NovitatesAMERICAN MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2798, pp. 1-37, figs. 1-37, tables 1, 2 November 13, 1984

Comparative Nesting Biology of the Bee Tribe Exomalopsini (Apoidea, Anthophoridae)

JEROME G. ROZEN, JR.' CONTENTS Abstract ...... 2 Introduction ...... 2 Acknowledgments ...... 2 Specific Observations ...... 3 Paratetrapedia (Trigonopedia) oligotricha (Moure) ...... 3 Paratetrapedia (Paratetrapedia) gigantea (Schrottky) ...... 3 Lanthanomelissa species ...... - 3 Tapinotaspis (Tapinorhina) caerulea (Friese) ...... 3 Tapinotaspis (Tapinotaspoides) tucumana (Vachal) ...... 4 Monoeca lanei (Moure) ...... 7 Monoeca schrottkyi (Friese) ...... 10 Monoeca species ...... 10 Exomalopsis (Exomalopsis) aureopilosa Spinola ...... 10 Exomalopsis (Exomalopsis) globosa (Fabricius) ...... 10 Exomalopsis (Exomalopsis) similis Cresson ...... 10 Exomalopsis (Megomalopsis) fulvofasciata Smith ...... 10 Exomalopsis (Phanomalopsis) solani Cockerell ...... 10 Exomalopsis (Phanomalopsis) solidaginis Cockerell ...... 12 Exomalopsis (Anthophorula) compactula Cockerell ...... 16 Exomalopsis (Anthophorula) crenulata Timberlake ...... 17 Exomalopsis (Anthophorula) torticornis Cockerell ...... 17 Exomalopsis (Anthophorisca) chionura Cockerell ...... 17 Exomalopsis (Anthophorisca) consobrina Timberlake ...... 17 Exomalopsis (Anthophorisca) sidae Cockerell ...... 17 Eremapis parvula Ogloblin ...... 21 Ancyloscelis apiformis (Fabricius) ...... 21 Ancyloscelis panamensis Michener ...... 24

1 Deputy Director for Research and Curator of Hymenoptera, American Museum of Natural History.

Copyright © American Museum of Natural History 1984 ISSN 0003-0082 / Price $3.27 2 AMERICAN MUSEUM NOVITATES NO. 2798

Analysis of the Nesting Biology of the Exomalopsini ...... 24 Profile of the Biology of the Exomalopsini ...... 34 Literature Cited ...... 35

ABSTRACT New data on nesting behavior and ecology of provisioning, development, defecation, cocoon the Exomalopsini are presented. The information spinning, and cuckoo bee associations. A formal is incorporated with previously published ac- synoptic overview ofthe biological features ofthe counts to provide an overview of the nesting bi- tribe also is presented. The evolutionary relation- ology ofthis New World tribe throughout its geo- ships of the Exomalopsini with other anthophor- graphic range. The account treats 23 species in ine taxa are briefly discussed, as are relationships seven genera. The specific subjects are: choice of ofthe tribe with the cleptoparasitic subfamily No- nesting site, social organization, nest structure, madinae.

INTRODUCTION The present paper records new information Exomalopsines are restricted to the New on nesting behavior and ecology of bees of World, as are their close relatives, as defined the anthophorine tribe Exomalopsini. Pre- by Michener and Moure (1957), except for viously published data are synthesized with the Old World tribe Ancylini, about which this information to present a comprehensive no biological information has been pub- understanding ofthe biology of this group of lished. Classification for the Exomalopsini bees, and to shed light on the evolutionary used here follows Michener and Moure (1957) relationships of the taxa within the tribe and and for the genus Exomalopsis, Timberlake also ofthe tribe with other anthophorid taxa. (1980). Although the last objective can scarcely be realized because so few data are available, the great diversity within the Exomalopsini ACKNOWLEDGMENTS promises evidence for evolutionary interpre- I have been assisted in many ways in gath- tation. ering data and preparing this manuscript. The relationships of the Exomalopsini to Finding and excavating nests is no easy task, the other taxa in the family are particularly and the following persons have worked with interesting. Analysis of adult anatomy sug- me to this end: Dr. Fred D. Bennett, Mrs. gests that the tribe is the sister group of the Marjorie Favreau, Ms. Ghisela Kreuger, other nonparasitic anthophorids (Michener, Padre Jesus S. Moure, Mrs. Barbara L. Ro- 1944; Michener and Moure, 1957). Because zen, Mr. Kenneth C. Rozen, and Dr. F. C. exomalopsines still retain some primitive Thompson. features, they may be similar to the ancestor Mr. John L. Neff allowed me to see his of the nonparasitic anthophorids. The evo- manuscript on the biology of Eremapis par- lutionary relationships of the parasitic an- vula, and Dr. Frank D. Parker showed me thophorids, especially the large subfamily his manuscript on Exomalopsis crenulata. Nomadinae, to the nonparasitic ones is not The information contained in those manu- completely understood, but again, adult mor- scripts has added to the completeness of the phology indicates that the nomadines and ex- analysis section. omalopsines may have shared a common My fieldwork in Brazil and Trinidad, and ancestor. The possibility that the nomadines some of my studies in southwestern United arose from a pollen-collecting exomalopsine- States were supported by grant nos. GB5407 like ancestor leads to the question: what is and GB32 193 from the National Science there about exomalopsine behavior that might Foundation. have led to the evolution ofcleptoparasitism? All the original observations on the genus 1984 ROZEN: EXOMALOPSINI 3

S

iR C ,kX,i M wt( A I i. I- '} t ...

FIG. 1. Tapinotaspis tucumana, single nest, near equipment on the ground in the foreground, Vila Velha, Paranfa, Brazil. FIG. 2. Monoeca lanei, nests located in path in front of figure, near Rincao Tijucas do Sul, Paranfa, Brazil.

Exomalopsis presented here were carried out SPECIFIC OBSERVATIONS at the Southwestern Research Station of the American Museum of Natural History over Paratetrapedia (Trigonopedia) oligotricha a period of 15 years. The long-term existence (Moure) of the Station therefore has contributed sub- Michener and Lange (1958) provided lim- stantially to the information in this paper and ited data on this species that nested in a ver- for that matter to many other papers by me tical bank (illustrated in Michener, 1 974b, p. and other persons interested in the Apoidea. 40). I extend my thanks to Mr. Vincent D. Roth, Resident Director, for his hospitality and co- Paratetrapedia (Paratetrapedia) gigantea operation on many occasions. (Schrottky) I also extend my appreciation to the New York Zoological Society for the opportunity Oliveira (1962) in her description of the to stay at the William Beebe Memorial Lab- larvae of the species indicated that they had oratory on Trinidad in 1966 and 1968 where been recovered from the trunk of a rotting I investigated a nesting site of Ancyloscelis fallen tree. apiformis. I thank my secretary, Ms. Deborah Hick- Lanthanomelissa species man, for the careful preparation of this According to Oliveira (1966) this species manuscript. nested horizontally in a vertical bank. The following bee specialists have kindly reviewed this paper and offered valuable Tapinotaspis (Tapinorhina) caerulea comments and suggestions: Mr. Robert W. Brooks, University ofKansas; Dr. George C. (Friese) Eickwort, Cornell University; Dr. Charles D. Claude-Joseph (1926) gave considerable Michener, University ofKansas; Mr. John L. information about the nesting biology of the Neff, Central Texas Melittological Institute; species under the name Exomalopsis caeru- and Dr. Philip F. Torchio, United States De- lea. More recently Vogel (1974) presented ad- partment ofAgriculture. Mr. Roy R. Snelling, ditional data and determined that the stored Natural History Museum, Los Angeles Coun- provisions collected from flowers contained ty, reviewed table 1. oil rather than nectar. 4 AMERICAN MUSEUM NOVITATES NO. 2798

Tapinotaspis (Tapinotaspoides) tucumana (figs. 3-5). Each was circular in cross section, (Vachal)2 without a special lining, and open, except for one that had loose, coarse fill partway down DESCRIPTION OF NESTING AREA: Dr. F. (fig. 4). The tunnels tended to be about 5 mm Christian Thompson, Padre Jesus S. Moure, in diameter, although one widened to about and I discovered and excavated five nests of 7 mm at the cell level. Except for laterals this moderately large, nearly all black species leading to cells, tunnels did not branch. in the vicinity of Vila Velha, a park in the Laterals, all of which were filled after cell state of Parana, Brazil, between January 31 closure, ranged in length from 2 to 6 cm mea- and February 5, 1974. One nest (fig. 1) was sured in a straight line from main tunnel to in the ground near the visitors center of the cell entrance. Approximately the same di- park, whereas all others were grouped several ameter as the main tunnel, they varied in kilometers away; three of these were within direction, some curving gradually downward 5 m of each other and the fourth was 100 m from the main tunnel, others leading hori- distant from the cluster of nests. Each nest zontally, and still others rising to join the was hidden in dense vegetation consisting cells. mostly of grasses less than 1 m high; none All cells were arranged singly (i.e., not in was near trees or tall shrubs. Most entrances a linear series), had maximum diameters of were not visible unless the surrounding plants 6.0 to 7.5 mm and maximum lengths of 10.0 were parted. These observations suggest that to 10.5 mm (five measurements each), and the species does not nest gregariously and were approximately 4.0 mm in diameter at prefers sites that are covered with abundant the opening. They appeared to be slightly flat- low vegetation. ter on the bottom than on the top, and they In all cases the ground was horizontal to sloped to the rear 20 to 35 degrees from the slightly sloping, moist, without stones or large horizontal (figs. 6-8). No masoned (i.e., pebbles, and easily excavated. Although nu- "built-in") wall ofsoil could be detected. The merous small roots penetrated the upper 15 entire cell surface was smooth but uneven to 20 cm of soil, no large roots were encoun- because ofthe coarse texture ofthe surround- tered. ing soil, and lined. In newly constructed cells, NESTING ACTIVITY: Contrasting with many the lining was conspicuous and shiny and other Exomalopsini, each nest contained a could be peeled with forceps. After construc- single female. Three entrances (figs. 3-5) were tion this lining quickly became dull, perhaps on the ground surface and surrounded by tu- as a result of a patina of fine black mold that muli of loose soil approximately 6 cm in di- grew sparsely over the surface. The lining was ameter and 2 to 3 cm high. In one (fig. 5) the waterproofwhen tested with a water droplet. loose material, when blown away, left a hard The lining darkened but did not melt at 700°F. core 1.7 cm in diameter surrounding the tun- Cells were arranged randomly around the nel. Although this feature suggested a turret, main tunnel. In general, older larvae were it was not noted around other entrances and encountered nearer the surface, whereas presumably was not a true turret such as that younger individuals and eggs were found constructed by Diadasia but was caused by deeper in the ground, facts suggesting that the drying of moist soil. Two other entrances order ofcell and lateral construction was from penetrated the fill of larger burrows (approx- top to bottom. Although mature larvae and imately 2 cm in diameter) which may have pupae were found lower than younger forms been scarab emergence burrows. One bee in one nest, we may actually have encoun- tunnel entered the side of the burrow about tered cells from another nest. 3 cm below the surface; the other (fig. 4) pen- The inside surface of the cell closure was etrated vertically through the fill of the large a concave spiral of about three to four coils burrow. that were somewhat indistinct because ofthe All main tunnels descended more or less coarse soil texture. Specimens of cells ex- vertically for their entire lengths, 23 to 33 cm amined in the American Museum ofNatural History revealed a closure that had no dis- 2 Identified by Padre Jesus S. Moure. tinct outer surface, as did, for example, the 1984 ROZEN: EXOMALOPSINI 5

BURROW

4.0 cm

5

3 .. . X ... .: :: :...... : ..... FECES 0^. 0-w 6 - POLLEN 8

7 FOOD LOAF

FIGS. 3-8. Tapinotaspis tucumana. 3-5. Nests, side view. 6. Cell, side view, in the process of being provisioned. 7. Cell and food loaf, side view, showing young feeding larva. 8. Rear of cell, side view, showing position of feces. Scale refers to figures 3-5.

