Noqvitatesamerican MUSEUM PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

NoqvitatesAMERICAN MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3066, 28 pp., 45 figures June 11, 1993

Nesting Biologies and Immature Stages of the Rophitine Bees (Halictidae) with Notes on the Cleptoparasite Biastes (Anthophoridae) (Hymenoptera: Apoidea)

JEROME G. ROZEN, JR.' CONTENTS

Abstract ...... 2 Introduction ...... 2 Nesting Biology of the Rophitinae ...... 3 Sphecodosoma dicksoni ...... 3 Conanthalictus conanthi ...... 11 Rophites trispinosus ...... 15 Profile of the Biology of the Rophitinae ...... 16 Mature Larvae of the Rophitinae ...... 17 Key to the Mature Larvae ...... 18 Sphecodosoma dicksoni ...... 18 Conanthalictus conanthi ...... 21 Rophites trispinosus ...... 23 Pupa of Sphecodosoma dicksoni ...... 24 Discussion ...... 24 References ...... 26

lCurator, Department of Entomology, American Museum of Natural History.

ABSTRACT

Information on the nesting biology ofthe ground- The mature larvae of the Rophitinae are char- nesting Sphecodosoma dicksoni (Timberlake) and acterized on the basis of six genera, and a key to Conanthalictus conanthi (Cockerell) from the available species is presented. The mature larvae southwestern United States is added to previously of Sphecodosoma dicksoni and Conanthalictus published data to provide an understanding ofnest conanthi are described taxonomically and com- architecture, cell construction, provisioning, egg pared with larvae ofother Rophitinae. Also treat- deposition, larval feeding behavior, cocoon con- ed is an immature larva (probably last instar) of struction (in the case of S. dicksoni), larval defe- Rophites trispinosus. Whereas the mature larvae cation, and voltinism of these species. No clep- of S. dicksoni and R. trispinosus share many fea- toparasitic bees are associated with either species tures with rophitine genera Dufourea and Xeral- at present. ictus, that of C. conanthi is very different, though Observations on nest provisioning and larval clearly sharing significant synapomorphies with the adaptations ofthe related Palearctic Rophites tris- others. Many ofits differences appear to be related pinosus Perez are included. Recovery of an inter- to the fact that it does not spin a cocoon. The pupa mediate-stage larva of Biastes emarginatus ofS. dicksoni is also described, the first such treat- (Schenck) (Nomadinae: Biastini) from the nest es- ment for any member of the subfamily. tablishes this host association of the cleptoparas- New information on nesting biology and im- ite. Its larva and the mature larva of the related matures of the rophitines, though supporting the genus Neopasites are compared and are found to monophyly of the subfamily, does not seem to share many derived features. Based on informa- demonstrate phylogenetic linkages with the Hal- tion presented here and on published and unpub- ictinae and Nomiinae or with the Andrenidae and lished accounts, a synopsis of the biology of the Melittidae at this time. However, a number of Rophitinae is presented based on 7 genera and 14 characters are identified and discussed that may species. The synopsis identifies features that seem eventually be helpful in resolving these relation- to be characteristic of the subfamily. ships.

INTRODUCTION This paper is an attempt to increase our permits a formal (but tentative) overview of understanding ofthe nesting biology and im- the biology ofthe subfamily for the first time. mature stages ofthe Rophitinae (= Dufourei- Larvae ofonly a few taxa ofrophitines have nae), one of the three generally recognized been described and the account of the pupa halictid subfamilies. Such information will of Sphecodosoma dicksoni below is the first eventually contribute to an understanding of treatment of any pupa of the subfamily. To the relationships ofthe rophitines to the other date mature larvae ofonly Dufourea (Torchio subfamilies of Halictidae and to other taxa. et al., 1967 [incomplete]; McGinley, 1981, The research started as a reinvestigation of 1987; and Eickwort et al., 1986) and Rho- the nesting behavior and ecology of Sphe- phitoides (Enslin, 1921) ofthe approximately codosoma dicksoni (Timberlake) and Con- 12 or 13 recognized genera have been treated. anthalictus conanthi (Cockerell). Nests ofboth In addition, Batra and Michener (1966) treat- species had been briefly studied earlier (Roz- ed the immature larva of Systropha. en and McGinley, 1976), but so much new The manuscript for the present paper was information resulted from the current re- initiated as a study ofthe nesting biology and search that an expanded presentation oftheir immatures of S. dicksoni; data on the other biology seems appropriate. In addition, lar- two taxa were subsequently added and com- vae and pollen-nectar masses ofRophites tris- pared with information on S. dicksoni. This, pinosus Perez preserved in the American Mu- and not presumed relationships, accounts for seum of Natural History (AMNH) allowed the sequence ofpresentation ofthe taxa here. interpretation of some aspects of the biology of this species in light of what is known of the other two species. The new information ACKNOWLEDGMENTS on these species, when combined with ac- I thank both George C. Eickwort and Ron- counts of the biology of other Rophitinae, ald J. McGinley for carefully reviewing this 1993 ROZEN: ROPHITINE BEES 3

manuscript and offering a number ofthought- for Sphecodosoma dicksoni. Precipitation in provoking suggestions. the area comes primarily in winter, resulting Roy R. Snelling kindly provided a copy of in a spring blooming period. During dry years a manuscript prepared by him and G. I. Stage the area is without bloom, but the precipi- on the nest and larva of Xeralictus timber- tation of the 1991-1992 season was copious, lakei. He also graciously donated the single and the flowering abundant. Although the known larval specimen of this species to the sand-dwelling Oenothera was past bloom and AMNH, thus permitting me to examine it Larrea tridentata had nearly finished flow- first hand. George C. Eickwort loaned the ering at the April 30-May 1 visit, Nama ap- larva of Dufourea novaeangliae (Robertson) peared to be at maximum flower. The plant and thereby added to the completeness ofthe was still blooming on May 11 although there study. I extend my appreciation to Richard were fewer flowers and a reduced number of H. Kruzansky who analyzed the soil from the bees visiting the plant. In addition to Sphe- nesting area. Noel Holmgren, ofthe New York codosoma, the following bees visited the plant: Botanical Garden, identified the Nama as- Calliopsis (Micronomadopsis) foleyi (Tim- sociated with Conanthalictus conanthi. berlake), Perdita (Pseudomacrotera) turgiceps Andrey Sharkov and Beatrice Brewster as- Timberlake, P. (Heteroperdita) arenaria sisted by translating some of the foreign lit- Timberlake,2 and P. (Perdita)flavicauda for- erature. mosa Timberlake. Sphecodosoma dicksoni nested in a sparse- ly vegetated area at the base of the east side NESTING BIOLOGY OF THE of the sand dune. The nest entrances were ROPHITINAE interspersed with the low-growing (ca. 15 cm high) Nama plants. Although the patches of Sphecodosoma dicksoni plants were scattered along the east side of The biology of this species was studied by the dunes as well as elsewhere in the region, Rozen and McGinley (1976) (Sphecodosoma nests were found only in two areas. The one then considered a subgenus of Conanthalic- site studied (fig. 1) consisted of a loose ag- tus) at 21 mi south of Animas, Hidalgo Co., gregation of nests, approximately 10 in an New Mexico, in August 1975. Information area of 1.5 x 0.5 m with other nests scattered in the earlier report agrees in most respects farther away. The other aggregation, 30 m with the new data, but differences that do away in an ecologically similar area, sug- occur are discussed within the description. gested that, iftime had permitted, nests might Measurements, counts, and other informa- well have been found at the bases of other tion in brackets are data from the 1976 re- sand dunes where there were sufficient num- port, presented here for comparison. bers of plants to support the foraging needs DESCRIPTION OF SITE: The nesting site of of the females. Sphecodosoma dicksoni was studied at 18 The plants and nests were in low elevations miles west of Blythe, Riverside County, Cal- adjacent to the sand dunes, where the ground ifornia, on April 30 and May 1, 1992, and sloped about 150 [20-30° in the 1976 study was again examined on May 11, 1992. This in New Mexico]. This low slope gave the im- locality has been visited by entomologists pression that the species prefers horizontal from the early 1950s to the present and was nesting surfaces, but observations on the nests referred to as Hopkins Well on some early presented below indicate otherwise. Loose on specimen labels. The dominant feature ofthe the surface, the soil was firmer below, es- site is the loose sand dunes that, though pecially where cells were encountered, but somewhat lower, have remained virtually could easily be excavated with a penknife. stationary for the last 40 years. Although the The surface was generally unshaded during dunes support little vegetation, adjacent ar- the day. The soil, a loamy sand without peb- eas are dominated by Larrea tridentata, other woody plants, and at the time of the current 2 Specimens of this species were found on Nama only study Sphaeralcea and Nama sp. (probably on May 11 by which time the flowers of Coldenia (its hispidum), the latter being the food source normal food source) had dried up. 4 AMERICAN MUSEUM NOVITATES NO. 3066

