) with cells close to one another. Dry tumuli (ca. 3.5 mm in diameter) were on the downhill side of entrances. Open entrance tunnels descended obliquely with some turning. Burrow walls were nonreflective and had no special lining. Cells were elongate ovals, 7.0–8.0 mm in maximum length and 4.5–5.0 mm in maximum diameter (based on actual measurements and on cocoon dimensions); thus, their lengths were somewhat more than 1.5 times their maximum diameters, contrasting with the elongate cells of Monoeca haemorrhoidalis. All were arranged singly (i.e., not in linear series). They were broadly rounded at the rear end, and their maximum diameter was one-third the distance from the rear end, forward from which they narrowed gradually to the front entrance, which was ca. 3.0 mm in diameter. Cells seemed to be radially symmetrical around their straight long axis, with one side not being straighter than the opposite one, contrary to the cells of Monoeca as described above. However, the true symmetry of the cells of Lanthanomelissa may have been obscured by their small size. Cell closures on the inside were a deep concave spiral of 3–4 coils.

The long axes of cells ranged from 35 to 80° from horizontal, with the front of the cell always higher than the rear. Cells possessed a conspicuous, waterproof, grayish lining beneath which the cell wall appeared dark, but whether the color resulted from mechanical compression of the substrate or from some applied substance (e.g., floral oils, glandular secretions) is unknown. When cell fragments dried, the lining tended to crack, but flakes could not be easily removed from the wall. When the lining was heated on a glass slide on a hotplate, it did not melt, as did the lining of the Monoeca cells. The source of the lining material is unknown. Laterals were of various lengths, with the shortest being ca. 1 cm long; all were filled with soil after cell closure.

The first provisions brought into a cell were placed as an unshaped mass occupying the bottom of cell (fig. 31). The final provisions were shaped as an elongate, mealy-moist loaf, in one case 4.2 mm long and 2.8 mm high (fig. 32). When newly formed, the loaf was, at least in a few cases, attached by its ventral or posteroventral surface to the cell wall, so that the rear surface was separated from the rear of the cell, and sometimes the anteroventral surface was also separated from the cell wall (fig. 32).

The female deposited her egg on the top surface of the provisions near the front end (fig. 32). Young larvae crawled over the loaf while feeding, and intermediate-stage larvae circled the food mass, which had by this time slumped in the cell so that its posterior end became attached to the rear end of the cell. Provisions at this time developed a central peak as the feeding larva circled the periphery with its dorsum against the cell wall. Information on the larva's final feeding position was not noted.

Externally brown, the cocoon of Lanthanomelissa betinae was 7.0–7.4 mm long (x̄ = 7.3 mm) and 4.5–5.0 mm in maximum diameter (x̄ = 4.8 mm) (N = 7). Each filled the entire cell and had a shape (fig. 33) identical to the inner topology of the cell (i.e., broadly round at the rear end), tapering forward from its maximum diameter to the narrowly rounded or faintly truncated front end. All but the front end of the cocoon was composed externally of fine, densely appressed silk strands. The anterior end consisted of coarser silk strands, openly and loosely arranged (thus, far less dense than elsewhere). The entire inner surface of the cocoon including the front end (fig. 35) was coated uniformly with a smooth layer of pale, almost white feces. Thus, in cross section, the cocoon consisted of two layers, a brown, silken outer layer (with the front end more open and less dense than elsewhere) and a somewhat thinner inner layer of feces.

The cocoons of Lanthanomelissa betinae and Monoeca haemorrhoidalis were similar in that their walls conformed to the similarly shaped cells, and both were composed of dense, fine, brown silk on the outside. However, they differed in a number of distinctive ways irrespective of dimensions: (1) Cocoon symmetry reflected the shape of the cell; thus, the cocoons of L. betinae appeared roughly radially symmetrical around their long axes whereas those of M. haemorrhoidalis were bilaterally symmetrical. (2) Whereas the front end of the cocoon of M. haemorrhoidalis was sharply truncated, with the periphery usually being accentuated by a slightly elevated rim, the front end of the cocoon of L. betinae was narrowly rounded, although on some specimens a small, flattened anterior surface was discernable. (3) Although both taxa start spinning their cocoons before the onset of defecation, L. betinae spins its entire cocoon before applying feces to the inner surface. Monoeca haemorrhoidalis continues to spin layers of fiber while defecating, so that the feces are incorporated into the fabric of the cocoon as well as on the inner surface of the completed silken cocoon. (4) Feces of L. betinae completely cover the front interior surface of the cocoon as well as the rest of the cocoon, whereas the feces of M. haemorrhoidalis are incorporated throughout the cocoon except for the front surface. (5) The front end of the cocoon of M. haemorrhoidalis was far removed from the cell closure, whereas the front end of the cocoon of L. betinae approached the cell closure closely. This feature is the result of the more elongate cells of M. haemorrhoidalis.

