The Chilean Bees Xeromelissa nortina and X. sielfeldi: Their Nesting Biologies and Immature Stages, Including Biological Notes on X. rozeni (Colletidae: Xeromelissinae)

ABSTRACT Herein are described the nests and their contents of Xeromelissa nortina (Toro and Moldenke) and of X. sielfeldi (Toro and Moldenke), found in the dry, high Atacama Desert of northern Chile. Nests of the former, discovered in 2014, contained linear cell series in the central pith channels of dead, broken twigs of Baccharis, revealing clear, cellophanelike cell linings that presumably control cell humidity. From the cells, postdefecating larvae were obtained, permitting their description and comparison with our meager understanding of other larval xeromelissines. Nests of X. sielfeldi, also found in broken dead twigs, were discovered and first studied in 1971 before the species was described and named, thereby delaying publication until now. Although similar in most respects to nests of X. nortina, they occupied abandoned beetle burrows. Toro and Moldenke provided information on eggs, predefecating larvae, and pupae, described herein. At the time of that discovery, adults of X. rozeni (Toro and Moldenke), a bee with an exceedingly long proboscis, were also active, permitting observation on their feeding habits, which are included herein.


INTRODUCTION
On a recent field trip to Chile, E.S.W. and Laurence Packer found adults of the bee Xerome¬ lissa nortina (Toro and Moldenke) (body length 3.5-4.5 mm) visiting its presumed food plant, 1 Division of Invertebrate Zoology, American Museum of Natural History.
Copyright © American Museum of Natural History 2015  Acantholippia tarapacana Botta (Verbenaceae), at Aguas Blancas, in El Loa Prov., Region II, elevation 2540 m. At the same site they discovered 12-15 nests of the same bee in broken twigs of Baccharis (Asteraceae) lying on the ground, often in the vicinity of live bushes. From nests they recovered and preserved a number of diapausing larvae, making possible the first descrip¬ tion of the postdefecating larva of this species and only the third modern account of a larva of any taxon of the Xeromelissinae, although Claude-Joseph (1926) early on had illustrated larvae and nests of Chilicola inermis (Friese) and C.friesei (Ducke) (McGinley, 1989).
In 1971 J.G.R., accompanied by Luis Pena, found two species of Xeromelissa visiting flowers of Nolana2 (Solanaceae) at Puquios, Atacama Province, Chile, and found nests of the smaller of the two. He made extensive notes on its nest and preserved an egg, several predefecating larvae, and pupae with adults in the collections of the American Museum of Natural History (AMNH). At the time of discovery, both species were unnamed, and there¬ fore the project was suspended. However, Michener (1995: 333) commented on nesting habits of the smaller species by referring to it as an unknown species of Chilimelissa (now Xeromelissa according to Packer, 2008) collected from Puquios, Chile, by J.G.R. Toro and Moldenke (1979) revised the Chilean Xeromelissinae, describing and naming a number of new species. Now Packer has been able to identify adults of the species that made the nest collected by J.G.R. and Pena as X. sielfeldi (Toro and Moldenke). This makes possible a further account of its nesting biology and the first description of the egg, predefecating larva, and pupa of any Xeromelissinae. The larger species flying at the site was identified by Packer as X. rozeni (Toro and Moldenke).
Although these two field investigations occurred far apart in time, they both treat members of the same genus and together they provide a more complete understanding of the nesting biology and immature stages of this mostly South American subfamily. Specimens described herein are deposited in the AMNH.