:. -. :t'-, 1 Fig. 1. Nesting site ofSphecodosoma dicksoni, 18 mi west ofBlythe, Riverside Co., California. Nest entrances most abundant to the right of the umbrella shadow. bles and containing few roots, consisted of and 2 cm wide at maximum width. They were the following: 82% sand, 11% silt, and 7% sufficiently more vivid in color than the sur- clay. Its calcium content was high (200 + ppm) face sand, permitting easy identification of which according to Kruzansky (in litt.) prob- likely nests. ably accounted for its firmness, because alu- Main tunnels (fig. 2), 2.0 mm in diameter, minum (trace) and clay contents were low. turned from side to side as they slowly de- NEST ARCHITECTURE (fig. 2): All nests ob- scended; below 10 cm, though still mean- served here were occupied by single females, dering, their paths tended to become more whereas some at the New Mexico site had vertical. Open above, they were partly filled been occupied by two or three females. The with loose sand in their lower reaches. This open entrances (N = 7) penetrated into the material could be blown away with an aspi- slope ofthe surface at a low angle, somewhat rator because the substrate below 12-14 cm less than 200 from horizontal. A number of was faintly moist and more consolidated. females tried (some successfully) to initiate Laterals, of approximately the same di- tunnels in the vertical face of the excavation ameter as the main tunnel, extended hori- I made to examine other nests. These facts zontally from the main tunnel and generally as well as the low descent rate of the upper ended in linear series ofcells. Lengths ofthese parts of main tunnels indicate that females filled laterals were 10-25 mm (N = 5) [7.0- prefer surfaces that enable them to penetrate 13.5 mm, N = 6]. However, two laterals, both the soil obliquely, as was also suggested at the lowest in their respective nests, ended in the New Mexico site. single cells. These cells were apparently the Tumuli of loose soil accumulated on the first ofa series yet to be constructed, and their downhill side of entrances. Generally low, laterals were open and 45 and 46 mm long. some tumuli were approximately 3 cm long Completed laterals were soil-filled and could 1993 ROZEN: ROPHITINE BEES 5

9 FECES o 5-

1.0 mm 10 Fig. 2. Nest of Sphecodosoma dicksoni, side view. Figs. 3-6. Nests of Conanthalictus conanthi, side views. All nests drawn to same scale as indicated. Figs. 7-9. Eggs. 7. S. dicksoni. 8. C. conanthi. 9. Rophites trispinosus. Eggs drawn to different scales as indicated. Fig. 10. Cell of C. conanthi side view, showing fecal deposition. Fig. 11. Young larva of S. dicksoni, side view. not be detected except by the fill being some- In sharp contrast to the many cell series what softer than the substrate. found at the California site, cells at the New All cells in linear series (fig. 2) were more Mexico site had all been arranged singly. or less end to end. The only exceptions were Series ranged in depth from 20 to 33 cm single end cells of a new series under con- (N = 9). Completed nests obviously consisted struction. Completed series consisted of 2 to of a number of cell series at various depths. 4 cells (N = 8) with cells of a series being Although no complete nests were excavated, uniformly spaced. Cell partitions were ap- an incomplete one contained cell series at the proximately 1 mm in the middle and 2 mm depths of 24, 26, and 29 cm and another at at the periphery. No intercalary spaces, either 25, 27, and 33 cm. A nearly complete nest filled or empty, were noted between cells in had a series at 16 and 21 and two series at series. 24 cm. Deeper cell series were always younger 6 AMERICAN MUSEUM NOVITATES NO. 3066 than series above them, indicating that nest- structing the cell in front. This was revealed cell arrangement was progressive ramifying- by two nests with single cells at the end of linear sensu Malyshev (1935). Cell depths at long open side tunnels and by an open cell the New Mexico site were much shallower, which contained a small preliminary food 10-14 cm (N = 12). sphere in front of a closed cell containing an Cells were uniform in shape, rounded at egg on the completed provisions. the rear, with a maximum length of 4.0-4.5 PROVISIONING AND DEVELOPMENT: Fe- mm (N = 11) [4.0-5.0 mm, N = 3] and a males carried Nama pollen primarily on their maximum diameter of 3.0-3.1 mm (N = 12) hind tibia and femur, although fully laden [3.0 mm, N = 3] somewhat to the rear of the females had some pollen also on the ventral midpoint ofthe long axis ofthe cell. Although base of the metasoma and the sides of the cells in a series were essentially arranged end propodeum. Transported pollen seemed to end, there was some variation in this ori- sticky but not moist. The small (one being entation. This variation in conjunction with 1.5 mm in diameter) spherical pollen-nectar their small size made it difficult to judge masses found in open cells indicated that ear- whether or not they were symmetrical around ly loads of pollen were shaped into small their long axis, especially because cell series spheres and subsequent loads were added, as were excavated from the top rather than from was the case at the New Mexico site (and is the side. Cells were arranged nearly horizon- the case with the higher panurgines, such as tally with the front end slightly higher than the Calliopsini). Complete mealy-moist food the rear, but at the New Mexico site they masses were completely spherical and 2.15- tilted to the rear at about 300. 2.4 mm in diameter (N = 3) [2.0 and 2.5 mm, Cells lacked walls distinctly harder or dif- N = 2]. These spheres were coated with a ferent in texture from the substrate. They were thin transparent and nonreflective water- coated with an essentially invisible lining that proof coating analogous to the transparent tended to glue together the sand grains on the waterproof shiny food coatings of the Cal- cell surface, making it semiwaterproof. When liopsini (Andrenidae: Panurginae). Such a droplet of water was placed on the surface, coatings were not suspected at the New Mex- it was absorbed slowly over a period of 5 to ico site and therefore were probably over- 10 sec, in contrast to an immediate absorp- looked. Food spheres rested on the lowest tion of a water droplet placed on the surface part of the cell floor and were not attached of a piece of substrate. The cell surface, as at to it by any fluid such as nectar or secretions. the New Mexico site, was nonreflective and One egg was discovered on the top of the not darker than the substrate. The surface food with both of its ends attached to the was, however, more regular than the broken mass, as was reported to be the case at the face of a clod of substrate. There was no in- New Mexico site. Eggs, 1.20-1.30 mm in dication that larvae deposit a clear glistening length and 0.33 mm in maximum diameter secretion over the cell surface as they feed, (N = 3), were white and curved and possessed as is reported for Conanthalictus conanthi be- a clear, smooth, shiny chorion. They tapered low. posteriorly but were otherwise unremarkable The cell closure on the inside consisted of in shape (fig. 7). Two preserved eggs in which a concave spiral of approximately three to embryological body segmentation was al- four distinct coils. The closure material was ready evident revealed the embryo to be ori- loose soil that was not waterproof. For all but ented so that its venter was on the outcurving front cells in series, the outside surface of a (upper) side of the egg. In another egg pre- cell closure (i.e., partition) was also the rear served when further developed, the embryo of the cell in front. Closures of the front cell had rotated 1800 so that its venter was now in a series did not appear to have a special on the incurved side, a late embryogenetic smoothed surface, but such a feature might reorientation typical of many bees (see Tor- have been obscured by the fill of the lateral. chio, 1989, and references therein). At the Females constructing a series obviously con- time of hatching the first instar would have structed the rear cell first, provisioned, ovi- been oriented with its venter on the food sur- posited, and then closed the cell, before con- face. 1 993 ROZEN: ROPHITINE BEES 7

