Stelis ater (Hall and Ascher, 2010)

Rozen, Jerome G. & Hall, H. Glenn, 2011, Nesting and Developmental Biology of the Cleptoparasitic Bee Stelis ater (Anthidiini) and Its Host, Osmia chalybea (Osmiini) (Hymenoptera: Megachilidae), American Museum Novitates 2011 (3707), pp. 1-40 : 16-26

BIOLOGY OF STELIS ATER

EGG DEPOSITION: We discovered eggs of Stelis ater on the front surface of the provisions and others deeply buried in the provisions. Because we detected no mechanism whereby eggs could have been inserted after cell closure, we assume that they are introduced only into cells that have yet to be closed by the host female, as also reported for other Stelis by Torchio (1989b). It is unclear how the eggs are buried since there are no special modifications at the end of the female metasoma of this species that would assist in inserting the eggs deeply. A likely possibility is that all eggs are deposited on the existing exposed surface of provisions and, if the provisions are incomplete, deposited eggs are buried when foraging host females return to add more larval food. However, this assumption needs confirmation, because in other cases of bee cleptoparasitism, we think that returning host females search out and destroy cleptoparasite eggs. It seems unreasonable that Osmia chalybea females would not do so.

According to observations by P.H. Torchio (personal commun., October 25, 2010), Stelis montana appears to modify its egg laying behavior to match that of its host. When parasitizing Osmia (Cephalosmia) californica Cresson and O. (Cephalosmia) montana Cresson , it constructs an egg chamber within the provisions, as do those hosts. However, when parasitizing O. (Osmia) lignaria Say , it lays its egg on the surface of the provisions, as does that host. He commented that all three Osmia species and Stelis montana attach their eggs to provisions only by their posterior ends, and they all begin feeding as second instars.

DEVELOPMENT: Stelis ater , like its host and S. montana ( Torchio, 1989b) , has five larval instars evident from accumulated cast skins. We assume that the first is pharate because of the lack of sclerotization of its head capsule and because mandibles were not detected on any first instar skins, although spiracles were. Although we did not find a band of spicules above the spiracular line on each side of the body of first instars, we would not be surprised if future studies identified such bands; we simply had insufficient material.

J.G.R. had anticipated that the last larval instar would be found to be the prime hospicidal stage because of the extreme modification its head capsule and mouthparts as found in figures 83–85, 95, and 96. However, our study of this species clearly indicates that most stages are capable and behaviorally inclined to attack host immatures and one another if they encounter them. Although most cells contained a single cleptoparasite, not uncommonly two occurred in a cell, and in two cases three cleptoparasitic larvae coexisted apart from one another in a cell. Figures 35–41 show various examples. This variability in timing of encountering host or competing cleptoparasitic larvae results from the diversity of where the parasite eggs are deposited in the provisions and the now-realized fact (elaborated in the Discussion) that early stage megachilid larvae are incapable of crawling from the position where the egg is deposited.

Hence, if a recently hatched cleptoparasite is far removed from the host immature, both it and the host will increase in size substantially before they meet. A case in point was the discovery of a fifth instar of Stelis ater feeding on the surface of the provisions while a fourth instar of Osmia chalybea fed alongside it. About an hour after discovering the two larvae feeding side by side, we returned to discover the fifth instar of S. ater repeatedly chewing into the side of the host larva, an activity that continued for more than 10 min. The attacking larva then fell away from its host but continued to defecate until preserved. The wound on the side of the host larva was large and obviously fatal (fig. 42). The host larva did not react in any way to the attack, suggesting it had no defense response to deflect or avoid its attacker. Examination of the provisions afterward revealed that the egg of the S. ater had been deposited deeply, close to the cell wall. The pathway to the surface was a channel in the provisions close to the cell wall extending to the surface of the provisions. Near the egg-insertion point the channel was narrow but gradually widened as it progressed forward to the surface. The cast larval skins of instars 1–4 were layered against one another, and the flattened mass was pressed against the channel wall. At the farthest rear of the channel, presumably where the egg had been deposited, were two fecal pellets associated with the posterior end of the larval exuviae. Because only last instars defecate, these two pellets indicated that the posterior end of the fifth instar had been positioned at the approximate point where the egg had been deposited. One of these fecal pellets was presumably encased in the posterior end of the cast-off exoskeleton on the fourth instar, suggesting that it was the meconial mass first extruded after the intima of the fourth instar’s hindgut had been withdrawn from the fifth’s alimentary tract, allowing passage of fecal material from the midgut to the exterior.