smooth, lined, concave outer surface of clo- also occurred in the fill of the adjoining lat- sures of Exomalopsis. Instead, the closure eral. The fibers may have been mold hyphae material seemed to grade into the fill of the because one vacated cell (but not others) had lateral. All six closures preserved in the col- similar strands extending from the cell wall lection had short strands of fine brown fi- into the cell lumen and even into the feces. brous material running between the soil par- However, strands did not occur in the soil ticles in the plug. These twisted, flattened surrounding either the cell or the lateral in strands, somewhat like silk in appearance, any of the samples stored in the Museum. It 6 AMERICAN MUSEUM NOVITATES NO. 2798 seems likely that the fibers were either hyphae rounded anteriorly than posteriorly. Young or special material that the female bee de- larvae fed as they slowly crawled over the posited in the closure and lateral fill. The soft, moist food surface (fig. 7), and, as they closure was waterproof on the inside when moved, made channels in the mass similar tested with a droplet, possibly because ofthe to those made by larvae of Diadasia. Man- hydrofuge properties of the fibrous material. dibles were used to bite into the mass, as is The fill in the lateral was also water repellent, also the case with Diadasia olivacea (Cres- in sharp contrast to the absorbent quality of son) (Rozen, Eickwort, and Eickwort, 1978) the surrounding soil. The nature, source, and and Monoeca lanei. Both large and inter- possible adaptive significance ofthe material mediate-sized larvae pressed their dorsa deserves further investigation. against the cell wall as they crawled over the PROVISIONING: Although females in flight food, but how very young larvae moved is appeared to be transporting dry pollen to the unknown. By the time larvae became inter- cells, subsequent examination of pollen-lad- mediate in size, the pollen masses no longer en females in the collection suggested that the adhered to the cell floors; the larva, curved pollen on the large, dense scopae was prob- around the mass, supported it entirely and ably somewhat moist. The source of the pol- prevented it from touching the cell wall at len was not ascertained although pollen-bear- any point, as has been described for Prote- ing females were seen on a number of plant peolus singularis (Linsley and Michener) species. Early loads were placed evenly in a (ibid.). Shortly after the larvae finished eating, semimoist condition on the floor of the cell they defecated. Feces were pressed to the up- (fig. 6). After the total quantity had been per rear of the cell (fig. 8) as elongate yellow- brought in, the female shaped the food into ish ribbons running parallel to the saggital an elongate loaf 7 mm long (three measure- plane ofthe cell. None ofthe excavated larvae ments) and 5 to 6 mm in diameter (two mea- became inactive after defecating but soon pu- surements) and attached it to the rear floor pated. They spun no cocoons even though so that most ofthe loaf surface did not touch the labiomaxillary region, including palpi and the cell wall (fig. 7). A number ofloaves, bear- the salivary lips, was fully developed as in ing live eggs, had a highly convoluted, brain- cocoon-spinning larvae. like surface that possessed a stubble of whit- ADULT AcTIvITY: In contrast to that ofmost ish mold hyphae. In contrast, a number of bees, the flight ofthis species (both sexes) was food masses bearing young and intermediate slow. Furthermore, individuals flew low over larvae had a moist, yellow, fine, more even the vegetation and often encountered the tall- surface that lacked hyphae. Because it seems er stems and blades of grass, reducing their unlikely that early instars could consume speed even more. The nests of this species enough surface material to remove all the would not have been discovered hidden in hyphae and convolutions, the convoluted the dense vegetation had we not been able to surface may have been an abnormality, per- follow the slow-moving, pollen-laden fe- haps an outbreak of a mold induced by cer- males with their black color and bright yellow tain weather conditions. Because all nests but pollen. We were further assisted in finding one contained such provisions, the matter nests because returning females tended to fly warrants further study. directly to them rather than altering their DEVELOPMENT: One egg was found in situ course from one flower to another as do for- in the midsaggital plane near the middle of aging females. the top of the pollen mass, its long axis par- We saw no matings but collected numerous alleling that of the cell. Because oviposition males flying through tops of the low vegeta- behavior shows little variability in solitary tion in the same fashion as females. Both bees, this placement will probably be char- sexes were active during most of the day and acteristic of the species. Eggs were distinctly a few females bearing pollen were still ob- elongate, 3.0 to 3.15 mm long and 0.6 mm served at 5:30 P.M. on one of the days. in diameter (two measurements each). Little can be stated concerning the seasonal Smooth, whitish, and somewhat shiny, they cycle except that this species obviously has were slightly curved and perhaps slightly more more than one generation a year as evi- 1984 ROZEN: EXOMALOPSINI 7 denced by larvae, after feeding, pupating in gles. At least one was highly polished on the the laboratory. inside as was the main tunnel below. Tunnels PARASITISM: Immatures ofno parasitic bee turned downward immediately below the en- were recovered from the nest, but an appar- trances and descended almost vertically. Cir- ently new species of cuckoo bee belonging to cular in cross section and 5.5 mm in diam- the nomadine genus Parepeolus (Michener, eter, they were unusual in that the wall was in litt.) was commonly seen flying in associ- shiny and coated with a waterproof material ation with Tapinotaspis tucumana at a num- (tested with a droplet of water). Between the ber oflocalities in Parana. This mostly black depths of 19 to 27 cm the vertical tunnel of parasitic form was difficult to distinguish in each of -the five nests excavated angled flight from T. tucumana. Parepeolus is be- abruptly and then gradually curved down- lieved to be a parasite of this species both ward until it reached an open cell. The lower because of coincidence of occurrence of the curved section, here termed a secondary, had two and because a single adult female of the the same diameter as the main tunnel. In parasitic bee was recovered in one of the ex- small nests a grouping ofcells seemed to sub- cavated nests of T. tucumana. tend a single secondary tunnel (fig. 15), as if all the cells had been constructed from the Monoeca lanei (Moure)3 one secondary. However, in larger nests (fig. 10) the apparent random arrangement ofcells DESCRIPTION OF NESTING AREA: This suggested that each secondary branch led to species nested along a narrow dirt path (fig. a separate cell. If this is the case, the second- 2) approximately 20 to 25 cm wide near Rin- aries were actually "laterals" that were ob- cao Tijucas do Sul, Parana, Brazil. Dr. F. viously filled with soil after cell closure. Sec- Christian Thompson and I made the follow- ondaries were shiny on the inside like the ing observations on February 10, 1974. The main tunnel and presumably were also water- path led across a level soccer field that was proofed. Secondaries narrowed only slightly overgrown with tall grasses and other scat- at the cell opening. tered plants. The path was 4 to 5 cm below Cells (fig. 11), arranged singly, were ori- the field surface in places, and during the ented vertically. Maximum cell length was course ofour observations was intermittently 15 to 17 mm (three measurements) and max- trodden by people and farm animals. Trees imum diameter, 7.0 to 7.5 mm (five mea- and shrubs were sufficiently distant so as not surements). To keep cells more or less intact to shade the site. Dry on the surface, the soil during excavation was difficult, but one cell was moist immediately below and consisted was obviously flatter on one side than the of a fine claylike material, scattered coarser side opposite, and other cells suggested a sim- sandlike particles, and some organic debris ilar asymmetry. This asymmetry as well as but no stones, rocks, or large roots. The pol- that of the provisions is similar to certain len plant was not discovered. features in Tapinotaspis tucumana. Cells NESTING ACTIVITY: Six nests, each with a ranged in depth from 24 to 37 cm. single female, were discovered along a stretch The cell wall showed no indication ofbeing of6 m, and several other nests may have been plastered with specially worked soil, and the overlooked. All were at the edge of the path cell lining was extremely thick, cracked easily where the surface rose to meet the playing so that it could not be peeled from the wall field. No cells from nests of previous gener- in spite of its thickness, and turned opaque ations were encountered. Several nest en- whitish on drying. A piece of lining melted trances were surrounded by distinct turrets at slightly below the boiling point of water, (fig. 14) similar to those of some Diadasia. contrasting sharply with the mere darkening Turrets on other nests may have been de- of the cell lining of Tapinotaspis tucumana stroyed by people or animals walking on them. at 700°F. The lining dissolved slowly in ether The turrets were approximately 1.5 cm high but apparently not in xylene. The cell closure and curved in some cases at nearly right an- was a moderately concave spiral on the inside where it readily absorbed a water droplet when 3 Identified by Padre Jesus S. Moure. tested. The closure graded into the fill of the 8 AMERICAN MUSEUM NOVITATES NO. 2798

SECONDARY

4.0cm

: fo8 L; ______k______

TO ENTRANCE 80~ ~~~~~'5Q7 I0 9~~~~-- 90 "F

J z.-'. L :' :'''"'.:- ': -'': .: ': ':t 100 ., 6SC-': 12 13

FIGS. 9-13. Monoeca lanei. 9. Nest, side view. 10. Nest, same, top view. 11. Cell with egg and provisions, side view. 12. Bottom of cell with partly consumed provisions, side view. 13. Same, with provisions even more depleted. Scale refers to figures 9, 10.

secondary; there was no distinct, smooth out- that of Tapinotaspis tucumana in spite ofthe er surface to the closure. difference in orientation of the cells. After PROVISIONING: Females transported pro- transporting the final load, the female shaped visions as a conspicuous, moist mass on their the mixture into a tall, moderately dense loaf large scopae. During foraging, the female (fig. 1 1) in the center of the lower end of the stored unworked loads of provisions in the cell. This loaf, without a foot, was "cement- bottom of the cell, a condition that parallels ed" almost vertically by a somewhat more 1 984 ROZEN: EXOMALOPSINI 9

4.0 cm.

14

TOENTRANCE 1 4t15

FIGS. 14-15. Nest of Moneoca lanei, side and top views, respectively.

moist food mixture at the bottom of the cell. The cementing mixture had a smoother sur- The surface of this cementing material tilted face than the loaf. slightly upward toward the flat side ofthe cell DEVELOPMENT: The slightly curved egg was in each case. The surface of the loaf was deposited toward the top ofthe provision loaf somewhat uneven, but no mold grew on it. and, at least in one case, was positioned to- 10 AMERICAN MUSEUM NOVITATES NO. 2798 ward the curved side of the cell (fig. 1 1). The Monoeca schrottkyi (Friese) anterior part of the egg, pointing toward the cell closure, was more rounded than the pos- Schrottky (1901) reported that this species, terior. In most cases the ends but not the referred to as Pachycentris schrottkyi, nested middle of the egg were attached to the pro- in a bank, but he gave no further information. visions, but, in one case (fig. 11), only the anterior part was attached. Because they were Monoeca species not glued to the food, eggs were easily dis- Michener and Lange (1958) briefly referred lodged accidentally when cells were opened. to a nest of an unidentified species in this Somewhat elongate and white, with a smooth, genus. shiny, transparent chorion, they measured 2.75 mm (two measurements) in length and Exomalopsis (Exomalopsis) aureopilosa 0.6 mm (one measurement) in maximum di- Spinola ameter. Upon hatching, the first instar re- mained in situ and apparently ingested liq- Zucchi (1973) carefully examined nests of uid. Second instars crawled over the loaf as this species and recorded his findings in his they fed. As the larva grew it consumed the doctoral thesis. From the description and his provisions in such a way that the loafbecame well-prepared drawings, it is clear that he smaller but more acutely pointed at the top misinterpreted Rozen and MacNeill's (1957) (figs. 12, 13). This species did not channel the description of the "foot" on the food loaf. provisions as do Diadasia and Tapinotaspis What he recorded as the foot seems to be a tucumana. At no time did the food mass sep- special structure at the bottom of the food arate from the bottom of the cell (figs. 12, mass which is in a vertical cell. 13). While curled more or less horizontally, last instars fed by ducking their heads to chew Exomalopsis (Exomalopsis) globosa at the provisions on which they rested. Al- (Fabricius) though a few cells contained considerable liq- Considerable explicit ecological and life uid, apparently most food masses did not history information on this species and the normally acquire water but rather remained one following was recorded by Raw (1977). moderately firm while being consumed. In addition to details about nests, he included Our investigations were completed prior to information on predation, mortality, fecun- defecation and possible cocoon spinning. Al- dity, and cooperation among adult females though the mouthparts of the larvae pos- in nest building. sessed such cocoon-spinning modifications as projecting salivary lips, these features may Exomalopsis (Exomalopsis) similis not always signify cocoon spinning on the Cresson part of the larva; see, for example, the dis- cussion of Tapinotaspis tucumana. The biology of the species was treated and ADULT ACTIVITY: Females were observed compared with the one above by Raw (1977). only over the nesting area as they returned with provisions. They approached moder- Exomalopsis (Megomalopsis) fulvofasciata ately slowly, but after landing at the nest en- Smith trance, they entered without hesitation. In the Zucchi (1973) reported that two nests of larger nests, eggs as well as fully grown, pre- this species were found on a narrow path with defecating larvae were encountered, suggest- a number of nests of E. aureopilosa. ing that nesting had been taking place over a considerable period of time. However, since Exomalopsis (Phanomalopsis) solani no pupae or postdefecating forms were found, Cockerell this suggests that these bees do not nest and provision throughout the year. Although both Michener (1966) and Hurd PARASITISM: No cuckoo bees were associ- and Linsley (1975) have noted that a number ated with this site. Meloid larvae were found offemales occupy the same nest, nothing else in two cells. has been recorded about the nesting biology 1984 ROZEN: EXOMALOPSINI I1I of this common species from the southwest- spiral with four to five coils to the radius on ern part of the United States, except that the inside and smooth and deeply concave Paranomada velutina Linsley is a cleptopar- on the outside. Two of the closures were 2 asite in its nest (Rozen, 1977b). The following mm thick at the periphery and about 1 mm information is derived primarily from cells thick in the middle. The outside, like the cell preserved in the American Museum of Nat- wall, was waterproofwhen tested with a drop- ural History, and from photographs of the let of water, but the inside surface readily cells and provisions taken in 1967. absorbed water. Laterals were filled with soil. DESCRIPTION OF NESTING AREA: Two nests PROVISIONING: Several photographs of the ofthe species were excavated at 1 mile north food loafftaken in 1967 show that it was sim- ofRodeo, Hidalgo County, New Mexico. One ilar in general shape to those of E. chionura, nest, excavated on August 28, 30, and 31, solidaginis, and consobrina, with a flattened 1967, by Ghisela Kreuger and myself, yielded top surface4 on which the egg was positioned, the Paranomada velutina larva described by and a somewhat truncated front end as viewed Rozen (1977b), and the other nest, excavated from above. As viewed from the side, the on August 15, 1974, by Kenneth C. Rozen, front end sloped downward and backward as was occupied by at least 12 females. The if drawn into a ventral foot, which may or ground surface was probably horizontal or may not have been present. The surface of nearly so in both cases. Cells from the 1967 the loaf was smooth and moist, and the nest were in soil of coarse granular particles rounded rear ofthe mass fitted evenly against intermixed with a fine reddish claylike ma- the rear of the cell. Dimensions (single mea- terial. The soil from the 1974 nest was fine surements) were loaf length, 5.0 mm; height, and even-grained. 3.3 mm; and maximum width, 3.2 mm. NESTING ACTIVITY: Cells collected in 1967 DEVELOPMENT: Two eggs were in the mid- ranged in length from 8.0 to 9.1 mm (six saggital plane on top of the food masses. At measurements) and 5.2 to 5.6 mm in maxi- least one had its anterior end closest to the mum diameter (eight measurements). Two cell closure, and slightly less than 1.0 mm cells from the 1974 nest were 5.2 and 5.5 mm from the front end of the loaf. Its position, in maximum diameter. No apparent differ- therefore, was essentially as shown for the ence could be detected between the cell wall early instar larvae of E. solidaginis (fig. 20). in cross section and the surrounding soil in It was white and curved. No cocoons were the 1967 nest, so that the cell wall apparently collected from either nest, and the salivary had not been plastered. Similarly, no ma- lips on mature larvae were only slightly pro- soned cell wall could be detected in the 1974 duced. Consequently, no cocoons are appar- nest, but samples in the collection suggest ently spun by this species. Several cells in the that the cell wall may have been somewhat collection had linear sequences of nearly harder than the surrounding soil. In cells from spherical fecal pellets over most, if not all, of both nests, the lining was partly shiny, water- the cell wall. Although the quantity of feces proofwhen tested with a droplet, and smooth, in these cases suggested that larvae had only except where large soil particles protruded. begun to void, the pattern suggests that feces No visible change could be detected in the are applied to the entire cell wall. No infor- color or nature of the lining when heated on mation is available on whether or not feces a hot plate to 700°F. Cell orientation was not are applied to the inner surface of the cell recorded, but cells were slightly flatter on one closure. side than the other, and the angle ofthe plane PARASITISM: Rozen (1977b) described the ofthe closure did not form a right angle with mature larva and pupa of Paranomada ve- the "longitudinal axis" (as defined under E. lutina collected from the 1967 nest. The rate solidaginis) of the cell, as was also the case ofparasitism by this cuckoo bee was high; at with E. compactula (fig. 16) and solidaginis least 50 percent of the cells contained Para- (figs. 17-24). Consequently, the flatter side of nomada eggs or larvae. the cell was shorter than the opposite side. The cell closure on two cells was 4.0 mm 4Description of food loaf and egg here assumes that in diameter and in all cells was a concave the cell is essentially horizontal. 12 AMERICAN MUSEUM NOVITATES NO. 2798