Figs. 12, 13. Cocoons of Sphecodosoma dicksoni, front end of cocoons uppermost. 12. Of single layer of silk, spun by larvae that did not diapause. 13. Of two layers of silk, spun by larvae that later diapaused. Scale units = 1.0 mm.

Small larvae were uncovered in various po- in holding the food mass away from the cell sitions on the food sphere which became more wall while feeding, or both, as is further dis- or less angulate as the larva fed. These facts cussed for Conanthalictus conanthi below. indicate that larvae crawl on the pollen mass Because similar features are found in many as they graze its surface, as is also suggested (if not all) other rophitines, they should be by the larva's elongate form (fig. 11). Similar examined on species with a larger body size conclusions were reached at the New Mexico so they can be observed more easily. Abdom- site. Large larvae were encountered encircling inal segment X was attached dorsally rather the pollen masses as the masses became than medially to IX giving an appearance of somewhat dumbbell shaped. an elevated segment X as seen in lateral view Small larvae and those intermediate in size (fig. 11). The function of the somewhat dor- were elongate, and possessed conspicuous sally directed segment X is discussed under rounded (not transverse) paired dorsolateral C. conanthi. Whereas the paired dorsolateral tubercles on all body segments except the body tubercles, bulging venter of abdominal prothorax and abdominal segments IX and segment IX, and dorsal position of segment X (fig. 11). Middorsal tubercles and subspi- X persist through the last larval instar, the racular tubercles, reported for a partly grown ventrolateral tubercles disappear in the pen- larva of Systropha punjabensis (Batra and ultimate instar. Michener, 1966) were absent. In addition As described in the following section, last Sphecodosoma larvae possessed rounded larval instars possess a strongly projecting la- paired ventrolateral tubercles (presumably the biomaxillary region (fig. 27) (as do the larvae same as the "broad transverse ventral tuber- of Dufourea); this feature is even more pro- cles" of S. punjabensis) that were evident on nounced than in most other cocoon-spinning the mesothorax and metathorax and con- bee taxa. spicuous on abdominal segments I-VIII. The Two types ofcocoons were uncovered. One venter ofabdominal segment IX lacked paired was associated with larvae that molted to the tubercles but was strongly produced ventral- pupal stage without diapausing, the other with ly. The modification ofthe venter ofabdom- larvae that entered diapause and then over- inal segment IX as well as the paired ventro- wintered. Two cocoons (fig. 12) of the first lateral tubercles of the other body segments type from which pink-eyed, active pupae were probably assist the larva either in crawling or removed about a week after being collected 8 AMERICAN MUSEUM NOVITATES NO. 3066

Figs. 14-17. Scanning electron micrographs of cocoons of Sphecodosoma dicksoni. 14. Fabric of single-layered cocoon, inner surface; scale = 20 ,m. 15. Close-up of same; scale = 2 ,um. 16. Fabric of double-layered cocoon, outside surface; scale = 20 ,um. 17. Filter area of double-layered cocoon, cross section, showing multiple silk layers; scale = 200 ,um. on April 30-May 1 were externally (and in- ily detected even with a stereoscopic micro- ternally) pale tan, almost white. Similar va- scope except for fine fibers along torn edges. cated cocoons were uncovered on May 11, However, when examined under a com- after pupae had matured and adults emerged. pound microscope, fine silk strands were quite The fabric of cocoons of this type was very obvious. thin (roughly 0.01 mm thick, exclusive of As revealed on scanning micrographs (figs. feces), single layered, nearly white, and sem- 14, 15), the fabric was composed of fine silk iopaque where there was no fecal material. strands. Some strands were broadly fused to The cocoon (fig. 12) lost its shape when re- form broad sheetlike areas which were seem- moved from the substrate because the fabric ingly fibrous but actually had scattered small lacked rigidity. Silk strands could not be eas- holes. Other strands fused only where they 1 993 ROZEN: ROPHITINE BEES 9

='7 f 7I 7;1W', ,,!

Figs. 18, 19. Scanning electron micrographs of cocoons of Sphecodosoma dicksoni. 18. Inner surface offilter area showing nature ofinner layer ofsilk; scale = 200 ,um. 19. Close-up ofsame, showing opening in cellophanelike material;~~~~~~~tscale = 20 ,m. crossed one another so that the fabric there (similar in structure to that of Dufourea no- was more netlike, with many fenestrations. vaeangliae; Eickwort et al. 1986) encountered The fabric was pliable and soft, and scattered was a darker tan, more opaque, leathery, and sand grains adhered to the outside surface. (except for the front end) roughly 0.02 mm Fecal material (vacuolated, flattened pol- thick along the cut edge. On excavation, this len exines) as elongate flattened pellets or type tended to retain its shape because of its smears radiated from the rear point of the rigidity. It consisted ofan outer tan mat layer cocoon and covered only the rear two-thirds. which contained numerous strands of mod- No layer of silk covered the feces inside the erately fine silk (fig. 16), some of which were cocoon, so that larvae and pupae were in di- darker than others, visible with a stereoscopic rect contact with the dried fecal material. microscope. This layer was spun first and was Where there was no fecal material, the inside followed by the deposition of a layer of feces surface ofthe cocoon was smooth and slightly over the entire surface with the apparent ex- shiny, in general tissue-paperlike. Cocoons of ception of a small area at the front of some the first type, like cocoons described below, cocoons. Feces were appressed to the silk as completely filled the cell lumen from the clo- longitudinal ribbons radiating from the front sure to the rear and possessed no outward and rear poles of the cocoon. Silk strands protuberances (i.e., nipples) at either end, al- could not be detected running through the though the front end that fitted the spiral clo- feces, suggesting that cocoon spinning was sure conformed externally to the shape ofthe not carried out during defecation. This mat- closure. No special gas exchange area was ter, however, needs to be confirmed by ob- detected at the front end of this type of co- serving larvae in the process of defecating. coon. After feces deposition, the larva laid down a The second type of cocoon (figs. 13, 20)3 thin glistening layer of silk, fibers of which