A similar attack occurred when we found a moderately small last larval instar of the host and a Stelis ater of similar stage feeding next to the cell wall in the provisions. We moved the cleptoparasite with forceps and placed it by the host, with the result that we were able to watch the attack involving the repeated opening and closing of sharp-pointed Stelis mandibles. On the other hand, figure 39 represents the situation where two Stelis ater eggs were presumably deposited on the surface of the provisions close to the host egg with the result that the two early cleptoparasitic instars attacked the young host larval almost immediately. Presumably the two attackers would then have battled one another had we not preserved them.

Motions of an attacking early instar are not particularly agile: They consist merely of a repeated but routine opening and closing of mandibles usually unaccompanied by a thrusting or turning head (although fig. 41 seems to be an exception). When a mandibular apex catches onto some part of the victim, this action is followed by continuing mandibular opening and closing until the victim is mortally damaged. After becoming the fifth instar, the larva begins defecating, with the fecal material extruding slowly from the anus as a single, coiling, grayish-yellow strand, slightly more than 0.25 mm wide (fig. 14), in sharp contrast to the short cylindrical pellets of the host. The occurrence of such coils in a cell permits certain identification of Stelis , as Torchio (1989b) reported a similar phenomenon with respect to feces of S. montana . Of course, we do not know whether all species of Stelis produce such feces. As the larva increases in size, the strand often breaks into sections of irregular lengths though rarely as short as those of the host. By the time of cocoon spinning, feces have become very dark and more fusiform (note mixture of pale and dark feces in figs. 15, 16). After cocoon spinning, the larva becomes totally quiescent, with its head end closest to the front end of the cocoon. Its flaccid body no longer responds to being touched by forceps FIGURE 43. Diagram of front end of cocoon of Stelis ater, showing elements of filter area, side view, or by being moved, unlike the host larva. Ste- with arrows pointing direction of inflow air through lis ater is now in diapause. The first parasite filtered gas-exchange system of nipple.

pupa to emerge from a cocoon stored at ambient room temperature was preserved January 17, 2011, by which time almost all hosts had emerged as adults.

COCOON: Externally the cocoon of Stelis ater is similar to those of other species of Stelis that have been described or pictured (see Parker, 1986; Parker et al. 1987; Rozen and Kamel, 2009, and references therein). All are characterized by having the anterior end drawn out into an elongate, tapering nipple (figs. 15, 16, 43, 68–70). For S. ater , six cocoons in U-shaped tunnels were oriented with their long axis at about 45° from horizontal. All had their front ends pointed upward and all but one had the front ends pointed toward the tunnel entrances. The nipple is not hidden in a mass of fine silk as is the case with Osmia chalybea , although it has a thin surface covering of whitish strands. The body of the cocoon, but not its nippled front end (figs. 15, 16), is submerged in a mixture of uneaten pollen and fecal pellets loosely held together by strands of silk. Fecal pellets closest to the cocoon tend to be short and dark, as mention above. The cocoon is about 8.5– 12 mm long and approximately 4.7– 6 mm in maximum diameter (N = 4).

The fabric of the body of the cocoon consists of three distinct layers (figs. 43, 52, 53). The outer layer is a thin sheet of semitransparent grayish-tan silk, more or less covered by loose strands of white silk (fig. 15). This layer is loosely connected by tan strands to the thick, ridged middle layer, which is opaque, very dark brown to nearly black, leathery, and strongly resistant to compression. The source and composition of this layer was at first unknown, since silk fibers could not be detected in it. However, Torchio’s (1989b) description of cocoon construction in Stelis montana almost certainly explains that the material comes from the larva’s anus and is the excreta of the Malpighian tubules. 3 The thin inner layer consists of shiny strands of silvery silk (fig. 44) when first deposited, that later sometimes become faintly coppery against the dark background of the middle layer.