FECES c f SPIRAL SURFACE_ _-- COCOON - ~FL

CLOSURE OUTER FACE eCLOSUR0 CLOSURE CLOSURE EAR ~~~~~~~~~~OF 16 18 FLOO 17 -LARV

FOOT 21 LIQUID ..2 F~~~~~~~~~OODLOAF 3mm 19 POLLENOLLEN 20 ~ ~ ~

iT~~~~~~~~~--//-,/ ,//I- Ii, ~~~~~~~~FECES

22 ~~~FOOD LOAF23 -4

FIG. 16. Exomalopsis compactula cell containing feces and cocoon, side view. FIGS. 17-24. Exomalopsis solidaginis. 17. Cell, side view, showing method of taking measurements. For further explanation, see text. 18. Same, showing terminology used in text. 19. Same, being provi- sioned. 20. Same, showing fully formed provisions with early instar. 21. Same, front view, showing position of provisions and early instar. 22. Same, side view, with third instar. 23. Same, showing fourth instar circling grooved food loaf. 24. Same, early fifth instar starting to defecate while feeding on grooved food loaf. Scale refers to all figures. roadway 20 miles south of Animas, Hidalgo Exomalopsis (Phanomalopsis) solidaginis County, New Mexico, on September 8, 1977, Cockerell5 and excavated it on September 14, 1977. DESCRIPTION OF NESTING AREA: I discov- Twelve females were collected either enter- ered a single nest of this species in the center ing, exiting, or in close association with the of a nearly horizontal, infrequently used, dirt nest; and 14 others, possibly representing re-

sIdentified by J. G. Rozen, Jr., but the identification of Natural History, lead me to suspect their distinctive- is not certain. Extensive series of adult E. solani and a ness and therefore their validity as separate species. Fe- shorter series of E. solidaginis, both identified by P. H. males taken from the nest display more of the charac- Timberlake, in the collection of the American Museum teristics ofsolidaginis as identified by Timberlake (1980) 1 984 ROZEN: EXOMALOPSINI 13 turning foraging females, were collected while The Apache nest was essentially similar ex- flying as iflost over the roadway after I start- cept that its main tunnel descended only to ed excavations. The entrance as well as the about 25 cm in depth. At this point it surrounding ground was unshaded by the branched numerous times with subsequent ground cover of low herbaceous vegetation tunnels leading nearly horizontally. Appar- and was not in danger ofbeing flooded in the ently the hard, pebble- and stone-filled lower event of heavy rains. The compact, claylike soil prevented the females from descending soil was hard and dry on the surface but moist farther, with the result that numerous freshly and more easily excavated below 10 cm. It provisioned and vacated cells were found be- contained few stones or roots. tween the depths of 25 and 35 cm. Two more nests were found within 1 m of Cells at the Animas site (figs. 17-24), ar- each other at 13 miles southwest of Apache, ranged singly, were oriented somewhat less Cochise County, Arizona, on August 30, 1983 than 45 degrees from horizontal, usually near (fig. 25). The larger nest was partly excavated main tunnels, and had the closure end higher the next day, and 35 females and 1 male were than the rear. They ranged from 7.5 to 8.5 recovered from it. The smaller one contained mm long and 5.0 to 5.5 mm in maximum 17 females and 1 male. Like the Animas site, diameter (eight measurements). The inner the area was nearly level, unshaded, and sup- surface was extremely fine grained, smooth, ported low sparse vegetation such as Lepid- and moderately shiny throughout, and the ium, Mentzelia, and Euphorbia. The soil was earthen cell wall was thick and usually some- dry on the surface and had few roots and what harder than the surrounding soil. As a pebbles, but at about 25 cm depth it con- consequence, cells often could be identified tained many rocks and pebbles. and extracted from the ground without being NESTING ACTIVITY: The nest entrances, opened. The hardened wall was thickest (ap- without tumuli or turrets, were circular in proximately 1.0 mm) at the rear of the cell, cross section, and slightly beveled. The An- whereas elsewhere, especially the floor, it was imas nest was excavated more completely often noticeably thinner (approximately 0.3 than the Apache one. The main tunnel ofthe to 0.5 mm). Except for hardness, there was Animas nest, 4.5 mm in diameter, descended no indication that the wall had been fabri- vertically with only slight meandering to a cated by the female. Although the hardness depth of 42 cm. The tunnel possessed no may have been caused by the female mixing chambers or branches in the upper reaches, soil with some secretion, it could also have but between depths of 42 and 58 cm it resulted from a secretion applied to and ab- branched a number of times at irregular in- sorbed by the inner surface of the cell. tervals, with resulting tunnels radiating out- Cells at the Apache site agreed with those ward in all directions in a meandering fash- from Animas except that the walls were not ion. At this general level 130 cells, containing noticeably harder than the surrounding soil. all stages oflarvae and eggs as well as meloids, However, there was a strong suggestion that were excavated as were cells from previous not only the cell walls but also the burrow generations. The cells extended over a hori- wall might have been masoned. For example, zontal area ofabout 58 by 40 cm and the nest parts of dead Exomalopsis individuals were was probably even more extensive. The tun- buried in the walls. Cell orientation at the nels seemed unlined and most were open, Apache site may have been more variable except that those tunnels (laterals) leading to (perhaps because ofthe pebbly substrate), with cells were plugged with soil. Some tunnels cells seeming to range from 35 degrees to 80 other than laterals may also have been filled, degrees from horizontal. as a number of cells seemed considerably re- Because of the numerous cells recovered moved from open tunnels. from the Animas nest and the hard consis- tency of their walls, a detailed investigation of cell symmetry was possible. Several cells than those of solani, and therefore I provisionally use were opened and trimmed so that their shape the former name. Cells of this species were also smaller could be carefully drawn with the camera lu- than those ofE. solani, corresponding to a recorded size cida and analyzed in longitudinal cross sec- difference between the two. tion. As can be seen in figure 17, the major 14 AMERICAN MUSEUM NOVITATES NO. 2798

Aft i6aiihw

i.-

FIG. 25. Exomalopsis solidaginis, nesting area 13 miles southwest of Apache, Arizona. Two nests were found at edge of barren area just behind the ice chest. FIG. 26. Exomalopsis sidae, nesting area 5 miles north of Willcox, Arizona. Nests were scattered from umbrella next to car to net in middle foreground. part ofthe cell was divided into an upper and edge, and three measured 3.2 to 3.4 mm in lower half by drawing line ab from the rear diameter. point ofthe cell so that areas above and below PROVISIONING: Two cells were encountered the line and behind a perpendicular (cd) to as they were being provisioned. The moist it were approximately equal. This line only pollen load was appressed to the rear of the approximated the "longitudinal axis" of the cell and the cell floor. One cell (fig. 19) had cell because of the asymmetry of the upper a drop of liquid (presumably nectar because and lower halves and because the plane of it was miscible with water) extending from the periphery of the cell closure (line ef) was the pollen surface onto the cell floor. Newly not at right angles to this line. From figure formed provisions were shaped as in figure 17 it appears that the rear (lower) part of the 20, and in each case the rather wide, rounded cell was nearly, but not quite, symmetrical, rear end was attached to the rear of the cell the floor being slightly flatter than the ceiling. by a droplet of liquid. The foot, which in Unquestionably, the ceiling was longer than some other species of Exomalopsis (e.g., the floor, because the plane ofthe cell closure chionura, Rozen and MacNeill, 1957, figs. 1- (ef) was not perpendicular to ab. This cell 4; consobrina, Rozen, 1977b, fig. 1) touched and closure configuration was characteristic the cell floor, was smaller than in the other of all cells in the nest. The extremely thin, species, and did not reach the floor. The gen- transparent lining did not melt when heated eral shape and position ofthe food mass (fig. to 700°F. on a hot plate, nor did it change in 20) were constant in the numerous cells ex- color or texture in any other way. amined, although the size varied. The or- The cell closure at both sites was a shallow, ange-yellow, moist loaf, somewhat flattened concave spiral on the inside with four dis- on top, was 5.0 to 6.1 mm in length (nine tinct, and well-defined rows to the radius. measurements), 3.2 to 4.0 mm in width (nine The inner surface of the cell closure was not measurements), and 2.9 to 3.5 mm high (eight waterproof when tested. The outside of the measurements), and had a smooth surface. cell closure was smooth and more concave Cells with provisions gave off a fermented, than the inner side. The outer surface was cheesy odor when first opened. darkish and waterproof, not unlike the inner DEVELOPMENT: Curved, translucent white, surface ofthe cell itself. Animas closures were shiny eggs were deposited on the anterior top approximately 1.5 to 2.0 mm thick in the ofthe food masses in the midline of the cells middle, and 3.0 mm at the edge, and their with only the front and rear ends touching diameter was approximately 4.0 mm. A sin- the surface. gle Apache closure was 1.5 mm thick at the The following comparative anatomical 1984 ROZEN: EXOMALOPSINI 15 study of larvae from the Animas nest re- body segments. These tubercules, though vealed five larval instars, a figure correspond- small, were quite visible on most specimens ing to Oliveira's (1966) documentation in the in which the integument was somewhat dis- case of Lanthanomelissa. Nonetheless, Ro- tended. zen, Eickwort, and Eickwort (1978) observed The mandibles of the fourth instar were only four instars in Diadasia olivacea, so that even more darkly pigmented than those of the number of larval stadia deserves further the third, with the ventral subapical tooth careful attention. The anatomical study was now more apical in position. The maxillary based on a total of 94 larvae preserved from palpi were longer than their basal diameter. the nest, including six postdefecating, quies- This instar was able not only to crawl while cent, mature larvae; 56 actively feeding and/ feeding over the now oblong, ridged food or defecating fifth instars; 17 fourth instars; mass, but also to detach the mass from the nine third instars; four second instars; and rear of the cell (fig. 23) and support it in the two first instars. Because the structures ofthe center of the cell with its body. The food second instar were visible through the nearly surface, by this time, had become much more transparent cuticle of the first instar, the first moist, so that distinct feeding channels re- instar was difficult to distinguish from the mained in the soft surface as the larva crawled. next instar. Nonetheless, the cuticle of the Fourth instars were found in various posi- head and anterior part ofthe body ofthe first tions in the cell. None was seen to defecate. stage larva displayed distinct mouthparts (in- The fifth and last larval instar (fig. 24) con- cluding a short unpigmented but apically sumed most of the provisions. The mandi- pointed mandible), as well as spiracles and bles were thinner subapically in relation to tracheae. As indicated above, the first instar their overall size than those of other instars, apparently remains in the same position as and maxillary palpi were longer than their the egg, and there was no pollen in the dis- basal diameter. The larvae continued to move sected intestinal tract ofeither specimen. Per- in relation both to the wall and to the food haps first instars consume only liquid from mass. The loaf diminished in size with the the food mass, as has been suggested for Dia- larval feeding while the larva cradled it so dasia olivacea (ibid.). that it did not touch the cell wall or the feces. Second instars bore mandibles, each of Larvae fed rapidly; one took 36 bites during which was slightly pigmented at the tip and a 10-second period, or about four bites per possessed a small, pointed subapical ventral second. During the beginning ofthis stadium tooth. Maxillary palpi were distinct, but small larvae started defecating, applying yellow and nipple-like, and not as high as their basal feces as rows of small ovoid smears on the diameter. Because the abdominal venter IX cell wall and closure. After provisions were was slightly protuberant, this instar may have consumed, the feces appeared as somewhat been able to crawl, but I made no direct ob- flattened pellets appressed more or less evenly servations of this. The intestinal tract of one over the entire surface of the cell, although about to molt was packed with abundant pol- the cell closure may not have been so densely len, proving that this instar ingests more than covered as the wall. No cocoon was spun, as liquid. Increase in body girth and length also revealed both by six postdefecating larvae attested to its eating. lacking them and by the recessed labiomax- The mandibles of the third instars were illary region of the larval heads. The dark distinctly pigmented at the entire apical end, color of the feces in cells of postdefecating including the ventral tooth. The maxillary larvae suggested they may have been over- palpi projected and were about as long as wintering from the previous year. No pupae their basal diameter. One larva was found on occurred in the nest at the time ofexcavation. the top front ofthe pollen mass, as illustrated The Animas site produced a wide range of in figure 22. This instar is believed to be am- instars on September 14, as discussed above, bulatory in that the abdominal venter IX pro- whereas the Apache site yielded only eggs and jected strongly and small median ventral tu- very early instars on August 31. Although the bercules (probably eversible in nature) sample size was small at Apache, the differ- occurred on the midline between most ofthe ences between the two may have been real, 16 AMERICAN MUSEUM NOVITATES NO. 2798 a result of the different dates of collection. trance was open, somewhat irregular in shape, ADULT ACTIVITY: Females were seen en- not near an obvious marker such as a stone tering and leaving the nest from 10:00 A.M. or twig, and without a tumulus. One nest was to 3:30 P.M. Although some females were occupied by at least two females, and both found in the Animas nest, they were few in old and current cells were present in the nests. number (four or five) and no adult males were The main burrow, circular in cross section, observed either in, or in the vicinity of, the was 4.0 mm in diameter. Its wall seemed to nest. However, the partly excavated nest at be lined with finer material than the sub- the Apache site yielded nine females, and a strate, and was not waterproof when tested male emerged from each of the two nests with a droplet of water. Each main burrow there. descended vertically in an irregular fashion PARASITISM: No cuckoo bees, either as for 20 to 25 cm and then branched, and at adults or as immatures, were found in the least one then branched numerous times. Animas nest. Of the 130 cells uncovered, 10 Some sections seemed filled with soil, where- were inhabited by large, active meloid larvae. as others were open. Laterals were filled with At the Apache site, four females ofthe cuckoo soil that was less consolidated than the cell bee Paranomada nitida were collected while closure. examining the entrance ofthe larger nest, one There seemed to be no pattern to the dis- while flying around the entrance ofthe small- tribution of occupied cells in relation to va- er nest, and still another as it emerged from cated cells within the nest. Most cells seemed the smaller nest. The excavated nest pro- to be tipped 10 to 30 degrees from the hor- duced three first instars and one second instar izontal with the closure end higher than the of the cuckoo bee. Several of these larvae rear. Cell arrangement needs further study were associated with flaccid Exomalopsis eggs but in one case, two in linear series were 4.0 which they had undoubtedly killed with their mm apart. Others may have been arranged strongly curved mandibles. Two first instars singly, and most seemed clumped side by side measured 1.5 mm long, the second instar 1.8 in close proximity. Cells in the collection mm. The head capsule of the first but not ranged from 6.2 to 7.2 mm in length (three second instar was darkly pigmented. A single measurements) and from 4.2 to 4.5 mm in oviposition insertion in the cell wall consist- maximum diameter (six measurements), and ed of a hinged, irregular flap, 0.5 mm long two were 3.5 mm in diameter at their con- and 0.3 mm wide, similar to the one de- nections to the laterals. Cell walls were harder scribed for Paranomada velutina (Rozen than the surrounding soil, and may have been 1977b). plastered with specially prepared material be- cause their texture seemed to be finer than Exomalopsis (Anthophorula) compactula that ofthe substrate. Cell shape seemed iden- Cockerell6 tical with that ofE. solidaginis, that is, some- what flatter on the bottom, and, at least in Two nests of this species were excavated some cells, with the plane of the closure not by me and Barbara L. Rozen on August 30, at right angles to the "longitudinal axis" of 1966, and September 2, 1966, at 5 miles north the cell. The internal surface of the cell was of Rodeo, Hidalgo County, New Mexico. Its extremely smooth and polished, and was immature stages as well as those of its par- waterproof when tested with a droplet of asite, Triopasites penniger (Cockerell), were water. One but not the other of two cells that recovered (Rozen, 1977b). Briefnotes on nest had not yet been provided with a closure was structure were taken, and samples of nest lined with a nearly transparent material over components were preserved in the American the rear wall in addition to the normal pol- Museum of Natural History. ished lining. When tested by being placed on NESTING ACTIVITY: Both nests were situ- a hot plate, this lining shriveled but did not ated in nearly horizontal ground. Each en- puddle, vaporize, or darken at 700°F. Frag- ments ofthis and other cell walls, when placed on a hot plate, did not change in appearance, 6Identified by P. H. Timberlake. even at 700°F. 1984 ROZEN: EXOMALOPSINI 17