3The type of Sphecodosoma cocoon described here closely parallels (excluding size) cocoons of Dufourea ner so that at first it did not appear to have a specialized mulleri which were examined for comparative purposes filter area. SEM examination, however, revealed a small during this study. The main difference seemed to be that multilayered filter, thinner and more restricted than, but the front (closure) end ofthe Dufourea cocoon was thin- otherwise similar to, that of Sphecodosoma. 10 AMERICAN MUSEUM NOVITATES NO. 3066

1.0 mm 1.0 mm 1.0 mm

'WHITE MECONIAL MASS Fig. 20. Sphecodosoma dicksoni, diagram ofdiapausing larva in cocoon, side view. Fig. 21. Rophites trispinosus, young intermediate stage larva encircling provision, side view. Fig. 22. R. trispinosus, abdominal segment II of older larva, lateral view. Scales as indicated. were apparently sufficiently moist to fuse with flow of gas molecules while the cellophane- one another, although the fibrous nature of like baffles may be a filter, excluding para- the silk was still evident. This layer was thin, sites. faintly tan and cellophane-like, so that the Two complete or nearly complete cocoons longitudinal fecal streaks were visible (more and one incomplete cocoon of this sort were so on some cocoons than on others) through encountered on April 30-May 1. The incom- it. SEM examination (figs. 17-19) revealed plete one contained a larva that had not yet that this material contains relatively few started to defecate, but the dark outer cov- openings. ering containing visible strands of dark tan At the front end of the cocoon, the fabric silk identified it as the outer layer ofan over- (figs. 17-20) covering the cell closure con- wintering cocoon. One larva (fig. 20), kept sisted of a number of loosely appressed, but alive in its completed cocoon, became qui- not fused, layers of nearly opaque, dense, escent and assumed a looped shape described parchmentlike silk that, combined, were below. The larva from the incomplete cocoon thicker (approximately 0.13 mm or perhaps was removed from it before defecating. After more) than the fabric ofthe rest ofthe cocoon defecating it spun the glistening fused silk of because ofthe air spaces between layers. The the inner cocoon layer in a rearing dish in a outermost layer was composed of fine silk disorganized fashion, and then it too as- with many fenestrations, similar to the co- sumed the looped posture described below. coon fabric of the cocoon described above. Five completed cocoons of the second type Inner sheets appeared to be made from thick- containing looped, diapausing larvae were er strands that were fused in many places. On gathered on May 11. some cocoons the central part of this fabric In each cocoon containing a postdefecating seemed to lack feces and in other cocoons the larva, an irregular, lumpy, opaque, whitish quantity of feces appeared to be reduced but meconial mass was plastered to the inside of present. The innermost layer of silk (figs. 18, the rear of the cocoon. It contrasted so strik- 19) was cellophanelike, contained few fen- ingly in color and contour with the rest ofthe estrations, and seemed identical to the rest inner cocoon surface that at first it seemed of the inner surface of the cocoon. Presum- to have been deposited after the last layer of ably the multilayered front of the cocoon silk. In actuality, glistening, nearly transpar- served for gas exchange through the loose soil ent silk covered much if not all of it. This ofthe cell closure. The fenestrations in all the indicates that a larva finishes defecating the layers of fabric apparently permit adequate vacuolated pollen exines and then, while it 1993 ROZEN: ROPHITINE BEES I1I is still depositing the inner silk, discharges and orientation between the present study and the white meconial mass. the one based on the New Mexico site that Although only three cocoons were found cannot be satisfactorily explained. Nor can in the earlier study in New Mexico, they too the difference in the number of females per represented the two types. As in the present nest at the two sites be accounted for. Perhaps case the three-layered cocoon held a post- examination of sites in other areas, especially defecating larva, and one ofthe thin cocoons those areas geographically intermediate be- held a larva that soon pupated. These co- tween the two, might give insight into these coons, preserved in the AMNH, are identical matters by revealing intermediate condi- to the ones from the current study. tions. A less likely explanation might be in The looped posture (figs. 20, 25) of qui- the different ecological conditions of the two escent, overwintering larvae was consistent areas (e.g., different substrates, different sea- in all cases: the abdomen curved under the sonal rainfall patterns) or for that matter in front part of the body so that the venter of the different study seasons (i.e., spring versus the terminal two abdominal segments was late summer). appressed to the thoracic venter, and the apex ofabdominal segment Xjust reached the base of the labiomaxillary region. Although this Conanthalictus conanthi posture was not observed while the cocoons A nest with a single, partly provisioned cell were in place in the ground, larvae apparently of this species was described by Rozen and rest on their dorsa during diapause (fig. 20). McGinley (1976). Data in brackets below re- This is suggested by the fact that, in several fer to the earlier study. cases, a live quiescent larva had the dorso- DESCRIPTION OF SITE: The following inves- lateral body tubercles of the mesothorax and tigation was carried out 4 mi east ofWillcox, first two abdominal segments compressed Cochise County, Arizona, where this species presumably by the weight ofthe larva resting has been known to occur for many years. The on its dorsum. The posture ofthe hibernating nesting site (fig. 23) was discovered and stud- larva of Dufourea novaeangliae is different ied on May 17, 1992, immediately to the west (Eickwort et al., 1986: fig. 4). The looped pos- of an unpaved north-south road. Approxi- ture characteristic ofSphecodosoma dicksoni mately ten nests were scattered over the hor- is lost when diapausing larvae are preserved izontal ground among clumps of short grass in Kahle's solution. and the low-growing pollen plant, Nama his- ADULT ACTIvITY: The information regard- pidum var. spathulatum. The study was con- ing two types of cocoon associated with vol- tinued between August 16, 1992, and Sep- tinism suggests that the population is partly tember 7, 1992, when the area was again univoltine and partly bivoltine (i.e., parsi- visited. At this time, most of the nests, ap- voltine) as was also hypothesized in the New proximately 15 scattered over an area 20 x Mexico study. 10 m, were immediately to the east of the Adults were active during the heat of the same unpaved roadway (fig. 24) and the pol- day, and the numerous males flying about the len plant was now most abundant on the east Nama plants indicate that mating normally shoulder of that road. Although a few of the occurred there. No females attempting to find pollen plants were in bloom on August 16, nests were approached by males. Males and they appeared to be at maximum bloom at females were not seen to fly in copulo. the end of this second study period. During PARASITISM: No cuckoo bees were associ- the interval between the May observations ated with the nests of this species. and those in the late summer, the Nama plants DISCUSSION OF BIOLOGICAL FEATURES: The had obviously ceased blooming because of overall agreement between the current study the dry weather during June and early July. and the one carried out in New Mexico in The initiation oflate summer blossoming was 1975 is gratifying as are the similarities in the caused by the rains of late July and early conclusions independently based in each case August, a common weather pattern for south- on somewhat meager data. ern Arizona. Nonetheless, there are differences in cell Separated only by the width of the road- arrangement (i.e., single or in series), depth, way, the spring and late summer sites were 12 AMERICAN MUSEUM NOVITATES NO. 3066 1993 ROZEN: ROPHITINE BEES 13 virtually identical, both on horizontal ground their long axes, cells were elongate ovals with with an uneven surface dominated by clumps the rear broadly rounded and the front end of low-growing grass between which were more narrowly so. Their maximum diameter lower barren areas. Cattle trampled the sites (2.4-2.6 mm; N = 9) [2.2 mm] was wide from time to time, and their hoofprints cre- relative to the length of the long axis of the ated further unevenness in the surface and cell (3.2-3.8 mm; N = 6) [3.75 mm]. Their not uncommonly obliterated nest entrances. entrance was slightly more constricted than Most nests were found in the barren areas. the diameter of the lateral. The soil was sandy loam with the following The nature and deposition ofthe cell lining composition: 76% sand, 13% silt, and 11% in this species appears to be atypical of most clay. Calcium and aluminum contents were ground-nesting bees and is not fully under- low. During both periods of observation, stood because of too few cells and small cell moisture content at the cell level was high. size. Nonreflective (dull) linings on freshly NEST ARCHITECTURE (figs. 3-6): Active constructed cells immediately absorbed wa- nests could be identified by the tumuli ofvery ter droplets, indicating that the lining in early fine dry soil at their entrances. The tumuli stages at least was nonwaterproof. Several cells 1-2 cm in diameter were concentric, eccen- containing feeding larvae had glistening lin- tric, and acentric depending upon the slope, ings, as if they had been coated with a trans- but all entrance tunnels entered the surface parent plastic glue that caused the sand grains at an oblique angle of 10 to 200 from hori- of the wall to adhere to one another. The zontal. Tunnels normally were open or oc- surface was not smooth as is normally the casionally briefly blocked by tumuli. Several case with cell linings; the sand grains contin- tunnels as yet without cells were partly soil- ued to impart a rough texture to the surface. filled at the lower levels. Main tunnels, un- One cell containing a predefecating larva had lined and uniform in diameter throughout, the entire inner surface of the cell including were approximately 1.5 mm (N = 8) in di- the cell closure covered by this glistening, ameter [1.5 mm in 1976 nest] and lacked gluelike material. The cell containing a post- vestibules. They gradually curved away from defecating larva was similar but the feces at the surface over a horizontal distance of 3-4 the rear of the cell had been discharged over cm and thereafter generally descended ver- the glistening material. When tested with a tically (figs. 3-6). Laterals (about the same water droplet, this lining seemed to retard diameter as, and similar in appearance to, absorption but was not completely water- main tunnels) branched from main tunnels proof. These facts suggest that the feeding between the depths of 15 and 26 cm (N = larva may deposit much ofthe glistening ma- 19) [16 cm] and extended 6-17 mm [7 mm] terial as it feeds. Although no feeding larvae horizontally or slightly downward, ending at were observed doing this, few were found, cell entrances. Hence, cells tended to be close and they were difficult to examine under field to the main burrows. All laterals leading to conditions. The labiomaxillary regions oflar- closed cells were soil filled. vae of all ages were so recessed that it seems Cells were usually arranged singly, but sev- unlikely the salivary secretion, no matter how eral pairs were found in linear series sepa- fluid, could account for the shiny material. rated by a short distance (1.25-1.4 mm; N = Certainly no fibrous silk was found, and no 2). Their long axes were more or less hori- immature larva ofany species ofbee is known zontal or sloped to the rear by as much as to produce silk. On the other hand the agile, 300 from horizontal. Symmetrical around extensible and retractable, upward-pointed

