Hesperapis rhodocerata: Behavioral Biology, Egg, and Larval Instars, Including Behavioral and Larval Comparisons with H. larreae (Hymenoptera: Melittidae: Dasypodainae)

ABSTRACT This paper reports on a large nesting site of the ground-nesting solitary bee Hesperapis (Carinapis) rhodocerata (Cockerell) from southern New Mexico first discovered in the late summer of 2010 and active again in late summer 2015. Because the site was visited annually during intervening years without observation of any specimens, the species is believed to sustain a multiyear diapause that is broken in response to rain. It is judged to be univoltine, and females at the site collect pollen from Heterotheca (Asteraceae). Nests are briefly described as are the nest-digging behavior and pollen-transport system of females. The feeding behavior of larvae involves grazing on the surface of the food sphere, thus reducing its diameter. This is accomplished with the aid of paired ventral tubercles on each of the three thoracic and first eight abdominal segments and a single median ventral tubercle on the ninth abdominal segment. The second and last larval instars are described and illustrated. The first instar is essentially identical to the second instar except for size. Mature larvae are similar to other known Hesperapis larvae. The strongly curved egg of H. rhodocerata is described and illustrated with a diagram and SEM micrographs of the micropyle. Because the last larval instar does not spin a cocoon and freshly constructed brood cells are unlined by females, questions are evoked concerning humidity control and parasite exclusion during the long diapause of mature larvae. This information is compared with and found in some ways different from that uncovered in an earlier study of H. (Amblyapis) larreae Cockerell. It is hypothesized that the clear thin transparent material covering the postdefecating larva of H. rhodocerata may function to inhibit desiccation and furthermore may be the same material that hardens and waterproofs the cell walls of other congeneric species including H. larreae, thereby serving a similar function but in a different way. Because too few mature larvae of H. larreae had been collected at the time of drafting the study of that species, their description is added here as an addendum.


INTRODUCTION
Adults of Hesperapis (Carinapis) rhodocerata (Cockerell) are moderate in size compared with other Hesperapis species and are known from a number of localities in southeastern Ari¬ zona, southwestern New Mexico, and adjoining areas in northern Mexico (Ascher and Picker¬ ing, 2015). The first nesting site of this species was discovered at 2 mi southeast of Willcox, Cochise Co., AZ, where the nest-digging behavior of females was described, although nests themselves were not explored (Rozen, 1987). The second nesting site was found on August 28, 2010, in a remote area containing few human habitations at 28 mi south of Animas, Hidalgo Co., NM. It was discovered by the author, John S. Ascher, and other attendees of Bee Course 2010. Female bees were digging tunnels in a broad area, perhaps 75 m in diameter on slightly sloping ground just west of low hills ( fig. 1). The ground surface was extensively covered with grass, low growing herbs, and scattered yucca plants. In addition, the area supported a large amount of Heterotheca (Asteraceae), the pollen source of the large nesting population of H. rhodocerata. Intriguing was the observation of numerous individuals of a rather large species of the cleptoparasite Sphecodes (Halictidae), whose body size seemed to match that of H. rho¬ docerata. Could this be a nest parasite of H. rhodocerata?
In an attempt to explore this question we excavated several nests the same day ( fig. 2) and learned that the nests were difficult to follow because tunnels twisted and turned and cells were difficult to associate with their nest tunnels. Although we found no evidence of Sphecodes in cells, we did recover two eggs of H. rhodocerata (described below) and some early larval instars of H. rhodocerata. Because the nesting site was extensive and densely populated, it would likely persist and therefore suggested that it could be explored further at about the same time the following year.
After the end of the Bee Course the collection dates of specimens of this species were checked in the American Museum of Natural History (AMNH) and in Ascher and Pickering (2015). Adults have been collected only during August and September. This indicates that H. rhodocerata is a late summer, univoltine species (later also confirmed by J. Neff, personal commun., X-25-2015). Although this observation provided more promise that this large population  In several parts of this paper, I refer to a work entitled "North American Melittidae (Doc. 143153)," comprising pages 14-96 of typewritten manuscript by the late Gerald I. Stage.2 It pertains to the biology of the genus Hesperapis and the larger manuscript ends with a series of tables (pp. 406-410). Stage presented it to me years ago before the publication of Rozen and McGinley (1991), therein referred to as "Stage, G.I., and R.R. Snelling, unpublished ms.