The cell closure on the inside was a nearly Exomalopsis (Anthophorisca) chionura flat spiral ofapproximately four coils, in some Cockerell cases slightly convex and in others, slightly concave. On the external surface, the closures Rozen and MacNeill (1957) described were strongly concave and smooth. Both sur- nesting and other aspects of the life history faces were waterproofwhen tested with drop- of this species. Specimens of cocoons and lets. The closures were 3.4 to 3.5 mm in di- nests in the California Insect Survey, referred ameter on the inside (three measurements), to in that paper, on re-examination yielded 1.0 mm thick at the periphery the following new observations: Water ab- approximately sorption was difficult to test because the soil and 0.5 mm thick in the middle. preserved in the laboratory appeared very PROVISIONING: No information available. dense and absorbed a water droplet slowly. DEVELOPMENT: In several cells (fig. 16), Nonetheless, a droplet placed on the burrow there was clear evidence that most, ifnot all, wall was absorbed rather quickly in the two of the feces were applied to the cell closure cases tested, indicating that there was no spe- and very front of the cell, so as to obscure cial burrow lining. Several cell walls similarly the cell closure from the inside. In old cells tested seemed to absorb water at a slower rate from which adults had emerged, there was than the soil, suggesting a waterproof lining. an indication of a band of feces around the The nearly flat, spiral inside surface of two cell opening, and the cell contents consisted cell closures absorbed water moderately of mixed soil (presumably closure material) quickly, an indication of a lack of water- and feces. These observations suggest that the proofing. The strongly concave outside sur- feces are normally plastered at the front end face bore the distinct impressions of the fe- ofthe cell. In several other cells, feces seemed male's pygidial plate and absorbed water to have been distributed elsewhere, but those droplets rather slowly so that it may have cells may have been inhabited by larvae of been at least partly waterproof. Cell linings Triopasites. No cocoons were detected in the did not melt at 700°F. Laterals leading to cells field, but each of two cells preserved in the were filled with soil. Although samples were collection had an extremely thin cocoon of insufficient to determine whether cells were almost transparent parchment-like material slightly flattened on the bottom, a number of appressed to the cell wall and inner surface closures were not at right angles to the "lon- ofthe feces, with the result that the larva was gitudinal axis" of the cells. All preserved co- walled-off from the fecal mass. The presence coons were rounded at the front end and none ofcocoons and the placement offeces deserve conformed closely to the shape ofthe closure confirmation and further study. and cell rim. The front ends of several co- PARASITISM: Triopasites penniger was dis- coons in situ were so rounded that some space covered to be a cuckoo bee in the nests of existed between the cocoon and the front of this species (Rozen, 1 977b). Early stages ofa the cell. Only the most frontal point of the mutillid were also recovered. cocoon touched or nearly touched the center of the closure. Exomalopsis (Anthophorula) crenulata Exomalopsis (Anthophorisca) consobrina Timberlake Timberlake Parker (in press) provided some valuable The nesting biology of this species, called information on the nesting habits of this E. near chlorina, was briefly treated by Rozen species. (1977b). Exomalopsis (Anthophorisca) sidae Exomalopsis (Anthophorula) torticornis Cockerell7 Cockerell Barbara Rozen and I discovered numerous Hicks (1936) reported that four females adults ofthis and on were associated with a single nest of this species foraging mating species. 7Identified by J. G. Rozen, Jr. 18 AMERICAN MUSEUM NOVITATES NO. 2798

POLLEN

/MOIST AREA 27

WHITE WHITE MATERIAL MATERIAL '-~~~ ~30-. ,

* FECAL INNER SURFACE STREAKS OF COCOON FECES FECES 3mm 31 CAST SKIN 32 LI FIGS. 27-32. Exomalopsis sidae, cells, side view. 27. Open cell containing preliminary loads of fluffy pollen. 28. Egg and shaped food loaf. 29. Overwintering larva in cocoon. 30. Front end of cocoon, showing position of white fecal material. 31. Cell from which pupa had been removed, showing lack of cocoon and position of fecal material. 32. Cell containing hibernating larva of Melanomada sidaefloris and its feces. the flowers of Sida, 5 miles north of Willcox, two; and one had one. Males were not seen Arizona, on August 21, 1983. Returning on leaving any of the nests. August 29, I found and marked four nests. The entrance of each of the six nests was Another two were found when Marjorie Fa- open, not near a ground marker, and without vreau and I revisited the site on Sept. 3, 1983. a tumulus. Entrances ranged in diameter from A total of three nests were excavated. 2.75 to 3.0 mm. The open main tunnel, 3.5 DESCRIPTION OF NESTING AREA: As in the mm in diameter, descended almost vertically nest of the same species reported by Rozen to the level ofthe cells. Although one burrow (1977b), all six entrances were on the gently branched 8 cm down, branching generally be- sloping shoulder of a road between 2.5 and gan approximately 20 cm in depth, with 3 m from the edge of the pavement, among branches dividing, redividing and descend- the grasses and abundant Sida that grew there. ing at various angles. In three nests almost The plants were mowed periodically as part all cells were encountered between 20 and 30 ofhighway maintenance. Undoubtedly other cm in depth. nests occurred along the shoulder, consider- The main tunnel and branches were un- ing the large number ofadults on the flowers; lined, not waterproof, and often, if not usu- the six nests were scattered along 8 m. None ally, bore clear imprints of the females' py- of the entrances were shaded extensively by gidial plates. Although there was no clear the low, scattered vegetation. The soil was evidence of a masoned wall, such walls prob- fine grained, almost claylike, with few stones ably would have been undetected because of and roots. Dry on the surface, it was visibly the fine-grained nature of the substrate. Lat- moist below 5 to 7 cm and cracked in many erals leading to cells descended obliquely places because of the drying of the upper 10 downward and were filled after cell closure. cm. Cells, arranged singly, had their long axis NESTING ACTIVITY: Nests were generally varying in orientation from vertical to about occupied by more than one female: two nests 45 degrees from horizontal, the closure end had at least eight; one nest had four; two had always higher than the cell rear. Their shape 1 984 ROZEN: EXOMALOPSINI 19 corresponded to that of Exomlopsis solida- the coarse nature ofthe pollen obscured chan- ginis, with a maximum diameter of4.2 to 4.6 neling trails, although the mass developed a mm (eight measurements); length 7.8 to 8.5 somewhat uneven surface. Several older lar- mm (six measurements); and closure diam- vae were encountered circling the food loaves eter 2.3 to 2.8 mm (eight measurements), as that were now detached from the cell rear, so observed from inside the cell. In several cases that the provisions no longer contacted the there was some indication of a built-in (i.e., cell surface. Hence, the food was held away masoned or plastered) wall, although this fea- from the feces after the larva started defe- ture was difficult to detect in the fine-grained cation. Some ofthese food loaves were ovoid, substrate. The cell lining was clear, shiny, and but apparently later became dumbell-shaped waterproofand embossings from the female's as the larvae became larger. Detaching ofthe pygidial plate were in evidence to a greater mass took place before defecation, but soon or lesser extent, particularly toward the top afterward the larva, while still feeding, started of the cell. The wall immediately next to the to defecate. closure seemed somewhat rough in many in- In the case of larvae developing in late stances, compared with the cells ofother Ex- summer (see below), the feces were applied omalopsis. as yellow smudges on the entire cell surface. The inner face of the closure was a non- The larva then applied silk on the thin layer waterproof concave spiral ofabout four coils of feces. In the early stages of cocoon con- that were almost indistinct toward the mid- struction the silk consisted ofdistinct strands dle. One closure was observed to have a running in all directions and adhering to the waterproof, concave surface on the outside, feces, with many fenestrations in the fabric. indicating that this species agrees with other When completed, the cocoon (fig. 29) over Exomalopsis regarding this feature. Instead most ofthe cell surface was thin, slightly tan, ofbeing completely smooth this surface bore semitransparent, nonfenestrated, fine-fi- markings of the female's pygidial plate. bered, parchment-like material with a shiny PROVISIONING: A number ofopen cells (fig. but uneven (crinkly) inner surface (i.e., facing 27) contained quantities of fluffy, loose pale the cocoon lumen) and with an outer surface pollen that seemed far more copious and less to which fecal pollen adhered. Only toward compact than the preliminary stores of E. the closure (top) end was the cocoon more solidaginis. Such masses seemed to occupy complex (figs. 29, 30). Here it was thicker, half the cell and were moist only below. All approximately 1 mm thick at the closure it- completed food loaves were composed ofpale, self, and was composed of perhaps two or large-grained pollen of Sida. They were three layers of pollen-laden feces separated shaped by the females into elongate loaves by layers of silk. As a uniform feature, the (fig. 28), 4.3 to 5.2 mm long (seven mea- innermost layer offecal material consisted of surements), 3.3 to 3.75 mm in horizontal an opaque white, amorphous, fine-grained width (seven measurements) and 2.4 to 2.7 material (fig. 30) that contrasted with the tan- mm high (three measurements). The loaf in ner, granular feces elsewhere in the cell. This each case had a distinct, moderate-sized foot white blotch had an uneven outline, roughly at the front end that, at least in several cases, the diameter ofthe closure, and rested on the touched the cell floor. The loaf was mealy inside ofthe normal pollen-laden feces at the and moist throughout, slightly sweet to taste front end ofthe cell. In some cases it seemed and had a smooth surface. The rounded rear to be covered by a nearly transparent, ad- of the food mass was fitted to the cell rear as hering sheet of silk, and in other cases it may is characteristic in the genus. have been partly or completely exposed to DEVELOPMENT: White, curved, and shiny, the cell lumen. The completed cocoon, when the eggs (fig. 28) were placed on the top front extracted from the soil, was fragile in ap- offood loaves in the midsagittal plane ofthe pearance and collapsed easily except at the cell. Two measured 2.2 to 2.3 mm long and front end where the layers of feces and silk 0.45 mm in maximum diameter. Young lar- provided a firmer structure. Larvae in the vae crawled over the food surface as they fed, cocoons (fig. 29) were quiescent, pale over- gradually reducing the size of the mass, but wintering forms, curled on their backs with 20 AMERICAN MUSEUM NOVITATES NO. 2798 their heads somewhat lower than the tips of indicating that the three nests were used by the abdomens. more than one generation. Although all fully fed larvae encountered ADULT ACTIVITY: Adults were active dur- on September 3 were either spinning or had ing my three visits to the site. Peak activity spun cocoons (table 1) as described above, seemed to be late morning, but many indi- no individuals ofthe previous generation had viduals were still flying in midafternoon. For- spun cocoons. This is known because on Au- aging females flew rapidly from flower to gust 29, eight active and developing pupae of flower. Males flew in a similar fashion, prob- the previous generation were recovered from ably in search of females in flowers. Male cells (fig. 31) that had no cocoons and in which flights were swift, not the slow deliberate flight the feces were plastered against the bottom of certain other Exomalopsis. (rear) of the cell in a mass of flat, elongate PARASITISM: Only a single male ofMelano- tannish pellets that extended as streaks ap- mada sidaefloris was seen flying over the proximately two-thirds up the cell wall. In all flowers during the three days of observation, cases, most of the feces were at the bottom excavation, and collecting, and only a single ofthe cell and the amorphous, opaque, white, female8 was collected, as it emerged from a nongranular substance was the innermost nest. In spite of the lack of adults of this material, the last to be deposited. The cast parasitic bee, the cells of Exomalopsis sidae larval skin, with its characteristic mandibles, were heavily parasitized by this species (see could usually be identified adhering to the table 1). white material or to the tan feces. In the nests Whereas Exomalopsis sidae appeared to excavated on September 3, adults of the first have two generations a year, the number of generation had emerged, but their vacated annual generations for Melanomada sidae- cells, now filled with soil, revealed the feces, floris was not clear, nor was the time phasing larval skins, and no cocoons. of the parasite population with that of the Hence, it seems clear that one generation host. A possibly related problem that needs did not spin cocoons, whereas the subsequent explanation was the low number of adult one did. This phenomenon was suggested Melanomada compared with the high rate of earlier by Rozen, Eickwort, and Eickwort parasitism, as revealed by immatures in the (1978) with respect to Protepeolus singularis. nests. In the case of Exomalopsis sidae, the non- Two oviposition sites ofMelanomada were cocoon-spinning generation probably occurs found in cells of Exomalopsis sidae that al- in spring or early summer, and the cocoon- ready contained postdefecating Melanomada spinning generation apparently overwinters larvae. Each was a hinged flap of cell lining as quiescent larvae in cocoons. Surprising is and wall, as is characteristic ofthe Nomadini the marked difference in placement of the (Rozen, 1 977b), found halfway from the rear fecal material in the two generations-at the ofthe cell to the closure. Active Melanomada bottom and lower sides ofthe cell in the ear- larvae were found in various positions on the lier generation, and mostly at the top but also loaf, indicating that they moved while feed- totally covering the entire cell surface in the ing. Postdefecating larvae (fig. 32), rigid, late summer, overwintering generations. The white, and without cocoons, were invariably importance of the cocoons for overwintering found with the anterior end closest to the cell larvae is unknown. closure. The feces had been plastered to the Postdefecating larvae of the noncocoon- rear (bottom) of the cells, as elongate partly spinning generation were not observed. It is unknown, therefore, whether they had pro- 8 This female agreed with those Melanomada females jecting salivary lips and labiomaxillary re- associated with Exomalopsis consobrina (Rozen, 1 977b) gions characteristic of cocoon spinners, or except that the integument of its head, mesosoma, and whether they lacked such appendages was somewhat blacker. The first flagellar seg- projecting adapta- ment was nearly concolorous with the other segments, tions. Although seasonal larval dimorphism rather than distinctly redder as was the case with the seems unlikely, this matter should be inves- consobrina associates. Its clypeus was also somewhat tigated. more densely punctate. Until more individuals associ- Vacated cells ofprevious generations as well ated with both host bees can be examined, these Me- as cells currently occupied were uncovered, lanomada will be considered conspecific. 1984 ROZEN: EXOMALOPSINI 21