Figs. 23, 24. Nesting sites of Conanthalictus conanthi, 4 mi east of Willcox, Cochise Co., Arizona. 23. Site west of road, studied on May 17, 1992; most nests in center foreground. 24. Site east of road, studied between August 16 and September 7, 1992; most nests in front of umbrella shadow, center foreground. 14 AMERICAN MUSEUM NOVITATES NO. 3066 abdominal apex raises the possibility that an len-nectar ball in an open cell was only 1.25 anal secretion might account for the appli- mm in diameter and seems to indicate that cation of the glistening material. earlier provisions are shaped into small This matter, ofobvious interest, needs fur- spheres and that subsequent loads are added ther study. A somewhat similar situation was to the sphere, as is the case with Sphecodo- reported for Colletes (Torchio et al., 1988, soma. Complete pollen-nectar masses, mealy- and references therein) and for Hesperapis moist, and all spherical or nearly so, were larreae Cockerell (Rozen and McGinley, 1.45-1.48 mm in diameter (N = 4), and at 1991), suggesting that a broader and more least one was coated with a nonreflective, thorough investigation ofmodifications ofcell transparent material. This material floated linings/walls by noncocoon-spinning larvae4 from the food surface when the food ball was is warranted. Such a study should not over- placed in water. look larval anatomical structures possibly as- Eggs (fig. 8) were on top ofthe pollen-nectar sociated with applying secretions to cell walls balls in the median sagittal plane of the cell, such as may be the case with the abdominal almost certainly with their rounded anterior apex ofConanthalictus larvae. There are some end pointed closest toward the cell closure. striking similarities between these larvae (see Strongly curved and whitish, they possessed description ofthe larva of C. conanthi below) a transparent, smooth chorion. The anterior and those of some species of Hesperapis end was substantially larger and more round- (Rozen and McGinley, 1974). ed than the posterior end, and the egg tapered Walls of freshly constructed cells were from just behind the nearly spherical anterior slightly more consolidated than the substrate, end toward the posterior end. and those ofcells with glistening material were Young larvae, nearly identical in appear- obviously held together by the reflective ce- ance to those of Sphecodosoma dicksoni ex- ment. cept lacking prothoracic dorsolateral tuber- Of the 15 nests excavated, none contained cles, crawled over the food surface as they more than three cells, and often the second fed. The food surface was not channeled (as cell ofthe nest was much lower than the first. is the case with the food masses ofsome elon- Few cells per nest may be the result of the gate, crawling anthophorid larvae). The di- female's nest construction being disrupted by ameter ofthe food mass decreased as the lar- the cattle in the region, so that, unable to find va grazed over its surface. A larva moved by their nests, females were forced to start new detaching and extending the anterior body ones. However, some nests contained a ma- segments and, after securing the front of the ture or nearly mature larva in the upper cell body, by then detaching and contracting the and just an egg or open cell below, a sugges- posterior segments. On young larvae the ven- tion that the species is slow in nest excavation trolateral tubercles help establish contact with compared with many other bees. In all cases, the food surface, as is the case in Sphecodo- older immatures were higher in the nest than soma and probably Rophites as well. Abdom- younger ones, an indication that nests were inal segments X pointed dorsally away from progressive. the food surface and, when in contact with PROVISIONING AND DEVELOPMENT: Fe- the cell surface, extended and thus pushed males transported dry but apparently sticky the posterior part of the body forward. That pollen on the anterior venter of the metaso- is, it was a dorsally directed pygopod, able to mata and on their hind tibiae and to some contract and extend, and possibly to apply a extent on their hind basitarsi, femora, and secretion to the cell wall as the larva moved perhaps even trochanters. A single small pol- over the food ball, as suggested above. The anatomical mechanism involved with the of the apex should 4Stephen et al. (1969) appropriately regarded the lar- mobility abdominal val envelope of Colletes as a cocoon. The silk is secreted be studied further, but one component al- by the Malpighian tubules through the anus, as is the most certainly is a ventral area between ab- case for a number ofinsect groups. For the present study dominal segments IX and X (fig. 37) that is the term noncocoon-spinning larvae refers to bee larvae delimited anteriorly and posteriorly by dis- that do not construct cocoons with salivary silk. tinct transverse integumental lines at least on 1993 ROZEN: ROPHITINE BEES 15 a cleared specimen. This area appears to be Xeralictus) hibernate as mature larvae (as do part ofabdominal segment X because on both most bees) so that Conanthalictus may pass male and female larvae the imaginal discs are the winter as quiescent postdefecating larvae. in their normal position on segment IX de- Ifthis is true, then C. conanthi has an annual fined by the anterior boundary of this spe- bimodal adult activity pattern in favorable cialized area. This area helps elevate the rest years at this locality and seems to be multi- ofsegment X so that it points dorsally. Some- voltine so long as the food source is available. what similar areas, delimited (e.g., Xeralic- PARASITISM: No cuckoo bees were associ- tus; Snelling and Stage, MS) or not (Spheco- ated with the site, either in the spring or late dosoma dicksoni, fig. 28) by transverse summer. The single postdefecating larva from integumental lines, are present on other a nest containing but a single cell had an elon- known rophitine larvae causing the terminal gate white egg (fig. 40) attached to the front segment to point dorsally. right side of its body. The egg, presumably An older larva, mostly grown, was en- that of a mutillid, had a spiculated chorion countered encircling the pollen-nectar mass and one end more rounded than the other. so as to completely hold the mass away from The main burrow of this nest (fig. 5), unlike the cell wall. Presumably at this time the lar- burrows of other nests, was filled with soil in va no longer moves in relation to the pollen many places, presumably the result of the mass. The dorsolateral tubercles on the lar- parasite's activity. va's thorax and abdomen were the points of contact with the cell wall. A single cell containing a postdefecating Rophites trispinosus larva lacked a cocoon and had the yellow I was shown the nesting site of this species feces applied as a thin layer to the rear one- by Professor J. de Beaumont at Vercorin, Va- third of the cell (fig. 10). Blotches of white lais, Switzerland, elevation 1340 m, on July material appeared on the exposed surface of 4, 1964, where nests occurred along a barren, the feces, representing the final meconial dis- trodden foot path. Although I made no notes charge. The greatly recessed labiomaxillary on the excavations, the following is derived region and obscure palpi on all larvae col- from larvae and food masses preserved at lected (figs. 35, 36) indicate that this species that time. never spins a cocoon, in contrast to larvae of PROVISIONING AND DEVELOPMENT: A single Dufourea, Rhophitoides, Rophites, Specodo- food mass, spherical and 4.1 mm in diameter, soma, and