Although several nests of Lanthanomelissa betinae with only a few cells were excavated, one nest seemed to contain 21 active cells, another had 11 cells, and a third had 9 cells. However, the first two nests mentioned included many other cells from which the adults had emerged, leaving behind vacated cocoons. This strongly suggests that these main burrows may have been used by more than one generation, a possibility supported by the fact that a single bee pupa7 was encountered in one nest. Either the pupa was a member of the generation from the previous year or some members of the current generation may be bivoltine. So many cells in the nest favor the first alternative. Large nests contained offspring that had already constructed cocoons, freshly deposited eggs, and even open cells still being provisioned. The large range of developmental stages might be explained (assuming that only a single generation comprises the nest contents) either by a rapid development of immatures or, conversely, by a slow rate of emergence, nest construction, and provisioning. These matters cannot be determined by available data.

The observations on the nesting biology of Lanthanomelissa betinae (cited as goeldiana (Friese)) reported by Sakagami and Laroca (1988) concur with ours in that the species seems to prefer a sloping nesting surface, although the slope at their site appears to have been steeper than ours. In both cases, entrance turrets were absent, main tunnels tended to be short and remained open, and a single female inhabited each nest. Laterals tended to be short and were soil-filled after cell closure for both. Whereas cells we observed were arranged singly, they stated that cells were often in series of twos. Furthermore, they indicated that cells tended to be more vertical, whereas we found their orientation more variable, with their long axes ranging from 35 to 80° from horizontal. With both sets of observations, cell lengths were substantially less than twice their maximum diameters, and nests were quite compact and shallow because of short main tunnels and laterals. The provisions in both cases were arranged as loaves attached to the side of the cell, with eggs on the opposite surface from the attachment. They reported that intermediate larvae circled the provisions, so that they did not contact the cell wall, whereas we found such larvae circling the provisions while the food was still attached to the rear of the cell; this difference, however, may be a discrepancy in the ages of the two larvae observed. Sakagami and Laroca found no cleptoparasites associated with their nests.

The floral relationships of Lanthanomelissa betinae were reported by Cocucci and Vogel (2001) and Truylio et al. (2002). The latter found females of L. betinae to be the main visitors of Sisyrinchium micranthum in Rio Grande do Sul, Brazil. Roig-Alsina (1997) described the oil-collecting modifications of the foretarsus found in the genus Lanthanomelissa.

In an unpublished study in 1994, I.A.S. analyzed the provisions from 10 nests of Lanthanomelissa betinae in Rio Grande do Sul and found that all were provisioned with pollen only from Sisyrinchium. Thus, this bee species obtains both pollen and oil from the same plant.

BIOLOGY OF PAREPEOLUS MINUTUS ROIG-ALSINA

Adults of Parepeolus minutus were observed apparently searching for and entering nests of Lanthanomelissa betinae at the nesting areas at both Criciúma and Curitiba. Four postdefecating Parepeolus larvae, three of which came from one nest, were recovered from nests at Criciúma, confirming the host/ parasite association. In external appearance their cocoons (fig. 34) were indistinguishable from those of the host, except that in two examples (the third was not noted) the front end of the cocoon contained no dark, slightly thicker strands of silk. However, from the inside, the cocoons of the two taxa could be immediately recognized because the fecal linings at the front ends of the cocoons of P. minutus were incomplete, leaving a hole through which the open silk fibers of the cocoon were easily visible (fig. 36) and light could be transmitted.

Information on the mode of parasitism of Parepeolus minutus is not available. Other parasites were not observed in the host nests.

OVARIAN STATISTICS

Table 1 provides comparative data on the number of ovarioles and number and sizes of mature oocytes of Monoeca haemorrhoidalis and Protosiris gigas. Although we attempted to dissect the ovaries of Lanthanomelissa betinae from four females, we were unable to retrieve information concerning ovarian anatomy or oocyte development, either because of poor preservation or because of unusual morphology. Females of Parepeolus minutus were not available for dissection.