TABLE 1 Collection of Immatures of Exomalopsis sidae and Melanomada sidaefloris on Two Different Dates in 1983 August 29 September 3 Exomalopsis Melanomada Exomalopsis Melanomada IMMATURES sidae sidaefloris sidae sidaefloris Pupae 8 lb 0 0 Postdefecating larvae 0 2 18 8 Postfeeding larvae 1 1 1 4 Feeding larvae Large 0 2 5 2 Intermediate 0 1 2 0 Small 1 1 9 0 Eggs 2 0 1 0 Total 12 8 36 14 a (See footnote, p. 20.) b This pupa remained alive but inactive and did not develop in contrast with the pupae of Exomalopsis sidae collected at the same time and which were active and developed rapidly. fused pellets that extended roughly halfway Three to 4 m high, the bank was without to the cell closure. vegetation at the nest area, although else- where it was partly covered with plants. The Eremapis parvula Ogloblin lower level was sand intermixed with large rocks, whereas the upper level, where most Neff (in press) treated many aspects of the nests occurred, consisted of an even, fine- nests ofthis Argentinian species, the only one grained orange sand that was so hard and in the genus Eremapis, and compared its bi- compact that it had to be chipped with a ology with that of other exomalopsines. hunting knife to be excavated. The bees were common here but not elsewhere. They may Ancyloscelis apiformis (Fabricius)9 have chosen this site because it lacked any sign of moisture, unlike most roadside em- Aspects ofthe nesting biology ofthis species bankments on Trinidad in late February and have been treated in considerable detail by early March. The bank faced south and was Torchio (1974) and Michener (1974a), under exposed to the sun during most of the day. the name ofAncyloscelis armata (Smith), and Intermixed with numerous nests of this briefly by Linsley, MacSwain, and Michener species were a great many of Melitoma seg- (1980). Brethes (1909) also commented on mentaria (Fabricius),'I although most ofthese the nests ofthis species. Although my obser- were inactive. vations parallel theirs in most respects, new NESTING ACTIVITY: Although numerous fe- information and, in certain instances, infor- males of Ancyloscelis apiformis and one of mation that seems to disagree with their ob- Melitoma segmentaria were transporting servations, warrant yet another review ofthe pollen to their nests, plants ofthe genus Ipo- biology of this species. moea, to which both genera are restricted, DESCRIPTION OF NESTING AREA: Nests of were not noticed in the vicinity ofthe nesting this species occurred in a 10-m section of a area." Nest entrances ofboth species in 1968 vertical bank that extended for 30 m along a paved roadway in Maracas Valley, Trinidad. The site was studied by me and Dr. F. D. 10Identified by Dr. Thomas J. Zavortink. Bennett, who had discovered it in late Feb- II John L. Neff(in litt.) informed me that Ancyloscelis ruary and early March 1966, and again by apiformis "collects both Ipomoea and Convolvulus pol- me and Barbara Rozen in early March 1968. len in Texas and some ofthe oddAncyloscelis with hooked hairs on the mouthparts are apparently restricted to the 9 Identified by J. G. Rozen, Jr. Pontederiaceae." 22 AMERICAN MUSEUM NOVITATES NO. 2798

33

36 FIGS. 33-36. Ancyloscelis apiformis. 33. Cell with egg and food loaf, side view. 34. Same, showing food and egg in cross section at line ab in figure 33. 35. Cell, side view, showing cocoon and feces. 36. Three cells in series, separated by short tunnels. Scale refers to figures 33-35.