A revision of Nearctic Melittidae: the subfamily Dasypodinae (Hymenoptera: Apoidea)." It was edited by him with Snelling s name written in ink by Stage, suggesting that it would have been published jointly by them (Stage and Snelling, ms). It covers the following subtopics: life history, nest note, mating behavior, nest construction, nest architecture, provisions, for¬ aging behavior, and sleeping behavior with respect to seven species, two of which presumably have yet to be described.

METHODS
For study, all preserved larvae were cleared in an aqueous solution of sodium hydroxide after heads were removed from bodies, stained with Chlorazol Black E, and then examined and stored in glycerin on well-slides. The egg was critical-point dried and coated with gold/palla¬ dium before being examined with an Hitachi S5700 scanning electron microscope.
Scale bars on all diagrams =1.0 mm.

NESTING BEHAVIOR
Nest openings were scattered throughout the area mostly on flat surfaces between plants, usually surrounded by loose, course soil. Because a large species of Agapostemon was also nest¬ ing throughout the area, it was impossible to differentiate between nests of the two species unless a female was observed entering or cells were uncovered holding identifiable immatures.
Nest Structure and Configuration: To best understand the configuration of the nest of a ground-nesting bee it is desirable to select a single nest well separated from other nests and to carefully dissect the soil from one side of the descending main tunnel, as detailed in Ramos and Rozen (2014) for Psaenythisca wagneri (Vachal). Thus, confusion with neighboring nests is avoided. However, the area selecting for excavation was chosen because the abundance of nest entrances promised recovery of numerous cells of H. rhodocerata. Consequently, infor¬ mation concerning the nest is limited. It consisted of branching, mostly open tunnels 5 mm in diameter, twisting and turning as they extended downward. Ovoid cells about 15 mm long and 10 mm in maximum diameter were arranged singly and approximately horizontally, mostly between depths of 21-36 cm below the surface. Their walls were rough, without any evidence of a special lining. In general nests were therefore similar to that diagrammed for H. (Hesperapis) trochanterata Snelling (Rozen, 1987: fig. 3).
Nest-digging Behavior: As first reported by Rozen (1987), nest excavation by females of Hesperapis rhodocerata and H. trochanterata near Willcox, Cochise Co., AZ, involves an extremely rapid flinging of surface sand backward under their bodies at nest entrances. This behavior also occurs when a foraging female returns to her nest entrance. The result is a tumu¬ lus of fine loose sand that widely surrounds the entrance except at the very center from where the female discharges the soil. This was photographed for H. trochanterata (Rozen, 1987: fig. 2) where the sand was dry, uniformly fine, so that the tumulus formed a smooth mound. At that time it was determined that the females forelegs dug sand from the entrance and the hind legs flung the excavated sand backward so rapidly that the exact motion of the hind legs blurred. Now with the advent of smartphones (e.g., iPhone 6sTM)) with slow-motion video capabil¬ ity, a more detailed analysis was forthcoming for H. rhodocerata at 28 mi south of Animas, In the video, the returning female H. rhodocerata had both scopae filled with pollen, and the nest entrance was clogged with a mixture of sand and gravel obstructing her reentry. With her head partly inserted into the entrance tunnel, she rapidly moved her forelegs, thereby rak¬ ing the sandy mixture from the hole toward the ground surface under her body ( fig. 5). Her mid legs, partly flexed, extended laterally forming an anchor, so that she maintained her posi¬ tion relative to the entrance hole. The bracing is required to establish a stable platform to counteract the rapid action of her front legs and the strong strokes of her hind legs. Without slow motion the strokes were apparent because they resulted in flying sand and pebbles, but with slow motion, the rhythmic strokes are seen as a repetitive, nearly simultaneous perfor¬ mance by both hind legs. It starts with the tight folding of the femoral-tibial joint on each side of the body (fig. 5), followed by the forceful unfolding of both legs backward along the side of the metasoma (figs. 6, 7) and then outward (figs 7, 8) as she flings sand and pebbles backward and outward primarily from the troughs of both basitarsi. Immediately following this action she swings both legs back (figs. 8, 9) to the starting position (fig. 5) for the next fling. As she continues this routine, her body direction gradually changes relative to the entrance hole so that the tumulus tends to accumulate on all sides of the hole.