The egg index of Monoeca haemorrhoidalis was 0.65, whether calculated on the basis of the length of its mature oocyte or of its egg (see Methods and Terminology for methods of calculating the egg index). This value falls in the category of small. The only other tapinotaspidine for which we have information is Lanthanomelissa betinae, reported below. This egg index, based on a single female and her egg, was 0.99, close to the upper limit of category medium. It is unknown if such a range in values within a single tribe of solitary bees is unusual.

The surprising statistics in table 1 are the range in values of the egg indices between Protosiris gigas and Monoeca haemorrhoidalis. Cleptoparasitic bees tend to have lower egg indices than do those of solitary bees (Rozen, 2003), but here we find that the egg length of the cleptoparasitic bees is nearly as long as that of its solitary host (compare figs. 37 and 38). The egg index of P. gigas is 0.82 based on the mature oocyte (table 1) and 0.86 based on a sampling of eggs from females (see Egg of Protosiris gigas Melo), with both values being well within the medium category.

We can partly account for this apparently anomalous situation in a number of ways. First, the egg of Monoeca haemorrhoidalis is curved and hence measures shorter than its curved length, in contrast to straighter egg of Protosiris. Second, the egg of M. haemorrhoidalis is more uniformly thick; it does not taper posteriorly, as does the egg of P. gigas. Third, the host, a robust bee, has a very wide mesosoma compared to its body length, whereas the cleptoparasite is slender, with a narrow mesosoma. Hence, the denominators in both calculations of the egg indices are skewed. Finally, the egg of P. gigas is obviously much thinner than that of its host. When the two eggs are viewed side by side (figs. 37 and 38), clearly the cleptoparasite has a considerably smaller egg volume. Part of the problem is with our having used egg length and intertegular distance, two linear statistics, to evaluate volume.

While this explanation may partly explain the large size of the egg of Protosiris gigas, other Osirini tend to have egg indices higher than those of many parasitic bees: Epeoloides coecutiens (Fabricius), 0.70 (small) (Rozen, 2001); Osirinus lemniscatus Roig, 0.79 (medium) (Alexander, 1996); and Parepeolus aterrimus (Friese), 0.88 (medium) (Alexander, 1996). Thus, it appears that osirines as a group do not have egg indices that fall into the dwarf category, as do many parasitic bees. Now that we know that females of P. gigas deposit their eggs into a closed host cell, we can probably interpret these relatively high indices as indicating that all members of the tribe invade a closed host cell for ovipositioning; as Rozen (2003) pointed out, cleptoparasitic taxa that invade closed host cells tend to have larger eggs than do those that hide their eggs in cells still being visited by host females.

The remaining statistics in table 1 are normal. Cleptoparasitic bees tend to have more mature oocytes than do solitary bees, presumably because of the need for cleptoparasites to lay eggs whenever appropriate host cells are discovered. Four ovarioles per ovary is plesiomorphic in the Apidae (Michener, 2000).

IMMATURE STAGES

In this section we describe the immature stages of Monoeca haemorrhoidalis, Lanthanomelissa betinae, Protosiris gigas, and Parepeolus minutus, respectively, to the extent that collected material permits. Most of the information presented under Methods and Terminology is applicable to this section. In the descriptions of the intermediate larval instars of both M. haemorrhoidalis and P. gigas, dates and collector names refer to exemplars that were cleared for careful study.

DISCUSSION

The reader is referred to “Synopsis of the nesting biology of the Tapinotaspidini bees” (Aguiar et al., 2004) for a comparison of the nesting biology of the members of this tribe based on the information from this study and from previously published accounts. As the authors pointed out, members of the tribe nest in horizontal ground, earthen banks, and rotten wood. In addition, Chalepogenus rozeni Roig-Alsina has been observed 26 km south of Vicuña, Elqui Prov., Chile, nesting in deep cracks in the soil, as much as 30 cm below the surface. These nests have short main tunnels extending obliquely downward and cells appearing within 5 cm of the vertical crack surfaces (Rozen, unpubl. data). Such a nesting site can probably be considered akin to nesting in an earthen bank.

It would be interesting to compare the nesting biology and immatures of the Tapinotaspidini with those of the Exomalopsini because, until recently, the included taxa were placed in the same tribe, the Exomalopsini. However, such a comparison would be premature because so little has been published concerning the immatures and biologies of the various genera. Nonetheless, one biological feature seems to distinguish the Tapinotaspidini, that is, their behavior of collecting floral oils and their ability to transport and manipulate them.