seemed to be most abundant in the 10 to 15 females sometimes construct their own bur- cm deep excavations made by us in the face rows, as shown by both my observations and of the embankment in 1966. These depres- those ofMichener, indicates that Ancyloscelis sions had existed for two years without ap- is quite independent from Melitoma so far preciable change, a fact signifying the per- as any need to use another bee's burrow to manency of the bank face. Torchio's nests start its own. Further observations may shed (1974) from Mexico came from a vertical more light on this interesting microdistri- adobe wall that was "completely honey- butional overlap. combed" with old burrows and cells of two Because so many twisting and anastomos- species, A. apiformis and Melitoma Sp.,12 a ing burrows, both new and old, occurred in situation that closely parallels the Trinida- the nest area, we could not excavate a single dian nesting site. Michener's nesting area, "a nest so as to learn its form, as also was the small cave under a huge rock" near the Uni- case for Torchio's study (1974). Only Mich- versidad del Valle in Cali, Colombia, was ener (1 974a) gave a nest diagram. In Trinidad inhabited by "probably the same two antho- females of A. apiformis carrying provisions phorines." However, Michener could not be entered not only Ancyloscelis burrows (about absolutely certain that the Ancyloscelis was 3.0 to 4.0 mm in diameter) but also the con- apiformis because of the lack of males. His siderably larger (5.5 mm) ones of M. seg- nest site differed from mine and probably mentaria. The entrances constructed by An- Torchio's because it was in "deep shade." cyloscelis were both on the vertical cliff and Michener suggested that the association of also under the overhang where we had made Ancyloscelis and Melitoma might be because excavations in 1966. These entrances lacked "Melitoma plays a significant role in prepar- turrets, and the walls of the main burrows ing a site for later occupancy by Ancylosce- were moderately rough with no indication of lis." Although a reasonable hypothesis, no a soil lining as found in some tunnels by Tor- facts other than dual site occupancy suggest chio (1974), and with no evidence of liquid this. Such occupancy could as easily be a con- having been used by the females to soften the cordance of nesting requirements involving soil during construction. Because Ancylo- dry vertical banks near their common food scelis entered only those Melitoma burrows source, Ipomoea. The fact that Ancyloscelis that had no turrets, they were probably re- using vacated burrows of Melitoma. Ancy- 12 Linsley, MacSwain, and Michener (1980) identified loscelis females were also seen entering nests this Melitoma as marginella (Cresson). ofprevious generations ofAncyloscelis, as well 1 984 ROZEN: EXOMALOPSINI 23 as fresh ones. All tunnels were open except parted a faint shine. The glistening effect was for some that were blocked by thin septa; enhanced by placing a small fragment on a none was filled with soil. hot plate at 700°F. where the lining darkened Contrary to Michener's observation that but did not char. Another fragment of cell new cells are attached to previously occupied lining with substrate attached was placed in ones, Ancyloscelis in Trinidad probably did water, resulting in the substrate "melting not construct cells along old tunnels, be they away," leaving the curved cell lining undis- ofAncyloscelis or Melitoma, because current solved with the adhering soil particles at- cells always seemed to be attached to fresh tached. This remaining lining was nearly as Ancyloscelis burrows. Old nests of Ancylo- thick as the larger soil particles. These ob- scelis could be recognized immediately be- servations seemingly disagree with those of cause the sand surrounding them was darkly Torchio (1974), who stated that the inner stained, whereas no stain was seen in the vi- cell layer was composed of "small, tightly cinity of new burrows or cells. Ancyloscelis appressed soil particles" and "not coated with did not wall-off parts of old burrows or fill- secreted materials nor were secretions incor- in tunnels that were too large, as did Exo- porated into it." The coating was presumably malopsis chionura (Rozen and MacNeill, responsible for waterproofing the cell; a drop- 1957). However, Torchio (1974) reported that let ofwater placed on the wall was very slowly cell walls were sometimes partly built into absorbed. Other details of cell construction old cells. given by Torchio are difficult to relate to my Cell orientation in the nest ofA. apiformis observations. Further study is needed to rec- varied from nearly horizontal to almost ver- oncile these matters. tical, in agreement with the observations of After the cell was provisioned, it was closed both Torchio and Michener. The front ofmost with a moistened thin plug of soil that sub- but not all cells was as high or higher than sequently dried. The closure (fig. 33) on the the rear. Cells arranged singly were connected inside was a distinct spiral that was flat or, to an unlined lateral about 7 mm long; at more often, slightly convex, in contrast to the least some of these laterals were walled off concave cell closures ofmost bees. The outer from the main burrow by a septum. Most surface ofthe plug was very smooth and con- cells were in series of two or three, and were cave, and showed no indication of a spiral, either directly connected to one another, or, whether or not it formed the rear of the next more often, were separated by sections of cell in the series. Several closures measured tunnel, 5 to 7 mm long (fig. 36), the walls of 0.5 to 1.5 mm thick at the middle. Closures which were not waterproof. These facts cor- seemed to be water absorbent on the inside, roborate Torchio's observations, except I am but more or less waterproof on the smooth uncertain that the tunnels between cells are outer face. Cells were constructed only a short "generally filled with loose soil." distance into the bank, the deepest not pen- Slightly flattened on the lower side, the cells etrating much beyond 10 cm. (fig. 33) were about 7.0 mm long and 4.5 to PROVISIONING: Females of Ancyloscelis 4.8 mm in maximum diameter (four mea- apiformis transported the pollen dry to the surements), figures that correspond with those cell. Apparently the pollen was then mixed of Michener and are slightly less than Tor- with nectar, and placed in a slightly moist, chio's. At the closure, the cells ranged from fluffy condition at the rear of the cell. When 3.0 to 3.5 mm in diameter, somewhat less a sufficient quantity was accumulated, the fe- than the diameter ofthe tunnel. The cell wall male added considerably more liquid, and was smooth, more so than the tunnel walls shaped it into an elongate loaf (fig. 33) that or connecting burrows (but not quite as was no longer fluffy. This light tan food mass smooth as the inner surface of the Melitoma resting on the floor of the cell was homoge- cells), and was not obviously made of soil neously mealy, and 5.0 to 5.5 mm long, 3.75 mixed with some secretion, as seemed to be to 4.0 mm in maximum width, and 3.0 to the case with M. segmentaria. The inner cell 4.0 mm in maximum height (three measure- surface was the same color as the matrix, and ments). Occupying a large part ofthe cell (fig. lined with a varnish-like material that im- 33), the front part of the mass was slightly 24 AMERICAN MUSEUM NOVITATES NO. 2798 higher than the rear, and the top surface was not sandwiched between layers of silk. These broadly grooved longitudinally (fig. 34). The features were also observed by Torchio. provisions lacked a "foot" (Rozen and The larva was oriented in the cocoon with MacNeill, 1957). its anterior end closest to the closure. Al- DEVELOPMENT: The slightly curved egg (fig. though some larvae were found resting on 33) was placed in the midsagittal plane ofthe their sides, most rested on their dorsa. At cell on the top surface of the loaf, so that it least some such larvae were totally quiescent rested in the groove. As indicated by Torchio, and flaccid. Pupae were usually found resting but contrary to Michener's observations, on their dorsa with their heads closest to the some eggs (fig. 33) were situated on the mid- cell closures. dle of the top; others, more toward the rear. As all stages (eggs, larvae, pupae, and adults) The anterior and posterior ends of the egg were found at the same time, A. apiformis touched the pollen mass, and the middle part may have numerous generations a year; curved upward; neither end was firmly at- whether or not there is a seasonal inactive tached to the provisions. Measuring approx- period is unknown. imately 3.0 mm long and 0.7 mm at maxi- PARASITISM: No parasitic bees were asso- mum width, the egg had a smooth transparent ciated with the nesting site nor were meloids chorion that lacked sculpturing. The col- or mutillids encountered. The commonest lapsed chorion remained behind when the possible parasites were adult leucospids which larva emerged and crawled away. flew in a dancing fashion along the embank- We did not make detailed observations on ment, although none was found in the cells. the larval instars and their activities. How- Walls and closures ofmany old cells had been ever, the elongate young larvae moved about bored by some small creature, perhaps an ant, the pollen mass while they fed and were found and the contents removed, but none was en- both on top of and underneath the food loaf. countered during our observations. As the larva fed, the pollen mass became smaller but was not obviously channeled. Ancyloscelis panamensis Michener However, Torchio found that in Mexico the Nests of the species were discovered and food loaves were channeled by the feeding described Michener activity of the larva. by (1954). Larvae started defecating considerably be- ANALYSIS OF THE NESTING fore the provisions were totally consumed; BIOLOGY OF THE EXOMALOPSINI one larva was defecating while one-fourth to one-third of the food remained. Feces were This section summarizes my observations extruded as elongate pellets consisting of and those reported in the literature, on the partly collapsed pollen grains. The pellets, nesting behavior of exomalopsine bees. It is about 0.5 mm long, were smaller than those intended to provide an overview oftheir be- ofmost bees. Feces (fig. 35) were rather evenly havior and to identify areas where knowledge appressed to the floor, roof, and side of the is lacking so that future studies can be more cell but were not attached to the cell closure focused, especially for data of phylogenetic or to the wall near it, just as described by significance. Although our knowledge ofthese Torchio. bees has increased over the last 25 years be- When the still-active larva completed def- cause of interest in bee behavior, the follow- ecating, it spun a cocoon (fig. 35) first by at- ing account is based on only 13 of the 26 taching silk to the feces, then adding more genera and subgenera assigned to the tribe by silk to the feces, while at the same time, spin- Michener and Moure (1957) and Toro (1976). ning a cap near the cell closure. This cap was Information discussed below has been pre- not firmly attached to the closure and usually sented for the first time under the appropriate just touched it near the middle, whereas the species heading in the foregoing section, or rest ofthe cocoon could scarcely be separated the references to it can be found there. from the excrement. The finished cocoon was CHOICE OF NESTING SITE: The Exomalop- thin, semitransparent, light tan, and consist- sini appear to be predominantly ground nest- ed of a single layer: that is, fecal material was ing with only Paratetrapedia gigantea having 1984 ROZEN: EXOMALOPSINI 25 been discovered nesting in a fallen dead tree.'3 tricha, Tapinotaspis caerulea (Vogel, 1974), The surface slope of nest sites varies consid- Exomalopsis globosa, similis, chionura, per- erably. Both species of Tapinotaspis, Ere- haps sidae, crenulata, Eremapis parvula, mapis parvula, and all Exomalopsis bur- Ancyloscelis apiformis (Torchio, 1974; Mich- rowed into slightly sloping to horizontal ener, 1974a; and present paper), and Ancy- ground (even though females of E. chionura loscelis panamensis. descended into cracks rather than burrows, With other species, either only isolated nests in a gently sloping surface before digging more have been found or nests seem widely dis- or less horizontally). A nest of Monoeca pefetlas-opposed to being aggregated. These schrottkyi was reported in a vertical surface, are: Tapinotaspis tucumana, Exomalopsis and, although I found Monoeca lanei nesting aureopilosa, perhaps solidaginis, probably on a horizontal path, the entrances were at compactula, and consobrina. the edge of the path where it rose to the sur- The underlying adaptive basis (or bases) rounding field, a condition which might be for nesting in aggregations or singly is not interpreted as vertical or at least strongly in- understood in any of these cases and seems clined. Lanthanomelissa burrowed into a to have little phylogenetic significance be- vertical surface. Such surfaces were also the cause both behavior patterns exist in Exo- choice ofAncyloscelis apiformis in six widely malopsis and Tapinotaspis, and because even separated localities, but a flat sandy clay floor within Exomalopsis both patterns occur protected by the roofin an abandoned house within the subgenera Anthophorisca and Ex- was used at another locality (Linsley, omalopsis sensu stricto. MacSwain, and Michener, 1980). Ancylo- Communal nesting behavior, that is, more scelis panamensis nested in slightly sloping than one female actively occupying a nest, is ground. well documented in the tribe. Claude-Joseph To what extent exomalopsines prefer bar- (1926) recorded numerous females in a nest ren ground to a vegetated surface is not al- of Tapinotaspis caerulea, and Vogel (1974) together clear. Although nest entrances in also indicated that there may be more than barren or sparsely vegetated ground are easier one female to a nest of the same species. All for the researcher to find, the great number observed species of Exomalopsis seemed to of observations recorded here suggest that have more than one female to a nest, and a most exomalopsines prefer barren or semi- nest ofE. aureopilosa contained 884 females. barren nest sites. One notable exception was On the other hand, species in other genera Tapinotaspis tucumana, a number ofnests of are known to have only a single female to the which were found in dense, low vegetation. nest, including Paratetrapedia oligotricha, Numerous nests of Ancyloscelis panamensis Tapinotaspis tucumana, and Monoeca lanei. were partly hidden in grass. Ancyloscelis panamensis seems to have but a SOCIAL ORGANIZATION: A considerable di- single cell to a nest as was also suggested for versity exists among exomalopsines with re- Eremapis parvula (Neff, in litt.). spect (1) to solitary versus gregarious nest Several workers have investigated the so- associations; (2) to parasocial (sensu Miche- cial relationships among adults in Exoma- ner, 1 974b) as opposed to solitary nesting lopsis nests. Michener (1966) concluded that behavior; and (3) to the continued use ofnest females of Exomalopsis solani provisioned areas or even nests from one generation to cooperatively because foragers with unde- the next. The following species nest in loose veloped ovaries as well as those with devel- or tight aggregations: Paratetrapedia oligo- oped ovaries collected pollen. Raw (1977) de- cided that both Exomalopsis globosa and E. 13 Robert W. Brooks (in litt.), who kindly read this similis were at least quasisocial because more manuscript, has noted that some Paratetrapedia (Para- pollen gatherers inhabited a nest than there tetrapedia) and P. (Lophopedia) have "tridentate man- He dibles which may be related to wood nesting." He "saw were cells being built and provisioned. P. (Paratetrapedia) lugubris Cresson nesting in a leaf- further concluded that these bees were pos- cutter bee (Megachile rotundata) trapnest and also in sibly semisocial, because the large proportion what appeared to be beetle burrows in April 1981 at of foragers with vestigial o6cytes suggested a Curundu (Panama Province), Panama." division of labor with respect to egg layers 26 AMERICAN MUSEUM NOVITATES NO. 2798 and nonegg layers in a nest. He surmised that retarded the absorption ofwater droplets. This nests were started by a few females because appears akin to the waterproof lining of the two and three females per nest were discov- cells, and is very different from plastered ered in nests from which no bees had yet walls. emerged. He noted that "the numbers ofadult In general, available data do not reveal great bees in the larger Exomalopsis nests were very differences among taxa with respect to nest similar to the numbers of cells in them from configuration except as noted below. So far which bees had emerged, which suggests that as is known, all nest entrances and main tun- most of the emerging bees were females and nels remain open during nest construction that they remained associated with their natal and provisioning. Laterals leading to com- nest, forming sister colonies." pleted cells are filled or partly filled with soil. Occupation of a nesting site, or of even a Nest patterns for all representatives seem to single nest, for more than one generation, as consist of a main tunnel, which may (Exo- evidenced by the presence of numerous va- malopsis) or may not (Tapinotaspis tucu- cated cells, has been observed in a number mana) branch many times, with laterals ex- of species all of which (with the possible ex- tending variable distances to cells. The main ception of Ancyloscelis apiformis) are com- tunnel often is vertical, sometimes straight, munal: Tapinotaspis caerulea (Vogel, 1974), but more often somewhat meandering in those Exomalopsis aureopilosa, globosa, similis, species that nest in horizontal or nearly hor- solidaginis, compactula, crenulata, chionura, izontal surfaces. Nests built into vertical banks consobrina, sidae, and Ancyloscelis apifor- tend to have the main burrow and branches mis. The data suggest that communal bees essentially horizontal (Paratetrapedia oligo- tend to occupy nest sites and nests for more tricha; Lanthanomelissa species; and Ancy- than one generation. loscelis apiformis, Michener, 1 974a). Depths NEST STRUCTURE: Considerable informa- of nests vary considerably. Perhaps E. au- tion exists on the nest elements of exoma- reopilosa has the deepest nest of any known lopsines, which, except for Paratetrapedia gi- bee, recorded at 5.3 m. As one would expect, gantea, are known to nest in the ground. this large (deep) nest was correlated with adult Among exomalopsines, only Monoeca lanei population size: it was inhabited by 884 fe- has been recorded to construct turrets around males and 46 males. Of the 15 species for nest entrances. The adaptive and phyloge- which we have information, only Ancylo- netic significance ofturrets in this species (sit- scelis panamensis appears to construct but a uation not recorded for Monoeca species, or single cell to a nest. Its close relative, A. api- M. schrottkyz) is unknown, but some meli- formis, clearly has numerous cells to a nest. tomines and anthophorines do build distinct Although few differences can be detected turrets. in general nest pattern within the tribe, cell Exomalopsis chionura, crenulata, com- features vary considerably and give promise pactula, perhaps solidaginis, Ancyloscelis for further analysis. Most species arrange cells apiformis (Torchio, 1974), and probably oth- singly at the end of laterals, but linear series er exomalopsines at times line burrows with of cells are constructed by some species. Ac- soil that is different in consistency or particle cording to Claude-Joseph (1926), Tapinotas- size from the surrounding substrate. In some pis caerulea excavates cells in series, although cases, this lining appears to be an attempt on Vogel (1974) claims that cells of this species the part of the female bee to wall-off other are arranged singly. Exomalopsis chionura bee burrows or cells that are too large or per- builds cells both singly and in series, and An- haps in other ways unsuitable. To what extent cyloscelis apiformis usually but not invari- this is in the common repertoire of behavior ably has cells in series, with individual cells patterns for the tribe is unknown. In the few of a sequence often being separated by sec- cases where it has been identified, it seems tions of tunnel. that this type of lining does not retard the Cell orientation, that is, angle of the "long absorption of water droplets. In one case axis" ofthe cell to the horizontal, is extremely (Monoeca lanei), the burrows as well as the variable. Some species have uniform cell ori- cells were lined with a waxy material that entation, which may be horizontal (Exo- 1 984 ROZEN: EXOMALOPSINI 27 malopsis similis, Eremapis parvula); slightly grained. In certain other species (Exomal- inclined (20 to 30 degrees) toward the rear opsis chionura and Ancyloscelis apiformis) (Tapinotaspis tucumana, Exomalopsis com- there is clear evidence that old cells and bur- pactula); considerably inclined (30 to 45 de- rows can be walled-off with the cement-like grees) toward the rear (Exomalopsis globosa, material. With other taxa, there is no infor- consobrina); or vertical (Paratetrapedia oli- mation about a specially formed plaster wall gotricha, Exomalopsis aureopilosa, Monoeca (Tapinotaspis tucumana, Monoeca lanei, lanei, Monoeca species). With other species Exomalopsis solani, and consobrina). Al- cell inclination may vary from horizontal to though I noted no plaster cell wall with An- vertical in a single nest, as in Exomalopsis cyloscelis apiformis, Torchio (1974) clearly chionura, solidaginis, and Ancyloscelis api- noted such a wall. This suggests either that a formis. Considering the variation in cell ori- species may construct such a wall only under entation, one must ask the question, why are certain circumstances, or, more likely, that some species behaviorally locked into a fixed plastered walls can be easily detected when orientation oftheir cells, whereas other species there is a marked difference between the tex- are quite variable in this regard? The great ture of the substrate and the fine-grained ce- range of variation within the genus Exoma- ment-like material. This is not to say, how- lopsis seems to suggest that cell orientation ever, that all exomalopsines construct is not a significant feature for phylogenetic masoned walls. Data are lacking to draw such interpretation. conclusions. Female exomalopsines apply a Information on cell shape for exomalop- very thin coating of a transparent or semi- sines is sketchy: (1) because of difficulties in transparent, smooth, shiny material on the analyzing and describing cells, even in the cell wall. This film probably accounts, in all best of circumstances; (2) because cells are cases, for the waterproof nature of the inner normally extracted from the ground as bro- surface of cells, although the waterproof na- ken fragments that are difficult to reassemble; ture ofthe plastered part ofthe cell walls has (3) because cell shapes are often complicated; yet to be tested. The only noted exception and (4) because observers often tend not to (Parker, in press), recorded that droplets of examine cells closely under a microscope. water were quickly absorbed by the cell wall Cells described in the present paper were ob- of Exomalopsis crenulata. In most cases served through a stereoscopic microscope, in tested, this lining was nonwaxlike in that it the field, in the laboratory, or in both. These did not melt even at temperatures of 700°F. cells are not symmetrical around a straight However, the cell lining ofMonoeca lanei did longitudinal axis. With cells that are not ver- melt, somewhat below the boiling point of tical, the lower surface (floor) is flatter than water, when I tested it. Similarly, Vogel (1974) the upper surface (ceiling), which is therefore reported that the cell lining of Tapinotaspis slightly more vaulted, all as seen in longitu- caerulea melted when heated. He also noted dinal section. Such shapes appear to be char- that the lining became opaque, presumably acteristic of Tapinotaspis tucumana, species when the cell dried out, which is also char- of Exomalopsis described herein, and An- acteristic of Monoeca lanei. In contrast, the cyloscelis apiformis. Vertical cells, at least of cell linings of the presumably related Tapi- Monoeca lanei, also seem to be slightly flatter notaspis tucumana did not melt when heated. on one side than the other, so that this feature Zucchi (1973) claimed that the cells of Ex- may be characteristic of the Exomalopsini omalopsis aureopilosa were coated with a irrespective of cell orientation. waxy layer produced by glands associated with In a number of taxa (Exomalopsis chio- metasomal terga IV through VI. These rather nura, compactula, crenulata, Ancyloscelis profound differences do not correlate well with apiformis, Torchio, 1974), the female con- the presumed relationships of these exomal- structs the cell first by making a cavity in the opsine taxa, and therefore raise questions that soil and then by plastering its surface with a warrant further study. It is clear from inves- mixture of fine soil mixed with some other tigations ofBatra (1972) and Cane (1983) that substance, perhaps a secretion, so that the the chemistry ofcell lining in bees is far more surface of the cell wall is smooth and fine complicated than bee specialists had thought 28 AMERICAN MUSEUM NOVITATES NO. 2798 heretofore, and that sophisticated chemical case at least with some Anthophora and Em- analysis will yield considerable new infor- phoropsis. 14 mation. An unusual feature of the cell closure of Cell closures in Exomalopsini are appar- some exomalopsines is that it has a distinct, ently invariably spiraled on the inside, that strongly concave outside surface (fig. 18), is, the side of the plug facing the cell lumen. which abuts the lateral. This surface is usually Spiral closures are characteristic of many di- as smooth and hard as the cell wall, clearly verse groups ofbees (Apoidea) and are found different from the more loosely packed fill of in the following families: Colletidae (Rozen, the lateral, and apparently not absorbent to 1984); Stenotritidae (Houston, 1975); Ox- droplets of water. Such outer surfaces were aeidae [New Information]; Andrenidae (Ma- detected in Exomalopsis solani, solidaginis, lyshev, 1935); Halictidae (Torchio et al., compactula, chionura, sidae, and Ancylosce- 1967); Melittidae (Malyshev, 1935); Cteno- lis apiformis. This may well prove to be char- plectridae (Rozen, 1978); and Anthophori- acteristic of all Exomalopsis. Nests of Exo- dae (Malyshev, 1935). Spiral closures to cells malopsis chionura and sidae differ from the have not been recorded for the Megachilidae others of the genus in that the outer surface (including the Fideliinae) (Rozen, 1977a) ex- is clearly embossed with the impressions of cept in one case (Rust, Thorp, and Torchio, the female's pygidial plate, and therefore is 1974) and Apidae (Malyshev, 1935). If this not as smooth as the outer surfaces ofclosures is indeed the case, then the nonspiral closures of other species. In contrast to this type of such as found in Halictus sensu lato (Ma- closure is that of Tapinotaspis tucumana and lyshev, 1935), most Megachilidae (Rozen, Monoeca lanei, in which the soil of the clo- 1977a), Colletes (Malyshev, 1935), and Ap- sure grades into the fill of the lateral so that idae, are independently derived apomorphic there is no obvious demarcation between the traits. textures of the two substances. Although the spiral nature of the closures The matter of gas exchange between the is characteristic of the tribe, other aspects of interior of the cell and the outside has rarely the closures vary. For bees in general, most been addressed with respect to ground-nest- such closures are conspicuously concave on ing bees. With some taxa, the closure seems the inside. This is also true within the tribe sufficiently unconsolidated to permit oxygen for Tapinotaspis tucumana, Monoeca lanei, to enter the lumen and carbon dioxide to but not for Ancyloscelis apiformis, where the escape. This method of exchange seems par- closure is flat or even slightly convex on the ticularly plausible in those bees in which the inside. Within Exomalopsis, in some species cocoons are more porous where they contact (solani, solidaginis, sidae) the spiral is con- the cell closures. However, the cell closures cave, but in compactula, consobrina, and ofexomalopsines seem well consolidated, and chionura it is flat or nearly so. The spiral therefore, not very permeable. This consol- closure characteristic of this tribe may be idation is characteristic of all exomalopsines water absorbent on the inside, that is, water on the spiral face ofthe closure, and in certain droplets placed on the spiral are absorbed species, on the outer surface as well (fig. 18). nearly as rapidly as are water droplets placed This is another matter that needs further in- on the substrate. This seems to be true for vestigation. Monoeca lanei, Exomalopsis solani, solida- PROVISIONING: Vogel (1974) reported that ginis, chionura, sidae, and Ancyloscelis api- some Exomalopsini are known to provision formis. With other taxa, the closure is more their nests with oil and pollen rather than or less waterproof, when tested the same way, nectar and pollen. The oils are obtained from including Tapinotaspis tucumana and E. compactula. The reason for the retarded rates 14 Philip F. Torchio (in litt.) after reading this manu- ofwater absorption in the case ofthe last two script wrote: "In studying Pseudomasaris edwardsii, I species is not understood, but it is not the was amazed to learn how waterproof cell walls and caps result of wax or a shiny film of material cov- became when the wasp simply added nectar (sugar) to ering the inner face of the closure, as is the the soil it manipulated for cell construction." 1984 ROZEN: EXOMALOPSINI 29 certain plants with special oil-secreting glands. though he refers to the mass as having a More recently, Neffand Simpson (1981) sur- "foot," this projection at the bottom (rear) veyed the New World anthophorids with re- of the vertical food mass in a vertical cell spect to oil collecting and discussed the pos- seems in the wrong position to be the same sible roles of the floral oils in the biology of as the foot of other species. The elongate, oil-collecting bees. shaped provisions of Eremapis parvula, as Foraging females returning to the nest de- drawn by Neff (in press), appear similar to posit pollen loads associated with liquid (pre- those of Exomalopsis and are attached only sumably oil or nectar) on the lower part of at the front and rear of the cell, but a foot is the cell (that is, the rounded rear of the cell not shown or described. The food mass of in vertical and semivertical cells, and the floor Ancyloscelis apiformis is also smooth and of the cell in horizontal or nearly horizontal loaflike, but it lacks a foot, its midsection is cells) before bringing in another load. These attached to the floor ofthe cell, and the upper preliminary deposits are not shaped into the surface is shallowly grooved longitudinally. form ofa preliminary loaf, at least in the case Information regarding the food masses of of Tapinotaspis tucumana, Monoeca lanei, the other exomalopsines varies considerably. Exomalopsis globosa, similis, solidaginis, si- The food loaf of Tapinotaspis tucumana is dae, and Ancyloscelis apiformis. When the like that of Exomalopsis solidaginis, in that female has gathered the total quantity offood, it is apparently attached only to the rear of she then molds it into a form that is taxon- the cell. Fresh masses appear to have a rough specific and places the completed form in a surface, possibly the result of mold. How- taxon-characteristic position in the cell. The ever, the food of Tapinotaspis caerulea, as shape of the form, in most cases, is loaflike, shown by Vogel (1974), also has a granular that is, with the long axis that more or less surface, although its shape (fig. 37F), a ball parallels that ofthe cell, with the top flattened resting in a socket ofalmost liquid paste (oil) and with the rear rounded and the front at the rear of a horizontal cell, is quite dif- somewhat truncated. In some species the front ferent from the loaflike mass of Tapinotaspis end is drawn downward into a projecting foot tucumana, which is also in a more or less that normally rests on the floor of the cell horizontal cell. To further confuse the pic- while the rounded rear fits into the lower rear ture, Monoeca lanei has the food mass with end of the cell; hence the loaf, which has a an uneven surface "cemented" almost ver- smooth surface, touches the cell wall only at tically into a somewhat more moist food mix- these two points. This kind of loaf is char- ture at the bottom of the verticle cell. Al- acteristic of Exomalopsis chionura,l conso- though difficult to interpret on the basis of brina (fig. 37B) and sidae (fig. 28). Exoma- limited information, these food masses may lopsis solidaginis (fig. 20) has a similar food actually have some important similarities. mass, but the foot is short and does not touch The oily "socket" of Tapinotaspis caerulea the floor, and thus the loafis suspended only may correspond to the soft cement of the by its attachment to the rear of the cell. Al- bottom of the cell of Monoeca lanei.'6 Un- though the presence ofa foot in E. solani was even, granular, and convoluted surfaces may uncertain, the loaf is otherwise like these represent different degrees of a common tex- species. The descriptions of the loaves of E. ture in these three taxa. Spherical shape of globosa and similis by Raw (1977) seem to the main food mass of Tapinotaspis caerulea suggest a shape similar to those of the other may be a rather extreme form of the vertical Exomalopsis, but the smooth food mass of loafofMonoeca lanei, and the elongate rough E. aureopilosa seems quite different accord- loaf of Tapinotaspis tucumana in a horizon- ing to the diagrams of Zucchi (1973). Al- tal cell may be similar to the suspended loaf ofsuch bees as Exomalopsis solidaginis. The l Stephen, Bohart, and Torchio, (1969, fig. 293) in- correctly oriented the food mass in the cell ofthe species. 16 Robert Brooks and Charles D. Michener, both of The loaf is positioned in this species as diagrammed for whom read the manuscript, informed me that they be- Exomalopsis consobrina (Rozen, 1977b). lieve Monoeca to be an oil-collector. 30 AMERICAN MUSEUM NOVITATES NO. 2798