BIOLOGY
Nesting Biology of Xeromelissa nortina (Toro and Moldenke) All nests were in dead, dried stems of Baccharis, and most were recovered from stems 6-11 mm in diameter, although the largest was 16 mm in diameter ( fig. 2). It is unknown whether nests were also constructed in broken dead branches on living bushes. All nests occupied the central internal pithy channel, from which females removed the soft pith at the start of nest construction. No nests were recovered from burrows that had been created by beetle larvae, as evidenced by the lack of beetle feces in any of the channels, contrasting with the nest of X.
sielfeldi, described below ( fig. 19, arrows). Nest channels were approximately 3 mm in diameter and were filled with brood cells in linear series of 2-8 cells. Each cell was approximately 6 mm in length, and most in each series were arranged end to end. 1 FIGURE 1. Laurence Packer at nesting area of Xeromelissa nortina at Region II: El Loa Prov.: Aguas Blancas, -23.2670 -67.9830, Chile, elevation: 2540 m. All nests were recovered from dead dry twigs and stems on the ground, such as those on the foreground, left side (arrows). The food source, locally termed rica-rica, can be seen growing behind Packer in the midbackground. FIGURE 2. Nest consisting of a cell series of X. nortina in the central pith channel of a large twig (diameter 16 mm) of Baccharis from that nesting area. Cells were lined by a double layer of transparent, cellophanelike material. The outside layer was extremely thin3 while the inside layer, perhaps 0.1-0.2 mm distant from the outer 3 Rozen in Michener and Rozen (1999) was unable to identified an outer layer of the cocoon of the xeromelissine Geodiscelis megacephala Michener and Rozen but identified and illustrated the filaments that separate the inner layer from the substrate. This may suggest that in colletids the outer cocoon layer serves as a base to maintain the functional integrity of the inner layer. NO. 3838 FIGURES 3, 4. SEM micrographs of the cell lining of Xeromelissa nortina. 3. Circular closure end (a) with piece of lining wall, right (b), and second layer of folded closure (c) behind, all showing smooth, nonporous texture of surfaces. 4. Close-up of lining identified in figure 3 showing overlapping surfaces through which air exchange (arrows) may take place.
layer, was somewhat more substantial. Thin strands of silklike material connected the two layers, as has been noted for Colletes (Torchio, 1965;Rozen and Favreau, 1968). It seems likely that all known cases of double-layered cell linings in the Colletidae will be found to be similar to those described by Torchio (1984) for Hylaeus leptocephalus (Morawitz) (as H. bisinuatus Forster).4 The lining at the rear of the cell appeared as a curved continuation of the wall lining. At the front end (figs. 3, 4), cell linings were folded to cover the entrance after provisioning and egg deposition. Folds were presumably held in place with additional secretions, presumably as described for Colletes c. compactus Cresson (Rozen and Favreau, 1968). Thus, the front end tended to be flat, consisting of a number of transparent layers ( fig.   3). The addition of Malpighian tubule secretions as described by Torchio (1965) was not detected, but obviously should be searched for on future studies. A thick wafer of black and brownish fecal material was present at one end of cells containing mature larvae or dead 4 Torchio (1984) used a different terminology for cell linings from that used here. The outer layer of the cell lining he termed "the lower cell wall" and he called the inner layer "the upper cell wall." His "cell base layer" is the lining at the rear (or posterior end) of the cell. pupae and adults. Cells were never separated by partitions made of soil or other opaque construction material, as is characteristic of many Megachilidae.
The similarity in appearance of the cell lining of this bee to that of the cocoons of many bees is noteworthy considering that they are not homologous structures. Cocoons are created by the last-stage larva whereas cell linings of all colletids are made by the nest¬ ing female in preparation for provisioning and egg deposition (although larval Diphaglossinae also spin cocoons). However, both probably serve the same general functions: moisture control within the cell and exclusion of parasites and predators. Provisions of many colletids are liquids or semiliquids, and cell linings probably prevent loss of liquids into substrates. After feeding, these same linings almost certainly maintain cell humidity.
This may be especially important in the case of small-bodied Xeromelissa during the long hot dry periods in the Atacama Desert where water loss must be a major threat. Cocoons of cocoon-spinning bees are spun only after the larval feeding period and therefore do not provide provision maintenance; they protect only mature larvae or adults from water loss and invasion by parasites/predators.