Pollen Transport: Although females acquire pollen on many surfaces of their bodies because of long body vestiture, the tibial scopae are the structures on which females accumulate large agglutinated masses of pollen ( fig. 3) to be transported to the nest. Because scopal hairs are restricted to the anterior surface of the hind tibia, these masses are found there and do not sur¬ round the tibia. Very large tibial masses seem to extend apically over the basal part of the dorsal surface of the basitarsus, but it is uncertain whether the setae there are sufficiently long to hold the mass. However, the anterior row of long setae bordering the dorsal trough appears to fence off the pollen from invading the setal trough of the basitarsus. It is unknown how many provi¬ sioning trips are required to form the final ball of provisions, but the total amount is shaped into a sphere that is approximately 5.3 mm in diameter and placed on the cell floor ( fig. 11).
Larval Behavior: Numerous observations indicate that the size of the feeding larva increases as the diameter of the food sphere diminishes. However, the provisions do not change shape, thus indicating that the larva is feeding throughout the surface of the sphere. Masses, fresh as well as partly eaten, randomly collected from nest excavations during the field season in 2015, ranged from 5.3 mm in diameter down to 2.6 mm but always retained their approxi¬ mate spherical shape. The larval movements seemed unusually slow. The following description, referring to an unnamed species in the unpublished manuscript by Stage, is consistent with my fragmentary observations: The strongly curved, cylindrical egg is placed on top of the pollen ball in such a way that only its ends are in contact with it.... The duration of the egg stage is not known but presumably is short as in most other bees.
Upon eclosion the small larva starts feeding near the top of the pollen ball.... While still small it gradually works its way down and around the pollen ball until it achieves a characteristic position curled around and under the pollen ball.... At this time the larva is in a C-shape with one of its sides, not its dorsum, against the floor of the cell and its venter against the pollen ball. In this position the larva continues to feed but at the same time it slowly rotates the pollen ball by a constant twitching motion of the terminal segments of the abdomen. As the larva becomes larger this action tends to lift the pollen ball so that it becomes entirely supported by the larva. The effect of this unusual feeding behavior is that the pollen ball is evenly grazed and remains nearly spherical until it has been almost completely consumed. (Stage and Snelling, ms: 18) The anatomy of early instars, described below with paired ventral body tubercles and spiculate ventral integument, would seem to be suitable for crawling over the surface of the sphere and then for lifting it up and rotating it while feeding.
Nest Cell Environment: A series of recent studies points out that bee cocoons function to maintain appropriate humidity around the diapausing larva and developing pupa while preventing attacks by nest parasites and predators (e.g., . It is likely that special cell linings constructed by nest-making females form a partially waterproof bar¬ rier and also contribute to humidity control. Larvae of H. rhodocerata as well as all other Hesperapis (and presumably those of Capicola and Dasypoda) do not spin cocoons Rozen and McGinley, 1974). Furthermore, females of many (though presumably not all3) Hesperapis whose nests have been examined do not provide a specially prepared, waterretardant lining to their brood cells (Rozen, 1987;Stage and Snelling, ms.). How is the diapausing larva protected from desiccation between flowering seasons of the host plant?
Perhaps the following observations hint at an explanation.