topography and orientation of these masses are diagrammatically presented in figure 37. The orientation and size of the cells have A been modified to aid comparison. There is no way to determine at this time if there is any phylogenetic significance to the similar- ities. DEVELOPMENT: The elongate white, curved exomalopsine eggs are deposited on top of the provisions, somewhat toward the front end of the loaflike food masses, in all cases. Even the egg of Monoeca lanei (fig. 1 1), in a B vertical cell, is somewhat in that position. Larvae are elongate and, so far as known, all later instars crawl while feeding (Tapinotas- a pis tucumana, Monoeca lanei, Exomalopsis globosa, similis, solidaginis, chionura, sidae, Eremapis parvula, and Ancyloscelis apifor- mis). Partly grown larvae of some taxa circle the partly consumed provisions so that the mass no longer contacts the cell wall, sugges- tive of the feeding habits of Protopeolus sin- gularis (Rozen, Eickwort, and Eickwort, 1978). This is true for Tapinotaspis tucu- mana and Exomalopsis solidaginis and si- dae, but the food mass of Monoeca lanei ad- heres to the bottom of the vertical cell until consumed. DEFECATION AND COCOON SPINNING: Me- conial material may be voided either at the beginning of the last stadium while the larva is still feeding (Exomalopsis solidaginis, con- sobrina, sidae, Eremapis parvula, Ancylo- scelis apiformis) or when the food is entirely consumed (Tapinotaspis caerulea, Claude- Joseph, 1926; Tapinotaspis tucumana; E. F Where the feces are in FIG. 37. Diagrammatic representation of cells chionura). l7 deposited and food masses of various exomalopsine bees, the cell depends in part on whether or not side view. Diagrams not drawn to same scale and the species spins a cocoon, and in part on the all oriented at 45 degrees to aid in comparison of species-specific pattern demonstrated by the shape and position offood masses. This sequence larva. The following exomalopsines are is intended to show certain similarities when food known to spin cocoons: Tapinotaspis cae- masses are compared sequentially, as follows: rulea (Claude-Joseph, 1926), Exomalopsis A. Ancyloscelis apiformis-smooth food mass resting on bottom of cell. B. Exomalopsis conso- 17 This observation needs to be verified for E. chionura brina (after Rozen, 1977b)-smooth mass, now because of the more recent information regarding other with front foot resting on floor of cell and with members of the genus. rear ofmass at rear bottom ofcell. C. Exomalopsis solidaginis-smooth food mass, still with foot but foot not resting on cell floor, so that entire mass is suspended from cell rear. D. Tapinotaspis tucu- ing from rear of cell, but now embedded in softer mana-food mass now with uneven surface and food material. F. Tapinotaspis caerulea (after Vo- without foot, but suspended from rear of cell. E. gel, 1974), food mass with uneven surface attached Monoeca lanei-food mass still uneven and aris- to socket-like softer food material at rear of cell. 1984 ROZEN: EXOMALOPSINI 31 similis, globosa, apparently compactula, ently Exomalopsis compactula, crenulata, crenulata, consobrina, Eremapis parvula, and and Eremapis parvula. The only record of Ancyloscelis apiformis. The following species feces being on the inside of the cocoon was are known not to spin cocoons: Tapinotaspis by Claude-Joseph (1927) for Tapinotaspis tucumana, Exomalopsis aureopilosa, solani, caerulea, and this would signify spinning first and solidaginis. As reported by Rozen and and then defecation. Some species, however, MacNeill (1957), some individuals of Exo- start defecating first and then commence lay- malopsis chionura from the same nesting site ing down silk at the same time, so that fecal produced cocoons, whereas others did not. pellets are incorporated into the cocoon fab- In Exomalopsis sidae, one generation does ric. This has been recorded for Exomalopsis not spin cocoons and is followed by a late chionura, consobrina, and sidae. summer generation that produces cocoons in The placement of feces in the cell varies to which the postdefecating larvae diapause, a considerable extent, so far as we have clear presumably for winter. A similar situation information. Tapinotaspis tucumana (fig. 8) has been reported for Protepeolus singularis deposits fecal material at the upper rear end (Rozen, Eickwort, and Eickwort, 1978) and of the nearly horizontal cell, a situation may be widespread (although not universal) strongly reminiscent offecal deposition in the in anthophorids that have more than one gen- Halictidae, Andrenidae, as well as other ma- eration a year. For example, among the Ex- jor groups ofbees. Exomalopsis chionura and omalopsini, Tapinotaspis tucumana had well- consobrina apply feces over the entire cell developed salivary lips although it did not wall and closure, more or less uniformly but spin a cocoon but rather pupated after def- not always evenly or smoothly. Exomalopsis ecating as did the earlier generation of E. solidaginis, however, applies most ofthe fecal sidae. The discovery that some Exomalopsis material over the cell wall, with a smaller chionura spin and some do not may actually amount over the cell closure. Exomalopsis be the result of excavating a nesting site of a compactula (fig. 16) seems to place most of multivoltine species just as it was entering its the feces over the cell closure and the front terminal generation for the year. end of the cell, and Ancyloscelis apiformis Exomalopsine larvae that spin cocoons (fig. 35) smears feces over the entire cell wall have their labiomaxillary regions strongly but does not cover the cell closure. Eremapis projecting, with elongate maxillary and labial parvula smears most of the fecal material palps, and with the salivary lips produced. around the center of the cell, more or less Tapinotaspis tucumana and those individu- parallel to the equator of the cell, forming a als of Exomalopsis chionura that do not spin distinct ring. Exomalopsis solani and cren- cocoons also have these modifications. The ulata apply the feces over the cell wall, but salivary lips ofExomalopsis solani are slight- information is insufficient to determine ly produced, even though the species did not whether this also includes the cell cap. spin a cocoon. The last instar of Monoeca Exomalopsis sidae has two patterns offecal lanei has well-developed lips, although its co- deposition, depending on whether the larva coon-spinning habits are unknown. The last does not spin a cocoon but pupates imme- stage larvae ofExomalopsis solidaginis, which diately or spins a cocoon and undergoes dia- do not spin, have recessed labiomaxillary re- pause. In the former case the feces are pri- gions. So far as I now know, larvae ofa single marily at the bottom (rear) of the cell; in the species are not dimorphic, i.e., the generation latter, they cover the cell wall and become a that does not spin has a produced labio- part ofthe cocoon structure at the cell closure. maxillary region even though the other gen- Such differences may be expected with other eration spins. species if they have similar cocoon-spinning With forms that spin cocoons, onset ofdef- habits. ecation precedes the first production of silk CUCKOO BEE ASSOCIATIONS: Table 2 lists in most species. When defecation is com- exomalopsine species treated in this paper pleted and before spinning begins, all feces together with the cuckoo bees that have been are situated between the cell surface and the associated with them. These associations are cocoon fabric: Ancyloscelis apiformis, appar- made with various degrees of confidence, as 32 AMERICAN MUSEUM NOVITATES NO. 2798

C.) C.) -4C.) 0 .0 .0 C) °.° co co C.)0o CU . CU .0 .0 I II 0o Ilo .0 I I 2 -4Ut ,0 a C.) a. . . Cd .-O - ,0 Z ucn U cn U Z 0 -4 )CQ 4.) U C.) U z Q z gr~ 'a*1;t & .CU t., 0~ I - rN 0%~ ~ ~ Az& oCU r- C. (N ON II CN-S. CqO o. .1 o U)A 0% 3 o .CU CU . P - u P CL C .) 0 *a 0 co 0 I,~. -.) C.) CU I.) N a)- * C.) ' 0 *-4)U) u ItY 3 0..:~0&g8 I I 3 0 3 0 H - . .o U)).) 0 0 C.)t oa