Systropha. was preserved in Kahle's solution. No wa- ADULT ACTIvITY: Adults were most abun- terproofcoating was visible, although a coat- dant on flowers during the middle ofthe day. ing may have dissolved in the preservative. Mating probably takes place at the flowers A curved egg (fig. 9), 2.4 mm long and 0.6 judging by the abundance ofboth sexes there. mm in maximum diameter, possessed a re- Males were not seen around nests. ticulated chorion and was rounded at the an- There were two seasons ofadult activity at terior end and considerably more pointed at the site, spring and late summer, correspond- the posterior end. The embryo within was ing with the flowering ofNama. Because ma- oriented with its venter against the outcurved ture larvae and pupae were not uncovered side. An intermediate-stage larva (fig. 21) en- from nests on May 17, it is unknown how circled another food mass, suggesting that it individuals aestivate between spring and late gripped the mass tightly while it fed on it. summer. It does seem clear, however, that Intermediate-stage larvae (fig. 21) as well they are inactive during this time because of as a large larva (fig. 22) with salivary lips the absence of flowering food plants. Two (probably a young last-stage larva, described postdefecating larvae collected and preserved in the section of mature larvae, below) pos- in early September had developed pupal tis- sessed paired dorsolateral body tubercles. As sue, an indication that they would have main other rophitines (except for Conanthalictus tured the same season. Other known rophi- conanthi), the prothoracic tubercles were tines (Dufourea, Rhophitoides, Rophites, much smaller than the others, but the paired Sphecodosoma, Systropha and apparently dorsal tubercles of abdominal segment VIII 16 AMERICAN MUSEUM NOVITATES NO. 3066 were scarcely if at all evident. Middorsal tu- do not reach one another. Labiomaxillary re- bercles (reported for Systropha by Batra and gion greatlyfused. Larva unusually elongate; Michener, 1966) were absent, but subspira- most body segments possessing low, rounded cular protrusions (tubercles?) were evident at dorsolateral tubercles. Posterior part of ab- least on a medium-size larva. Intermediate domen (quite similar to that of its host) with larvae of all sizes (figs. 21, 22) (including the venter of abdominal segment IX protruding; specimen with protruding salivary lips) pos- abdominal segment X somewhat dorsal in at- sessed a pair of small, posteriorly directed, tachment to IX. ventrolateral tubercles toward the posterior margins of the mesothorax, metathorax, and PROFILE OF THE BIOLOGY abdominal segments I-VII. These tubercles OF THE ROPHITINAE were quite distinct, much like those of C. Sufficient information is on hand to present conanthi, and not as large and vague as those an overview ofthe biological features (except of S. dicksoni. The tubercles as well as the for pollination biology) of the Rophitinae, as strongly protuberant venter of abdominal has been done for some other higher taxa segment IX suggest that these structures (Rozen, 1977, 1984a, 1984b, 1989). This in- probably function in the same way as do their homologs in C. conanthi and S. dicksoni. The formation comes from the following ac- protruding (but very narrow) salivary lips and counts: a labium divided into a prementum and post- Conanthalictus conanthi (Cockerell) mentum found on the specimen (figs. 41, 42) Rozen and McGinley (1976); present study that was presumably an early last instar in- Dufourea malacothricis Timberlake dicates that this species spins a cocoon. Torchio et al. (1967) the cuckoo bee D. mulleri (Cockerell) PARASITISM: Adults of Torchio et al. (1967) Biastes emarginatus (Schenck) (Nomadinae: D. novaeangliae (Robertson) Biastini) were captured as they flew over the Kukuk et al. (1985); Eickwort et al. (1986) path, leaving little doubt that they parasitized D. pulchricornis (Cockerell) the nests of this host. However, it was not Torchio et al. (1967) until I examined preserved larvae in connec- D. trochantera Bohart tion with this paper that I realized that an Torchio et al. (1967) intermediate-stage larva of this cleptoparas- Rhophitoides canus (Eversmann) ite had been recovered. The anterior end of Enslin (1921); Malyshev (1925b) the specimen had been covered with pollen, Rophites hartmanni Friese Malyshev (1925b, 1935) its midsection somewhat damaged and dis- R. quinquespinosus Spinola torted, and the protruding venter of abdom- Stockhert (1922) inal segment IX and dorsolateral body tu- Sphecodosoma dicksoni (Timberlake) bercles were surprisingly similar to those of Rozen and McGinley (1976); present study the host. This is apparently the first associ- Systropha curvicornis (Scopoli) ation ofthis cleptoparasite with Rophites tris- Malyshev (1925a) pinosus although it has been associated with S. planidens Geraud R. quinquespinosus Spinola (Stockhert, 1922). Malyshev (1925a) The similarities between this specimen and S. punjabensis Batra and Michener the mature larva of the related Neopasites Batra and Michener (1966) (Biastini) (Rozen, 1966) are striking. Those Xeralictus timberlakei Cockerell italicized below are characteristic of mature Snelling and Stage, MS larvae ofthe Nomadinae (sensu Roig-Alsina, The profile is tentative because the nesting 1991); those in boldface are probably syna- biologies of many genera (Michenerula, Mi- pomorphies ofthe Biastini and in some cases cralictoides, Morawitzella, Morawitzia, Pen- Neolarrini. Head capsule with antennal pa- apis, Protodufourea, Systrophidia, and Trilia) pilla small but projecting; labrum very short, have yet to be examined, and even the most somewhat recessed behind frontoclypeal area detailed published accounts are still incom- (as seen in lateral view) and bearing pair of plete. Hopefully, the profile will encourage labral tubercles. Mandibles apically thin, studies of other genera and provide compar- sharp-pointed, and so short that their apices ative points ofreference. Certain similarities 1 993 ROZEN: ROPHITINE BEES 17 in the biologies seem distinctive for the sub- young larvae elongate, capable ofcrawling as family and are italicized. theyfeed so that pollen-nectar mass reduced GENERAL: Subfamilial distribution (Mich- in size but remaining spherical; older larvae ener, 1979) primarily Holarctic but reaching grasping food mass while feeding on it and southern Africa and southeast Asia; one ge- holding it away from cell wall; food mass now nus (Penapis) in South America; no repre- looses spherical shape. Cocoons of salivary sentation in Australia; most abundant in xer- silk constructed except in Conanthalicus and ic regions. Body size minute to moderate. presumably in Xeralictus, completely filling NESTING: All species ground nesting, soli- cell lumen; cocoon fabric (except at front end) tary, mostly noncommunal, often apparently thin, normally consisting oftwo layers of silk nesting in loose aggregations; none clepto- (except in case of nonhibernating larva of parasitic. Nesting surface horizontal, sloping, Sphecodosoma dicksoni) between which there and even vertical. Substrate texture appar- is layer offeces; defecation commencing after ently variable. Nests shallow (< 10 cm deep) all food consumed and outer layer of cocoon to moderately deep (maximum depth record- produced; feces applied to rear ofcocoon and ed 33 cm), apparently always progressive; often extending only part way to front of co- preexisting cavities and old nests not used. coon; inner layer of silk covering feces; front Tumuli present, usually eccentric or acentric