A number of recent papers have demonstrated that cocoons of megachilids Rozen, 2013;Gotlieb et al., 2014) provide a watertight barrier that regulates water loss. This barrier is the smooth inner surface of the cocoon, which has (usually at its front end) a cluster of small holes, termed the air portal; this portal allows air exchange between the inside of the cocoon and the exterior, necessary for the many months wherein the live animal estivates and then hibernates. Similar air portals on a broad spectrum of apid cocoons are interpreted to indicate a similar function (e.g., Michelette et al., 2000;Rozen and Buchmann, 1990;Rozen et al., 2006). This has also been the case for the fenestrated cocoon "opercula" of Diphaglossinae (Colletidae) (Rozen, 1984) and the "filter areas" on the cocoons of some Rophitinae (Halictidae) (Rozen, 1993). No tests of the porosity of the Xerome¬ lissa or any other colletid cell lining has been performed to date. However, the inner surface of the Xeromelissa cell also displays a smooth surface ( fig. 3), convincingly similar to that of a cocoon, strongly suggesting that it too is a barrier to air exchange between the inside of the cell and the exterior. However, there is no obvious air portal. We here suggest that the air exchange route may be between the folds of the front end of the cell lining ( fig. 4, arrows). NO with Notes on Biology of X. rozeni The following account of the nesting biology of X. sielfeldi with additional biological infor¬ mation on X. rozeni was extracted from a preliminary manuscript drafted immediately follow¬ ing J.G.R.s trip to Chile in 1971 on which he was accompanied and assisted by the late Luis Approximately 15-20 nests of X. sielfeldi were found, six of which contained live larvae or pupae. Some of the others were from previous generations. All were in dead, dry twigs 5-10 mm in diameter, lacking bark, from unknown plants. Most twigs ( fig. 18) were lying on the ground.
The immature stages of this species must be able to withstand high temperatures, for the surface temperature on the ground where sticks were found is normally high during the heat of the day in summer. The conspicuous cell lining described below almost certainly plays an important role with respect to water conservation, in regard to the provisions and to the insects body.
All nests were constructed in abandoned larval burrows of Acmaeodera ( linear series and oriented uniformly with front ends directed toward the nest entrance, though rarely an eclosed adult was found facing the wrong direction.
Females of both species of Xeromelissa transported pollen dry The scopal areas of both are poorly defined, for pollen grains adhered to the bases of fore-and midlegs and to the ventral regions of the metathorax. Most of the pollen, however, was found on the hind coxa, tibia, and femur. The scopa was also well developed on the venter of the metasoma, especially the entire area of the second metasomal sternum and the posterior areas of the third and fourth sterna.
The scopal setae were sparse, not obviously plumose, and moderate in length.
A partly provisioned cell of X. sielfeldi contained a semiliquid, buff-colored mixture of pollen and nectar, identical to that found in a completed cell. No odor was detected from either recently provisioned cells or cells containing mostly grown larvae. Food consistency did not change with time, and in all cases the pollen-nectar mixture was restricted to the rear of the cell. Because a cell must be constructed, provisioned, oviposited in, and enclosed before the next one can be started, the older cell in the series is always the one farthest from the nest entrance. Obviously the oldest immature bee is the farthest from the entrance.
In none of the cell series were cells found all the way to the nest entrance. The length of the tunnel between the last cell constructed and the entrance in one case was 8 mm long; in another case it might have been slightly shorter. The tunnel was filled with strands of silk, sometimes sufficiently dense to be cottonlike, to which adhered frass pellets from Acmaeodera.
Near the entrance some of these strands become attached to the cell wall, so that the last 1.5 With X. rozeni there appears to be a short segment 1 on its maxillary palpus followed by nearly as short segment 2 ( fig. 23). Segments 3 and 4 ( fig. 21), subequal in length, are greatly elongate, together longer than the stipes ( fig. 21). They are separate from one another by an unpigmented membranous basal part of segment 4, which presumably allows the two segments to bend relative to one another. These two segments are heavily sclerotized and pigmented.