The outer layer of the cocoon also covers the entire nipple, giving the nipple end its characteristic tapered appearance, with its length gradually narrowing to an acutely rounded, conical apex (figs. 15, 16). Viewed with an SEM, the fabric of the outer surface is dense but fenestrated (figs. 54, 55). Immediately under its apex is a dense mass of loose, pale tan silk fibers (figs. 56, 57) that surrounds and covers the apex of the strongly projecting middle layer of the cocoon that is continuous with the inner layer of the nipple. The latter is an externally dark, hollow tube (figs. 43, 48, 49, 58, 70), apically about 0.7 mm across and with a central circular opening about

3 This observation by Torchio (1986b) suggests that the dark substances combined with sand grains coating the inner surface of cocoons of Fidelia villosa Brauns ( Rozen, 1970) and Pararhophites orobinus (Morawitz) ( McGinley and Rozen, 1987) , both in the basal clade of Megachilidae , may also be from Malpighian tubules. It is worth noting that not all Stelis cocoons have a middle layer that is opaque, dark brown to black: Parker et al. (1987) reported that cocoon fabric of S. (Dolichostelis) rudbeckiarum Cockerell was amber colored, “less dense and the overwintering prepupal larvae were visible” through it.

of filter area. 50. Close-up of inner layer of nipple with one side removed showing pale cream color of silk and glassy surface, approximate lateral view. 51. Anterior end of cocoon viewed from inside, showing white inner lining to cocoon.

0.33 mm in diameter. This opening is the outer end of an open passageway leading to the interior of the cocoon. The tube is approximately 0.75 mm long. About halfway to the base of the tube, the passageway widens into an inner chamber about 1.0 mm in diameter at its base. Its base is covered with a disc of thick, reddish fibers that screens the passageway to the lumen of the cocoon (figs. 58, 60).

When the outer layer of the cocoon is removed, the middle layer is seen to connect flawlessly to the inner layer of the nipple externally (fig. 45). However, they are apparently made from two different materials. Whereas the middle layer of the cocoon wall is entirely blackish brown throughout, the inner layer of the nipple is dark only on the outer surface. Internally it pales to a creamy white toward its apex (fig. 50). Furthermore, its inner surface develops a glassy consistency with a transparent appearance (fig. 50). The ripplelike marks on the wall of the inner passageway of the nipple (figs. 50, 58, 60–63) suggest that it may be a glassy secretion of silk. This suggestion is reinforced by an earlier observation (seen in one of the observation nests) of a larva with its head moving inside the outer layer of the cocoon nipple, an unexplained behavior at the time. Exactly where the black material of the middle layer of the cocoon transforms into the glassy material of the nipple is difficult to identify because of the thinness of the cross section. Future observations of spinning larvae should be able to explain the phenomenon. Immediately behind the screen of the nipple is the inner layer of the cocoon wall with multiple small openings scarcely, visible without an SEM (figs. 64, 65); it is present behind the nipple end but not elsewhere (fig. 66). Incoming air (fig. 43, arrows) obviously enters through the opening in the outer layer, then through the open passageway of the nipple, through the screen, and finally through the openings of the inner cocoon layer. Gas exchange takes place along the same route. The study by Rozen, Rozen, and Hall (in prep.) currently underway shows that the cocoon of Stelis ater is even more airtight than that of Osmia chalybea .

We think that the airtight nature of cocoons of both species is derived from the thin inner cocoon layer with openings through this layer only at the nipple end. If true, then the smooth surface to this layer as seen in the SEM micrographs (figs. 27, 29, 52, 56, and 66) is evidence of air tightness. Although our original assessment was that this surface is formed by fusion of silk strands, we cannot ignore the possibility that it is a composite material, i.e., some other secretion spread over a foundation of silk strands.