A last larval instar was preserved as a postdefecating form on IX-15-2015. When brushed three weeks later to remove sand grains prior to being illustrated, its body surface was found to be completely covered by a thin transparent coating, which floated away from the surface as thin, nearly transparent flakes in the preservative. This solid material had not been detected until the brushing. Might it have to do with maintaining body moisture? On another specimen, this material was carefully removed from around several spiracles. While maintaining the sur¬ face feature surrounding the spiracular opening, the opening itself was not covered, i.e., it remained an aperture ( fig. 35). Almost identical observations had been recorded regarding H. trochanterata (Rozen, 1987), although then the material was described as "tannish." One won¬ ders if this coating helps to retain body moisture during the long larval diapause. With both species it is evident only on the postdefecating larva. Still unknown is the source of this coating and chemical nature of the substance.
A more recent study pertaining to the behavioral biology of H. (Amblyapis) larreae Cock¬ erell (Rozen and McGinley, 1991: 5-6) presented a somewhat different story. It stated that recently constructed cell walls of this species tended to be slightly more consolidated than the substrate. Small sections of wall could often be carefully teased from the substrate.... Either after feeding or perhaps shortly before finishing, large larvae produced a substance (source unknown but perhaps anal or salivary) that impregnated the cell wall and closure so that these structures became strong, took on a dark wet' appearance, glistened in places, and became water-retar¬ dant.4 This substance possessed no pollen grains (at least at first) and a section of impregnated cell wall did not 'dissolve when submerged in water for several hours.
Because of their new strength, cell walls and closures in this condition were extricated intact from the substrate.
In the same article: "Fresh walls were not waterproof and immediately absorbed water droplets. They were smooth, dull on the surface, and gave no hint as to what substance (if any) accounted for their slightly greater strength than the substrate. The soil of the cell wall was uniformly fine-grained in sharp contrast to the irregular particle size of the surrounding sub¬ strate. Hesperapis females are apparently capable of sorting out fine particles to construct the 3 Stage and Snelling, (ms: 70) reported that H. (Hesperapis) rufipes (Ashmead) and H. (Amblyapis) ilicifoliae (Cockerell) each had a cell "with a hard, smooth wall that was constructed on the crude surface of the coarse, heterogeneous substrate and that was apparently smoothed with the tongue and hardened with saliva." However, there is no indication in their report that cell walls at least of some Hesperapis are modified by the mature larva (Rozen and McGinley (1991). 4 A small piece of the cell wall collected in 1990 and preserved in the AMNH was placed in water for three days as this paper was being drafted. The piece remained unchanged, attesting to the stability of its water¬ proof condition.  Figure 15 shows the distinctly hardened cell walls and cell closures that had contained postdefecating larvae. These specimens were collected and preserved when the site was first discovered. The anterior end was determined by the orientation of the embryos. The larger of the two eggs seemed to be better preserved and therefore was used for illustration ( fig. 23)  The specimens described here are believed to be the second instar based upon the assump¬ tion that this bee has five larval instars, which seems likely based on a survey of head sizes of collected specimens. The second instars were selected from among the available specimens of earlier instars because they best illustrated the distinctive, unusual features of earlier larval instars of H. rhodocerata due to the quality of their preservation and size compared with both smaller and larger specimens. The following description indicates how the anatomy of this instar differs from that of the last larval instar. Remarks: These specimens were compared with a first instar collected at the same locality on October 30, 2010, by J.G. Rozen and J.S. Ascher. Although smaller than the second instar, the first instar otherwise agreed completely with the above description.
The third instar, while retaining the linear appearance and spiculated venter, has the ven¬ trolateral tubercles less pronounced. In the fourth instar spicules persist but paired ventrolateral tubercles have virtually disappeared, although the basal swelling of the venter of abdominal segment 9 is retained, so that in lateral view, that segment is as wide as the preceding one. In the last larval instar described below, there is no hint of paired ventrolateral tubercles.

Figures 26-29, 31-34
Diagnosis: The mature larva of H. rhodocerata agrees closely with described larvae of other species in the genus (Michener, 1953s;Rozen and McGinley, 1974;Rozen, 1987, and  attached dorsally to 9th segment; and venter of abdominal segment 9 much longer than the dorsum of same segment. Furthermore, the cranium of known species is much broader than its height in frontal view. In lateral view the profile of the head from vertex to labral apex forms a continuous even curve ( fig. 29).