ofcocoons often with specialized thicker fab- and bilaterally symmetrical; turrets absent. ric consisting of several silk layers separated Nest consisting ofsingle, open (at least above) by air spaces, without macropyle (Rozen and main tunnel and short to moderately short Jacobson, 1980). subhorizontal laterals that are soil-filled when ADULT ACTIVITY: Bees diurnal, probably completed. Main tunnels ofmany species en- flying primarily during middle of day; most tering ground at oblique angle to horizontal, species probably univoltine, but some par- meandering considerably in upper part, and sivoltine, and others multivoltine. Bees over- descending more vertically and with less me- wintering as quiescent larvae in cocoons (un- andering in lower part; main tunnels without known for Conanthalictus). Many species vestibules, blind laterals, or blind terminal oligolectic (Hurd, 1979). burrows. Nests with a number of cells; cells CLEPTOPARASITIc BEES: Among Biastini, arranged singly and also commonly in linear Neopasites definitely associated with some series, but not in clusters; cells in series sep- species ofDufourea (Hurd, 1979); Biastes with arated by only width of closure; cells ovoid, Rophites, Systropha, and Dufourea (Popov, apparently tending to be symmetrical around 1933); and Rhopalolemma almost certainly long axis and short relative to maximum di- with Protodufourea (Eickwort, personal com- ameter; cell orientation horizontal, subhori- mun.). In Townsendiellini, Townsendiella zontal, or tilted to rear by as much as 40 or suggested as cleptoparasite of Conanthalic- even 700 from horizontal. Cell lining dull, tus, but this yet to be confirmed. more or less water absorbent; cell walls not obviously soil lined. Cell closure more or less concave spiral of three to four coils. MATURE LARVAE OF THE PROVISIONING: Pollen transported primar- ROPHITINAE ily on hind femur, tibia, and basitarsus and The following is based on available spec- in many taxa also on sides of propodeum, imens of Conanthalictus, Dufourea, Sphe- and in Systropha, Systrophidia, and others codosoma, Xeralictus, and an immature spec- also abundantly on sides and undersurface of imen of Rophites. It also takes into account metasoma. Early pollen-nectar loads shaped information from the literature on Dufourea into smallsphere; completedprovisions spher- mulleri (McGinley, 1981, 1987); D. novaean- ical; provisions of some species with water- gliae (Eickwort et al., 1986); Systropha pun- proof or semiwaterproof, presumably secret- jabensis (Batra and Michener, 1966); and ed, transparent coating, but this coating Xeralictus timberlakei (Snelling and Stage, apparently absent in other species; provisions MS). not attached by liquid to cell floor. DIAGNOSTIC DESCRIPTION: Mature larvae DEVELOPMENT: Eggs apparently usually of the Rophitinae can be recognized by the strongly curved, placed on top of food mass; following characters (features in boldface 18 AMERICAN MUSEUM NOVITATES NO. 3066 should be particularly convenient to use): KEY TO THE MATURE LARVAE Clypeus and labrum usually short (except for 1. Labiomaxillary region (figs. 29, 42) not re- Xeralictus and perhaps Rophites), usually cessed; labium divided into prementum without obvious paired tubercles but these and postmentum; salivary lips (figs. 29, present in Rophites (and apparently Rhophi- 42) present; maxillary and labial palpi (figs. toides [Enslin, 1921: fig. 1] and Systropha 29,42) as long as or longer than their basal [Batra and Michener, 1966]); hypopharyn- diameters; prothoracic dorsolateral tuber- geal groove absent (Conanthalictus, Dufou- cles (figs. 25, 26) present but smaller than rea, Sphecodosoma), present laterally (as in- those of following segments ...... 2 dentation along articulating arms of stipites) - Labiomaxillary region (fig. 35) greatly re- but not medially (Rophites), or complete (only cessed; labium (fig. 34) undivided; sali- in Xeralictus). Intrasegmental lines (figs. 25, vary lips absent (fig. 35); maxillary palpi (fig. 36) scarcely evident; labial palpi (fig. 32, 37) not evident at least dorsally (weakly 36) undeveloped except for sensilla; pro- developed in Xeralictus); dorsolateral body thoracic dorsolateral tubercles absent, fol- tubercles conical (i.e., not transverse), present lowing tubercles well developed (fig. 37) on most body segments; those of pronotum ...... Conanthalictus conanthi absent (Conanthalictus) or noticeably small- 2(1). Labiomaxillary region not extending beyond er (all others) than those of following seg- labrum in lateral view (fig. 42); outer sur- ments;5 labiomaxillary region more (fig. 27) face of mandible with large seta-bearing or less (fig. 42) produced except greatly re- tubercle (figs. 43, 45); articulating arm of cessed in Conanthalictus (fig. 35); salivary stipital sclerite conspicuous (fig. 41) .. 3 opening transverse, with projecting narrow - Labiomaxillary region strongly projecting beyond labrum in lateral view (fig. 27); lips (figs. 28, 41, 42) except lips absent in outer surface of mandible with seta-bear- Conanthalictus (figs. 35, 36); abdominal seg- ing tubercles very small, inconspicuous ment IX (figs. 35, 26, 29) more or less pro- (figs. 29, 31) or moderately small (Dufou- duced ventrally; abdominal segment X (figs. rea, Eickwort et al., 1986: fig. 14); artic- 25, 26, 39) dorsal in attachment to IX and ulating arm of stipital sclerite inconspic- directed somewhat dorsally. uous (fig. 28) ...... 4 Some of the characters used before (Eick- 3(2). Labral tubercles absent; antennal papilla wort et al. 1986; McGinley, 1987) to differ- shallow, projecting less than half its di- entiate the larvae ofthe Rophitinae from oth- ameter ...... Xeralictus timberlakei er taxa are no longer as useful. For example - Paired labral tubercles (fig. 42) well devel- not all rophitines have strongly projecting la- oped; antennal papilla small but project- biomaxillary regions and transverse, projecting, about as long as basal diameter (fig. 42) ...... Rophites trispinosus ing salivary lips (Conanthalictus), and some 4(2). Antennal disc (McGinley, 1987: fig. 27.187a) rophitines (Rophites, Xeralictus) have a large on cleared head capsule moderate in size, seta-bearing tubercle on the outer surface of so that distance between disc and upper the mandible. edge of mandibular corium (as measured by arrow in fig. 28) is 1.5 to 2 times disc diameter; subatrium of normal length, of less than 13 chambers ...... 5 Eickwort et al. (1986) stated that the dorsolateral Dufourea mulleri and D. novaeangliae prothoracic tubercles ofthe then-known rophitines were - Antennal disc (fig. 28) large, so that distance "only slightly lower than the other thoracic tubercles" between disc and upper edge of mandib- rather than "markedly reduced" as reported by Rozen ular corium is less than 1.5 times disc di- and McGinley (1976). Through the kindness of George ameter; subatrium (fig. 32) abnormally C. Eickwort, I have been able to examine the mature long, of 15 or more chambers ...... larva of Dufourea novaeangliae and agree that the pro- ...... Sphecodosomadicksoni thoracic tubercles, though smaller than the mesothoracic ones, are not as greatly reduced in relation to the other Sphecodosoma dicksoni dorsolateral body tubercles as those ofDufourea mulleri. Figures 28-32 Nonetheless, prothoracic tubercles smaller than those of the meso- and metathorax (or absent) impart a charac- The format and terminology used here are teristic appearance to all larval rophitines that I have those proposed by Rozen and Michener seen. (1988). 1993 ROZEN: ROPHITINE BEES 19