They appear sulcate lengthwise along their medial side, which abuts the two segments on the opposing maxillary palpus, which are similarly sulcate. Thus is formed an elongate tube, pre¬ sumably functioning as a channel for nectar. Miklasevskaja and Packer (in press) independently also reach this conclusion. The terminal two segments ( fig. 24) are slender and short, together much less than half the length of segment 4, and appear incapable of connecting to opposite segments; they probably have a sensory function. The base of segment 5 displays a membra¬ nous unpigmented region even longer than this region on segment 4. Although the elongate membranous base of segment 5 obviously allows the maxillary apex flexibility, it may also provide an opening for nectar to be imbibed into the nectar channel formed by the basal seg¬ ments. Perhaps the membranous connection between segments 3 and 4 also assists in some way in passing along the nectar.
The anatomy of the labiomaxillary region of X. sielfeldi ( fig. 25) differs as follows. Although the cardo and stipes are extremely long and segment 1 of the maxillary palpus short, segment 2 is approximately twice as long as 1. Segment 3 is somewhat longer than 1 and 2 combined. Segment 4 has a long, membranous basal part that is as long as its distal sclerotized section, together almost as long as segment 3. Maxillary segments 5 and 6 are subequal in length, each only slightly shorter than segment 3. Segment 3 is the most heavily sclerotized segment and may form the nectar channel. Thus the entrance of the nectar channel might be between the long, membranous bases of segment 4.
To explore how mouthparts might be used during feeding in the case of females of X.
sielfeldi, flowers containing a foraging bee were pulled from the bush and immediately placed in a killing tube so that the bee died in situ (a technique attempted unsuccessfully with X. rozeni because individuals of that species were able to extract themselves before succumbing). Afterward the calyx and part of the corolla were stripped away (figs. 29, 31), the adult bee (male or female) could be observed with its head halfway down the corolla tube and with its maxillary palpi extended so that the palpal tips might contact the flower s ovaries at the bottom of the tube (fig. 29) Diagnosis: Although agreeing in many respects, the postdefecating larva of Xeromelissa nortina can immediately be distinguished from those described as Chilicola ashmeadi (Craw¬ ford) (Eickwort, 1967: fig. 20) andX. sielfeldi (as "Xeromelissinae species A," McGinley, 1981: figs. 80, 81, fide Packer, 2006, personal commun. with J.G.R.) by the lack of paired tubercles above the antennae on the upper part of the head capsule. The tubercles are low and longitu¬ dinally wrinkled in X. sielfeldi ( fig. 33, arrow). Although these tubercles are missing in X.
nortina, the integument is faintly wrinkled in the same homologous position (figs. 12, 13, arrows). Larvae of these three taxa, however, share the following suite of characters: body form in lateral silhouette prolonged, with dorsal and ventral outlines more or less parallel except at extreme ends (i.e., not gradually tapering at either end); paired dorsal body tubercles distinctly transverse; antennal papilla domelike, with length less than basal diameter; labrum short, with pair of small tubercles, well recessed compared to ventrally projecting apex; mandibular apex sharply turning adorally at nearly right angle; labial and maxillary palpi elongate, tapering to points, directed somewhat dorsally; postoccipital ridge rapidly weakening shortly above pos¬ terior tentorial pit to becoming nearly absent along top of parietals; abdominal segment 10 short in lateral view with length much shorter than diameter, with anus terminally positioned, not distinctly closer to dorsal surface than ventral surface. Although some of these features are shared with other colletid taxa, this combination is not known to be shared with others. Description: Total body length: 4.8-6.2 mm (n = 12). Head (figs. 11-13): Sensilla on parietals minute, scarce; spiculation possibly present on hypopharynx and epipharynx. Cuticular pigmentation limited to mandibular apex and apices of labral tubercle and apices of palpi. Coronal ridge absent; postoccipital ridge gradually dimin¬ ishing shortly above posterior tentorial pits and absent from top of head capsule; hypostomal ridge moderately developed, without dorsal ramus; pleurostomal ridge weakly developed; epistomal ridge evident but weakly developed laterad (below) anterior tentorial pit, absent between anterior tentorial pits; tentorium weakly developed. Parietal band evident as narrow depression.