Description: Head: Integument of head capsule and labiomaxillary region with surface extremely wrinkled but substantially more deeply so on postdefecating form than on predefecat¬ ing form; pigmentation evident primarily on mandibular apex and teeth, but integument of head capsule tending to be slightly darker than that of body presumably because of thickness; long setae entirely absent but minute setiform sensilla present; extensive spiculation on dorsal surface of maxilla and hypopharynx. Cranium wide in frontal view, with width distinctly greater than distance measured from lower clypeal margin to top of vertex. In lateral view profile curving forward, so that frontoclypeal surface farthest forward; antenna prominence not present. Tento¬ Remarks: Rozen and McGinley (1974) were able to distinguish the sex of at least some of the species of mature larvae of Hesperapis involved with their investigation. A feature for recog¬ nizing females involved identifying imaginal disc responsible for adult female structures seen through the predefecating larva's ventral integument. This sex-recognition method did not reveal any when applied to H. rhodocerata, possibly because the imaginal discs were not revealed against the white background of the predefecating form. However, one cleared and stained postdefecating larva bore cuticular scars on the venter of abdominal segments 7-9, visible because of the stain, which was not employed at the time of the earlier study. This suggests that with examination of more specimens, identification of the sex of mature larvae will become reliable.
In early treatments of melittid larvae, the dorsal ramus of the hypostomal ridge was not understood. It is now recognized as a secondary strengthening structure that extends posteri¬ orly from about the middle of the hypostomal ridge at the point where the hypostomal ridge bends sharply mesad to connect to the posterior tentorial bridge at the posterior tentorial pit.
In Rozen and McGinley (1974: fig. 64) the small triangular part of the parietal encompassed by the hypostomal ridge, the posterior thickening of the capsule, and the dorsal ramus was referred to as "swelling." The ramus is absent from known larvae of Meganomia, Melitta, Macrotera, and seemingly incomplete in Dasypoda (Rozen, 1977(Rozen, ,1978Rozen and Jacobson, 1980).

DISCUSSION
This study confirms the uniformity of mature larval anatomy within the genus Hesperapis, characterized by a broad head capsule that is almost hemispherical in lateral view and an elon¬ gate body ending with a narrow abdominal segment 10 attached dorsally to segment 9. These features are shared with Capicola capicola and less so with Dasypoda plumipes Panzer (Rozen and McGinley, 1974).
The biological information here highlights an interesting question: how can the larva of such a small bee survive while in diapause over a long to very long period when it must remain dor¬ mant until environmental conditions permit production of its food plant? It has been demon¬ strated that many bees (all Megachilidae, many Apidae) have cocoons that regulate cell environments, others (most Colletidae, many if not all Andrenidae) have cells in which the nest¬ making females have applied water-retardant cell linings. However, Hesperapis larvae lack such protections. Here it is hypothesized that the mature larva of H. rhodocerata produces a substance that covers its body with a thin layer of material that reinforces the water-retention quality of its FIGURES 31-37. Microphotographs of mature larva of Hesperapis rhodocerata. 31, 32. Right mandible, dorsal and inner views, respectively. 33, 34. Spiracle, side view and outer views, respectively. 35. Surface covering material from around spiracle. FIGURES 36, 37. Microphotographs of spiracle of mature larva of Hesperapis larreae, side and outer views, respectively. integument. Further, it is pointed out that the mature larva of H. larreae produces a liquid that hardens both the cell wall and cell closure making them waterproof, thereby providing the cell inhabitant protection against desiccation and parasite attack. Questions arise: where are the two substances produced, how are they applied, and might the two substances actually be the same material? The next step in this study will require analysis and comparisons of the materials pro¬ duced by the two larvae and the discovery of their source or sources. Coll. May 7, 1990, as predefecating larva;preserved May 25, 1990, as quiescent postdefecating larva (J.G. Rozen). Seven mature larvae, same locality and collector, May 3, 1994.