*C.) CUC v)- cn) 0 C.) * * .), mC.) C.) 0).) cn C.) 1-0 0 CUX '0 U U) _ C.) C.) C. C) 0CU - .0(A UC 00Q CU (A C.)U) I-- C.) *.) 0 0.0Q .t QCU e C,) *20 CO 0 U) UC.) u .0 R 00. E. 0 U)..., k .x . U) U) 0 C,.) 000c 0 *z~.U) .w U) o~~0 0 U-) 0; r'-. CU 0 In 031 -xv .D -E ,, U) C.) 0,o000: .C C.) C 0t$$ U)UQ.)1: U) .,o o o E II 08 U)U 0- U)U)U) -) U)U)U) Uc) LC QI 0.c 0.0q.0-.C 0. 0O OC.) .I 0. 0 00 03 p.4QXv D 1984 ROZEN: EXOMALOPSINI 33 indicated. The four separate genera of clep- subfamily Nomadinae have been another fo- toparasitic bees belonging to the Nomadini cus of the study. Are there features of exo- have definitely been linked to North Amer- malopsine behavior that might be precursors ican species of Exomalopsis alone: Hypo- to cleptoparasitism and thus indicate that chrotaenia, Paranomada, Triopasites, and these two taxa are sister groups? Several such Melanomada. A fifth genus, Hesperonomada, traits are suggested. (1) Communal nesting, is probably also associated with Exomalop- a common though not universal feature of sis. I know of no other genus of host bee that the Exomalopsini, may constitute the first has such a diversity of cleptoparasites.18 evolutionary step toward nest parasitism. Fe- In conclusion, the Exomalopsini are a di- males using a single nest may lead to some verse group with respect to nesting biology. of those females abandoning nest construc- Not unexpectedly, there are biological traits tion and foraging. Instead they oviposit in that reflect the relationships of many of the cells constructed and provisioned by their sis- taxa with one another, as for example the ters. Communal nesting behavior is an un- unity of Exomalopsis. However, available common phenomenon in other nonparasitic data are yet too scarce to be used as a major Anthophorinae; only certain Eucerini are source of information for cladistic analysis. communal (Michener, 1 974b).19 (2) All larval There is such diversity and incongruity of exomalopsines crawl while feeding, as do the characteristics that I am left with the impres- larvae ofthe Nomadinae. Such behavior sug- sions that (1) the relationships oftaxa within gests a common ancestor, but ofcourse most the tribe need further systematic evaluation ifnot all other anthophorid larvae also crawl. and (2) the relationships ofthe Exomalopsini (3) The fact that all five genera of the tribe with the Melitomini, Ancylini, and, indeed, Nomadini attack nests of various species of all the other so-called primitive, nonparasitic Exomalopsis is unusual, for I know of no anthophorids require further investigation. other host genus of bees that has so many The biological diversity offers promise that groups of cleptoparasites. Perhaps, in some further accumulation ofthe kind ofdata pre- way that is not altogether clear, these asso- sented here will be instrumental in resolving ciations may be the vestiges ofthe early evo- such relationships. Investigators should con- lution ofthe parasitic subfamily and they re- tinue to excavate nests and make careful notes flect the idea that cleptoparasites tend to be and detailed diagrams so that one taxon can related to their hosts (Wheeler, 1919; Bohart, be compared with the next. They are en- 1970). Although these traits might possibly couraged to use stereoscopic microscopes in indicate a close relationship between the Ex- the field on such subjects as cell shape and omalopsini and Nomadinae, they cannot be construction, cell lining, cell closure, and lar- regarded as strong evidence ofa relationship. val behavior. Saving cell samples, burrows, Other behavior traits of the Exomalopsini and turrets, as well as the immatures has been seem distant from those of the Nomadinae. an invaluable source of information for this Exomalopsines and other nonparasitic an- paper, and investigators are encouraged to thophorines place their eggs on the provi- label well and preserve such samples in mu- sions; in contrast, virtually all nomadines in- seum collections. The relationships ofthe Ex- sert their eggs in the cell wall, presumably omalopsini with the diverse cleptoparasitic hiding them from returning host females. First instars of the Exomalopsini (at least Exo- 18 Both Charles D. Michener and Robert W. Brooks, upon reading this manuscript, raised questions regarding 19 Anthophora marginata Smith is also communal in the distinctiveness of some of these genera. Jalisco, Mexico, according to Robert W. Brooks (in litt.). 34 AMERICAN MUSEUM NOVITATES NO. 2798 malopsis solidaginis) remain enclosed mostly temperate arid to moist tropical situations. within the chorion, and only the second in- Body-size moderate to small, often robust. stars crawl away from the chorion as they Individuals of some species solitary nesting, feed on the stored provisions. The crawling of others nesting in loose or tight aggrega- mechanism of the early instars of exomalop- tions. Although some species clearly non- sines is probably associated with the some- communal, others with numerous females to what projecting abdominal sternum IX. In a nest; all Exomalopsis communal and quasi- the Nomadinae the so-called first instar is social or possibly semisocial; nests of pre- extremely active, being able to crawl down vious generation often reused in communal the cell wall from the oviposition site, locate forms. Nest surfaces horizontal to vertical, and crawl to the host egg or larva, and kill it almost always in ground except for Paratetra- with elongate tapering mandibles. The crawl- pedia gigantea. Nests moderately to very ing mechanism usually, if not invariably, in- deep; those of communal species apparently volves a forked pygopod-like structure aris- without either turrets or tumuli; Monoeca la- ing laterally from abdominal segment X. nei with distinct turret and Eremapis parvula However, in the Nomadinae, studies have and Tapinotaspis tucumana with tumuli. not yet been made as to the number of larval Main burrows circular in cross section; bur- instars, and there are some suggestions that row walls in many Exomalopsis tamped by the cast skin of an earlier larval instar may pygidial plate, at times masoned, normally be left as the young larva crawls from the water absorbent, but in the case of Monoeca hidden chorion in the cell wall. More detailed lanei, lined with waxy, apparently secreted studies of larval behavior and development material that retards water absorption; main within the Nomadinae may be helpful in burrows open during nest construction and shedding light on the relationships of the provisioning; laterals leading to completed parasitic group with the Exomalopsini. In cells filled or partly filled with soil. Nest pat- summation, behavioral evidence suggesting tern consisting of main tunnel, which may that the Exomalopsini and the Nomadinae (Exomalopsis) or may not (Tapinotaspis tu- are sister groups is limited. cumana) branch many times, and of laterals extending variable distances to cells; these laterals radiating from main burrow in all PROFILE OF THE BIOLOGY OF directions; laterals not ending blindly. Nests THE EXOMALOPSINI normally with numerous cells, but Ancylo- In several recent papers, I have provided scelis panamensis apparently with only single in brief, telegraphic style the biological char- cell. Cells ofmost species arranged singly, but acterization of a particular higher taxon of linear series present in some cases. Cells of bees. Although such synopses repeat much of some species with uniformly horizontal to the information that is discussed within the vertical orientation, of others varying con- body of these papers, they also present the siderably in inclination; cells not symmetri- biological gestalt ofthe taxon involved in brief cal around straight longitudinal axis, usually form. In the profiles ofthe Fideliinae (Rozen, with lower surface flatter than upper, as seen 1977a) and Diphaglossinae (Rozen, 1984), in longitudinal section; cell walls usually such presentations seem reasonably success- smooth but in some cases bearing imprints ful, probably because of the distinctiveness of pygidial plates, sometimes masoned; cell and monophyly of the two taxa. The Exo- lining normally a thin film of transparent or malopsini seem more diverse and their re- semitransparent, smooth, shiny material, in lationships to other so-called primitive an- almost all cases waterproof; this material nor- thophorines, less clearly defined. Hence, the mally nonwaxlike, in that it does not melt at following lacks some ofthe unity ofthe other high temperatures, but in Monoeca lanei and two synopses, although within the Exoma- Tapinotaspis caerulea it melts at low tem- lopsini the genus Exomalopsis is very dis- peratures; material apparently secreted by tinctive. adults rather than transported from outside NESTING: New World species inhabiting of nest. Cell closures invariably spiraled on 1 984 ROZEN: EXOMALOPSINI 35 the inside; this surface concave, sometimes completely diurnal, no nocturnal or crepus- only shallowly concave, and sometimes es- cular forms reported. Mating not studied, but sentially flat; surface normally water-absorb- in some cases known to take place at pollen ent, but in several species water-retardant, plants. although not as a result of a shiny or waxy CLEPTOPARASITIC BEES: Exomalopsis film over the spiral; outer surfaces of closure species parasitized by cuckoo bees belonging in many species a strongly concave, smooth, to the Nomadini (Hypochrotaenia, Parano- waterproof surface that does not grade into mada, Triopasites, Melanomada, and Hes- the fill in the connecting lateral; with some peronomada). Cuckoo bees of other genera species (Tapinotaspis tucumana and Moneo- not reported, except for unnamed species be- ca lanei) closure without concave outer sur- longing to the subfamily Nomadinae, tenta- face distinct from fill. tively associated with Tapinotaspis tucu- PROVISIONING: Foraging females not shap- mana. ing deposits of pollen before bringing in en- tire food supply; stored provisions shaped into LITERATURE CITED loaflike forms, often with distinct "foot"; Batra, S. W. T. shape ofthese loaves somewhat variable and 1972. Some properties ofthe nest-building se- in some species attached to rear (bottom) of cretions of Nomia, Anthophora, Hy- cell in various ways; provisions never liquid laeus and other bees. Jour. Kansas Ent. or semiliquid. Pollen sources unstudied; pol- Soc., vol. 45, pp. 208-218. len transported on hind legs only; some Bohart, G. E. species seeming to be polylectic, others oli- 1970. The evolution ofparasitism among bees. golectic; source ofliquid in provisions known Forty-first Honor Lecture, the Faculty to be oil rather than nectar in some taxa; oil- Association, Utah State Univ., pp. 1- collecting habits need further investigation. 33. Modification of provisions during larval de- Brethes, Juan 1909. Una anthophorina parasita? Am. Mus. velopment through action of yeasts or bac- Nac. Buenos Aires, series 3, vol. 12, pp. teria not reported and not observed. 81-83. DEVELOPMENT: Moderately elongate, Claude-Joseph, F. curved egg, one to a cell, normally deposited 1926. Recherches biologiques sur les Hyme- on top of provisions, toward front end of nopteres du Chili (Melliferes). Ann. des loaflike food mass. Young larvae feeding while Sci. Nat., Zool., series 10, vol. 9, pp. ambulating over surface of food, sometimes 113-268. causing grooves in food masses; older larvae Cane, James H. often provisions so that food masses 1983. Preliminary chemosystematics of the circling Andrenidae and exocrine lipid evolu- no longer contact cell walls. Defecation com- tion of the short-tongued bees (Hyme- mencing either at the beginning of last stadi- noptera: Apoidea). Jour. Systematic um while larva still feeding or just after food Zool., vol. 32, pp. 417-430, 2 figs. entirely consumed; placement of feces in cell Hicks, C. H. quite variable, often species-specific, and 1936. Nesting habits of certain western bees. variable within a species, depending on Canadian Ent., vol. 68(3), pp. 47-52. whether or not larva spins cocoon; feces a Houston, T. F. moist solid, exuded as elongate pellets or 1975. Nests, behaviour and larvae of the bee smears. Some species spinning cocoons, oth- Stenotritus pubescens (Smith) and be- ers not, and in some, overwintering genera- haviour of some related species (Hy- summer menoptera: Apoidea: Stenotritinae). tions spinning cocoon, whereas gen- Jour. Australian Ent. Soc., vol. 14, pp. eration without cocoon; cocoon structure 145-154. quite variable, depending on species; cocoon Hurd, Paul D., Jr., and E. Gorton Linsley fabrics sometimes incorporating feces, other 1975. The principal Larrea bees ofthe South- times not; cocoon shape determined by cell western United States (Hymenoptera: shape and without nipple. Apoidea). Smithsonian Contrib. to ADULT AcriviTY: Bees predominantly ifnot Zool., no. 193, pp. 1-74. 36 AMERICAN MUSEUM NOVITATES NO. 2798

Janvier, H. rid bee with a metasomal scopa (Hy- 1955. Le nid et la nidification chez quelques menoptera: Anthophoridae). Pan-Pacific abeilles des Andes tropicales. Ann. des Ent. Sci. Nat., Zool., ser. 11, vol. 17, pp. 311- Neff, John L., and Beryl B. Simpson 349, 22 figs. 1981. Oil-collecting structures in the Antho- Linsley, E. G. phoridae (Hymenoptera): Morphology, 1945. A new species ofParanomada with notes function, and use in systematics. Jour. on Melecta thoracica Cresson. Ent. Kansas Ent. Soc., vol. 54, pp. 95-123. News, vol. 56, pp. 149-153. Oliveira, Bernadete Lucas de Linsley, E. G., J. W. MacSwain, and C. D. Mich- 1962. Morfologia externa de larva de Para- ener tetrapedia (Paratetrapedia) gigantea 1980. Nesting biology and associates of Me- (Schrottky, 1909) a considerag6es filo- litoma. Univ. California Pubs. Ent., vol. gen6ticas s6bre algumas larvas de An- 90, pp. 1-45, 8 figs. thophoridae (Hymenoptera-Apoidea). Linsley, E. G., J. W. MacSwain, and Ray F. Smith Bol. da Univ. do Parana, Zool., no. 12, 1954. A note on the nesting habits of Exo- pp. 1-21, 3 figs. malopsis solani Cockerell. Pan-Pacific 1966. Descricao de estadios imaturos de Lan- Ent., vol. 30, pp. 263-264. thanomelissa sp. (Hymenoptera: Apoi- Malyshev, S. dea). Studia Ent., vol. 9, fasc. 1-4, pp. 1935. The nesting habits of solitary bees. A 429-440, 5 figs. comparative study. Eos, Madrid, vol. Parker, F. D. 11, pp. 201-309. [In press] Biological notes on the bee Exoma- Michener, Charles D. lopsis crenulata Timberlake (Hyme- 1944. Comparative external morphology, noptera: Anthophoridae). Pan-Pacific phylogeny, and a classification of the Ent., ms. pp. 1-8. bees (Hymenoptera). Bull. Amer. Mus. Raw, A. Nat. Hist., vol. 82, pp. 151-326. 1976. Seasonal changes in and foraging activ- 1954. Bees of Panama. Ibid., vol. 104, pp. 1- ities of two Jamaican Exomalopsis 176, 155 figs., 3 tables. species (Hymenoptera: Anthophori- 1966. Evidence ofcooperative provisioning of dae). Biotropica, vol. 8, pp. 270-277. cells in Exomalopsis (Hymenoptera: 1977. The biology oftwo Exomalopsis species Anthophoridae). Jour. Kansas Ent. Soc., (Hymenoptera: Anthophoridae) with vol. 39(2), pp. 315-317. remarks on sociality in bees. Revista 1974a. Further notes on nests of Ancyloscelis Biologia Tropical, vol. 25, pp. 1-11. (Hymenoptera: Anthophoridae). Ibid., Rozen, Jerome G., Jr. vol. 47, pp. 19-22. 1977a. The ethology and systematic relation- 1974b. The social behavior ofthe bees. A com- ships of fideliine bees, including a de- parative study. Cambridge, Harvard scription of the mature larva of Para- Univ. Press, pp. 1-404. fidelia (Hymenoptera, Apoidea). Amer. Michener, Charles D., and Rudolf B. Lange Mus. Novitates, no. 2637, pp. 1-15. 1958. Observations on the ethology of neo- 1977b. Immature stages of and ethological ob- tropical Anthophorine bees (Hymenop- servations on the cleptoparasitic bee tera: Apoidea). Univ. Kansas Sci. Bull., tribe Nomadini (Apoidea, Anthopho- vol. 39, pp. 69-96. ridae). Ibid., no. 2638, pp. 1-16. Michener, Charles D., Rudolf B. Lange, Joao J. 1978. The relationships of the bee subfamily Bigarella, and Riad Salamuni Ctenoplectrinae as revealed by its bi- 1958. Factors influencing the distribution of ology and mature larva (Apoidea: Me- bees' nests in earth banks. Ecology, vol. littidae). Jour. Kansas Ent. Soc., vol. 51, 39, pp. 207-217. pp. 637-652. Michener, Charles D., and Jesus S. Moure, C.M.F. 1984. Nesting biology of diphaglossine bees 1957. A study ofthe classification ofthe mole (Hymenoptera, Colletidae). Amer. Mus. primitive non-parasitic anthophorine Novitates, no. 2786, pp. 1-33. bees (Hymenoptera, Apoidea). Bull. Rozen, Jerome G., Jr., and C. Don MacNeill Amer. Mus. Nat. Hist., vol. 112, pp. 1957. Biological observations on Exomalop- 395-452. sis (Anthophorula) chionura Cockerell, Neff, John L. including a comparison of the biology [In press] Observations of the biology of Ere- of Exomalopsis with that of other an- mapis parvula Ogloblin, an anthopho- thophorid groups (Hymenoptera: Apoi- 1984 ROZEN: EXOMALOPSINI 37

dea). Ann. Ent. Soc. Amer., vol. 50, pp. Anthophoridae). Jour. Kansas Ent. Soc., 522-529. vol. 47, pp. 54-63. Rozen, Jerome G., Jr., Kathleen R. Eickwort, and Torchio, Philip F., Jerome G. Rozen, Jr., George George C. Eickwort E. Bohart, and Marjorie Favreau 1978. The bionomics and immature stages of 1967. Biology of Dufourea and of its clepto- the cleptoparasitic bee genus Protepeo- parasite, Neopasites (Hymenoptera: lus (Anthophoridae, Nomadinae). Amer. Apoidea). Jour. New York Ent. Soc., vol. Mus. Novitates, no. 2640, pp. 1-24. 75, pp. 132-146, 9 figs. Rust, R. W., R. W. Thorp, and P. F. Torchio Toro, Haroldo 1974 The ecology of Osmia nigrifrons with-a 1976. TChilimalopsis, nuevo genero Chileno de comparison to other Acanthosmoides. Exomalopsini (Hymenoptera: Apoi- Jour. Nat. Hist., vol. 8, pp. 29-47, 14 dea). An. del Museo de Hist. Nat. de figs. Valparaiso, vol. 9, pp. 73-76. Schrottky, C. Vogel, Stefan 1901. Biologische Notizen solitarer Bienen von 1974. Olblumen und olsammelnde Bienen. S. Paulo (Brasilien). Ailgemeine Zeit. fur Tropische und subtropische Pflanzen- Ent., Band 6, pp. 209-216. welt. Akad. der Wissenschaften und de Stephen, W. P., G. E. Bohart, P. F. Torchio Literatur, Mainz, pp. 1-267. 1969. The biology and external morphology Wheeler, W. M. ofbees. Oregon State Univ. Press, Cor- 1919. The parasitic Aculeata, a study in evo- vallis, pp. 1-140. lution. Proc. Amer. Philos. Soc., vol. 58, Timberlake, Philip H. pp. 1-40. 1980. Review of North American Exoma- Zucchi, Ronaldo lopsis (Hymenoptera, Anthophoridae). 1973. Aspectos bionomicos de Exomalopsis Univ. Calif. Publ. Ent., vol. 86, pp. 1- aureopilosa e Bombus atratus incluindo 158. considera,6es sobre a evolugcio do com- Torchio, Philip F. portamento social (Hymenoptera- 1974. Notes on the biology ofAncyloscelis ar- Apoidea). Doctoral thesis, Faculdade de mata Smith and comparisons with oth- Filosophia, Ciencias e Letras de Ribei- er anthophorine bees (Hymenoptera: rao Preto, pp. 1-172.