Integument on front of parietal above each antenna with patch of integumental wrinkles. Maxi¬ mum diameter of basal ring of antenna somewhat greater than distance from ring to center of anterior tentorial pit; antennal papilla ( fig. 13) domelike, so that length less than basal diameter, bearing about three sensilla. Lower margin of clypeus projecting in lateral view ( fig. 13), strongly curving upward at midline, so that at midpoint margin at or above level of anterior tentorial pits ( fig. 12). Labrum with paired tubercles small, widely separated, and with apex strongly developed ventrally ( fig. 12); labral sclerite absent.
Mandible, as seen in dorsal or ventral ( fig. 14) view, moderately broad basally, narrowing rapidly and evenly to strongly curved apex; as seen in inner ( fig. 15)  Discussion: An unusual feature of this larva is its enlarged, downward-directed labral apex and the mandibular apices that curve adorally around the lower surface of the labral lobe.
Perhaps future observation of live, feeding larvae will demonstrate how the mandibles interact with the labrum during feeding.
Predefecating Larva of Xeromelissa sielfeldi (Toro and Moldenke) Figures 33, 34 Diagnosis: As indicated by the brevity of the following description, the anatomies of the mature larvae of X. sielfeldi and X. nortina are similar in many ways. This similarity is expressed even in the single character that can be used to differentiate them. While only X. sielfeldi exhib¬ its paired tubercles ( fig. 32, arrow) on its head capsule, the wrinkled surface where the tubercles would be is also found on the head ofX. nortina (figs. 12, 13, arrows).
Although most other head features seem shared (e.g., large palpi, similar mandibles, trans¬ verse salivary opening, small, well-separated labral tubercles), body features are another matter.
Likely, the straight body of X. nortina is merely a postdefecating form of the strongly curled body of the still-feeding larva ofX. sielfeldi. However, the distinct transverse dorsal body tuber¬ cles and pronounced lateral body tubercles of X. nortina seem remarkably different from the relatively even body surface of X. sielfeldi. The eventual discovery of both growth forms of one or the other of the two taxa will resolve this uncertainty. Diagnosis: Bee pupae are rarely described and none of any Xeromelssinae has been so treated. However, Torchio and Burwell (1987) provided information on those available for the Colletidae, and noted that the pupa of Hylaeus leptocephalus (Morawitz) (as H. bisinuatus) alone among the representatives of other colletid subfamilies bore a "terminal spine," possibly a homolog of the median tubercle on tergum 7 of X. sielfeldi. However, the leg 'spines" reported for pupal H. leptocephalus appear to be lacking in X. sielfeldi. Description: Head shape corresponding closely to that of adult, integument without spe¬ cial tubercles, spines, or setae. Mouthparts with cardo contained in proboscidial fossa, stipes, slightly exserted, and remaining distal elements bending sharply posteriad ( fig. 35).
Mesosoma shape also corresponding to that of adult; leg segments without spines or tuber¬ cles accommodating developing adult setae.
Metasomal terga without spines or spicules, but more distal ones each with subapical band of vague protuberances ( fig. 35) and metasomal tergum 8 with pronounced median tubercle at posterior margin, presumably corresponding to "terminal spine" of Hylaeus (Torchio and Burwell, 1987). Metasomal sterna each with posterior margin produced downward as transverse ridge ( fig. 35, arrows). Both authors wish to thank Laurence Packer for his cooperation and numerous contribu¬ tions to this project, and also express their gratitude to Margarita Miklasevskaja for allowing us access to the unpublished manuscript by her and Packer dealing with the mouthpart anat¬ omy of X. rozeni.