The nonparasitic Ctenoplectrini do collect floral oils (Michener, 2000) and the postdefecating larva of one species has been described (Rozen, 1978). While this larva is remarkably similar to those of the Tapinotaspidini, the similarities (body form, details of the head and mouthparts) appear to be plesiomorphic. To the extent known, the Ctenoplectrini are cavity nesters, unlike the Tapinotaspidini. However, a more detailed analysis of larval features is warranted to determine if synapomorphies can be identified that show a relationship between these two tribes in light of some of the analyses of Roig-Alsina and Michener (1993) that show them to be sister groups.

In attempts to determine the nonparasitic ancestor that gave rise to cleptoparasites, host lineages are often identified as being likely ancestors. Is there evidence that the Osirini are derived from a tapinotaspidine-like ancestor? The fact that osirine larvae can metabolize floral oils mixed with pollen may be a synapomorphy of the two tribes. Mature larvae of the two tribes show broad resemblances, but most of the features appear plesiomorphic and therefore are of little help in evaluating relationships.

The first instar of Protosiris gigas is obviously adapted to seek out and destroy host immatures and competing cleptoparasitic immatures. Its anatomy and biology therefore can be used to address the question whether Protosiris (and other Osirini) are closely related to other cleptoparasitic groups. Rozen (1991: table 1) identified 21 features of first instars that varied among the Nomadinae, Protepeolini, Melectini, Rhathymini, Isepeolini, and Ericrocidini. These same features are presented in table 2 here, to which two more columns have been added for Coelioxoides (Tetrapediini) and Protosiris (Osirini). Primitive features are scored 0, and derived features are scored 1, 2, and so on. Two features of the Isepeolini, then unknown, are now known: (1) females of Isepeolus introduce their eggs into host cells that are still open (scored 1), and (2) when deposited, the eggs are flat against the cell lining (Rozen, 2003). The latter is also scored 1, to be the same as “inserted in the cell wall” because of indirect evidence that the related Melectoides deposits its eggs this way in the cells of Canephorula (Michellete et al., 2000; Rozen, 2003). Data for Coelioxoides come from Alves-dos-Santos et al. (2002). It is now possible to analyze the behavioral and anatomical features of the first instar of Protosiris (the only member of the Osirini studied so far) with those of seven of the nine other cleptoparasitic lineages of the Apidae (Rozen, 2000b). Two cleptoparasitic lineages, Exaerete and Aglae (both Euglossini), are excluded, Aglae because its mode of parasitism is unknown, Exaerete because its second instar (not the first) is hospicidal (Garófalo and Rozen, 2001).

Table 3 shows the features of egg deposition and first-instar anatomy that Protosiris shares with each of the other seven lineages of cleptoparasitic apids. Based on the number of shared features, Protosiris and the Ericrocidini are the most similar, but they share no feature uniquely, a fact that suggests that these character states are not particularly strong. The complete sclerotization of the labiomaxillary region of the ericrocidines contrasts sharply with the totally membranous labiomaxillary region of Protosiris, as does the dorsoventrally flattened, large, projecting ericrocidine antenna with the small antennal papilla of Protosiris. The ericrocidine sclerotized labrum fused with the clypeus is also unlike the nearly membranous labrum of Protosiris. These strong dissimilarities do not support the idea that these two taxa are related. The only possibly unique character shared by Protosiris with any of the lineages is character 6, the presence of spinulae found also in Melectini. While spinulae have been a unique feature of melectines, present in all of them, the microscopic morphology of a spinula (Rozen, 1991: fig. 22) is not that of a simple setiform sensillum as appears to be true in Protosiris. Thus, while the spinulae of the Melectini and the band of setiform sensilla of Protosiris occupy the same position on the head capsule, they may not be homologous. The numerous differences between the first instars of these two groups also argue against a close relationship (table 2). One is left with the impression that Protosiris and thus the Osirini represent an independent origin of cleptoparasitism in the Apidae. The presence of the unique, ventral, integumental folds on the anterior edges of segments 3–7 seems to support the concept that many hospicidal bee larvae must be able to move with agility in order to find and kill host eggs (or larvae) and those of competing cleptoparasites, and that Protosiris (or its ancestor) found a new way to crawl. It is not out of the question that the Osirini arose from a tapinotaspidine-like ancestor despite the lack of evidence supporting such a conclusion at present.