COMPARISON OF COCOONS: A comparison of the cocoons of Osmia chalybea (the host) and Stelis ater (the cleptoparasite) is of interest if for no other reason than we have often devoted little attention to the function and structure of bee cocoons, 4 although there is a growing body of knowledge concerning the properties and use of silk by honey bees ( Apis mellifera Linnaeus ) (e.g., Sutherland et al., 2007, and references therein). It seems likely that the silk of O. chalybea and S. ater will be made of proteins in a coiled coil arrangement, as seems to be characteristic of aculeate Hymenoptera ( Sutherland et al., 2007) . In summary, cocoons of both species as

4 Contrary to this statement, Michener (1955) and Rust and Thorp (1973) provided rather detailed and thoughtful descriptions of the cocoons of Stelis lateralis and S. chlorocyanea , respectively, which seem in many ways comparable to our account of the cocoon of S. ater . However, without specimens to compare side by side, the differences cannot be evaluated and resolved.

described above provide larvae with an environment that screens out deleterious parasites, parasitoids, and predators and that protects the larva from dehydration with a nearly airtight cocoon fitted with a filtering nipple end to allow for gas exchange. At the same time, the structures of the two cocoons, particularly of the filtering devices, are significantly different and demonstrate the versatility of silk and the diversity of behaviors of larvae of the two species in applying the material. SEM studies (e.g., Rozen and Buchmann, 1990) strongly hint that cocoons of other bee taxa, though structured differently, may well afford similar protection.

DISCUSSION: There would appear to a selective disadvantage for a megachiline cleptoparasitic bee to have a host larva feeding on provisions that the cleptoparasite needs for survival. This problem was first implied when Baker (1971) correctly identified the third instar of several species of Coelioxys to be the chief hospicidal stage. J.G.R. in a study of Stelis elongativentris Parker ( Rozen, 1987) found that even the fifth instar was equipped to kill the large Ashmeadiella host larva. Recently, larvae of C. chichimeca Cockerell were discovered with their posterior ends attached to the egg insertion point until they reached the fourth instar ( Rozen et al., 2010a). The only known hospicidal first instars of cleptoparasites in the family are those that emerge from eggs deposited on host eggs, i.e., Radoszkowskiana rufiventris (Spinola) ( Rozen and Kamel, 2007) and C. (Allocoelioxys) afra (Ferton, 1896) and C. coturnix Pérez ( Rozen and Kamel, 2008) . We propose that the underlying reason is that earlier instars of the cleptoparasites cannot reach their host because of an inability to crawl. Examination of non-cleptoparasitic Megachilinae (if not all Megachilidae ) reveals that their early instars also lack that ability. Note the sessile posture of Osmia chalybea until it reaches the last larval instar (35, 36).

To the extent known, all Megachilidae spin cocoons; these structures may also afford protection from desiccation and parasite/parasitoid invasion. However, cocoon spinning is not a universal behavior of bees. Indeed, no other family consists solely of cocoon-producing species, and in two families ( Stenotritidae and Andrenidae ), no cocoon spinners have been reported. Future biological investigations should consider the mechanisms by which noncocoon spinners cope with brood parasitism/predation and desiccation.

Figures

FIGURES 35–40. Immatures of Stelis ater. 35. Egg deep in provisions; eggs were encountered both in the provisions and on the front surface of the provisions, as discussed in text. 36. Early instar (arrow) deep in the provisions and feeding larva of Osmia chalybea (right) on front surface of provisions. 37. Egg of Stelis ater (arrow) and probable first instar of Osmia chalybea (left) on front surface of provisions with trachea now visible. 38. Same cell some time later with Stelis ater (arrow) now early feeding instar having not yet encountered Osmia chalybea, which has become an intermediate instar. 39. Early instar of Osmia chalybea being attacked simultaneously by two early instars of Stelis ater; one of which (arrow) is barely visible to the left of the others. Presumably one S. ater larva would later kill the other, since we never found more than one cocoon in a cell. 40. Two feeding early instars of Stelis ater (arrows) from eggs deposited deep in provisions.

FIGURE 41. Early fifth instar of Stelis ater attacking fourth instar of Osmia chalybea. FIGURE 42. Wound (arrow) in same fourth instar caused by attack.