ANTERIOR TENTORIAL PIT ARTICULATING LABIOMAXILLARY REGION ARM OF STIPES

TUB RCLE 26~~~~~~~~~~~~~~~~~~~~~

to_SXFORE TIBIA

TEGULAR

MESOSCUTEL. TUBERCLE--

MESOSCUTELLAR TUBERCLE MID TIBIA (

TEGULAR TUBERCLE - LATERAL LOBE - BASAL TUBERCLE OF PRONOTUM OF HIND TIBIA

LATERAL OCELLAR 34

MAXILLARY PALPUS MESEPISTERNAL TUBERCLE GALEA Figs. 25-32. Mature larva of Sphecodosoma dicksoni. 25. Overwintering larva, lateral view. 26. Predefecating larva, lateral view. 27, 28. Head, lateral and frontal views, respectively. 29-31. Right mandible, dorsal, adoral, and ventral views, respectively. 32. Spiracle, side view. Figs. 33 and 34. Pupa of Sphecodosoma dicksoni, lateral and dorsal (left half) views, respectively. Scale refers to figures 26, 27, 33, and 34. 20 AMERICAN MUSEUM NOVITATES NO. 3066

DIAGNOSIS: At present the only known clypeus; labral sclerite not evident; labral tu- characters (in boldface) by which mature lar- bercles (fig. 27) very weak, inconspicuous, but vae of Sphecodosoma can be distinguished present, too vague to determine whether aris- from those of other Rophitinae are the large ing from labral disc or labral apex; lateral antennal discs and papillae and the very elon- areas bearing a few scattered spicules; epi- gate, multichambered subatrium. Rophites pharyngeal surface nonspiculate except lat- also has a long, multichambered subatrium erally. but can be recognized by its very small an- Mandible (figs. 29-31) moderately robust tennal papilla relative to the moderately large at base tapering to simple, slender apex; dor- disc (fig. 41). sal mandibular surface with a few small, HEAD (figs. 27, 28): Integument of head sharp-pointed spicules; outer surface ofman- capsule with scattered sensilla that are small dible with seta-bearing tubercle small, not and not obviously setiform. Integument un- pronounced; dorsal adoral surface with nu- pigmented except for mandibular apices, merous large sharp-pointed teeth in vicinity mandibular articulations, and, in postdefe- ofcusp; dorsal apical edge bearing one or two cating larvae, salivary lips. teeth; ventral apical edge apparently either Head size (figs. 25, 26) small compared to nondentate or bearing one or two teeth; apical body size; head capsule distinctly wider than concavity weakly developed, apparently rep- length measured from top of vertex to lower resented by shallow groove extending from clypeal margin in frontal view. Tentorium apex along ventral surface; this area nonspi- well developed, possessing well developed culate. Labiomaxillary region (fig. 27) greatly dorsal arms; anterior tentorial pits normal in produced. Maxillary apex not produced position, not immediately adjacent to ante- mesally; sclerotized cardo and stipes unpig- rior mandibular articulations; posterior ten- mented, difficult to see but represented by torial pits in normal position at junction of thickened integument best seen on cleared posterior margin of head and hypostomal specimen; articulating arm of stipital sclerite ridges; postoccipital ridge (i.e., posterior difficult to see but present in position shown thickening of head capsule) moderately de- (fig. 28); galea not produced as low swelling veloped, scarcely curving forward medially but perhaps represented by sensillum mesad as seen in dorsal view; posterior margin of of maxillary palpus; maxillary palpus elon- head normal in position; median longitudinal gate, longer than basal diameter. Labium di- thickening of head capsule absent except for vided into prementum and postmentum; pre- faint thickening of integument at very top; mental sclerite unpigmented; labial palpus hypostomal ridge strongly developed, with- equal in size to maxillary palpus. Salivary out ramus, of moderate length, forming ap- opening a moderately narrow transverse slit proximately 900 angle with posterior margin borne ofprojecting, well-developed lips. Hy- as seen in lateral view (fig. 27); pleurostomal popharynx a vaguely transversely grooved ridge only moderately developed; epistomal surface bearing spicules medially, gently ridge moderately developed, extending al- curved but not strongly bulging; hypophar- most to midline on front of head (fig. 28); yngeal groove not developed so that hypo- epistomal depression inconspicuous. Parietal pharyngeal surface grading into dorsal labial bands evident. Antennal prominence mod- surface. erately weak (fig. 27); antennal disc large (fig. BODY: Integument without setae but dor- 28); antennal papilla large (fig. 28), bearing solateral tubercles with small sensilla; ventral three sensilla. Vertex evenly rounded as seen areas ofmost body segments with small even- from side (fig. 27), without unusual projec- ly spaced spicules; dorsal integument of ab- tions; clypeus of moderate width, short so dominal segments VII-X with fine transverse that clypeolabral suture almost in line with linear asperities that become increasingly anterior mandibular articulations as seen in pronounced toward abdominal apex; dorso- frontal view (fig. 28); frontoclypeal area in lateral tubercles microscopically irregularly lateral view (fig. 27) scarcely projecting and roughened compared with rather smooth in- not produced beyond labrum. Labrum in lat- tegument elsewhere. Body form (figs. 25, 26) eral view projecting only slightly beyond moderately robust, not greatly elongate (al- 1993 ROZEN: ROPHITINE BEES 21 though early instars quite elongate); interseg- terad of anterior mandibular articulation so mental lines weakly incised; intrasegmental actually occurring on (or just above) pleuros- lines (figs. 25, 26) not evident; paired dor- tomal ridge; postoccipital ridge (posterior solateral body tubercles conspicuous on me- thickening of head capsule) (fig. 35) absent sothorax, metathorax, and abdominal seg- although posterior boundary of capsule de- ments I-VIII, comparatively reduced in size limited by constriction; median longitudinal on prothorax, absent on abdominal segment thickening of head capsule absent; hyposto- IX, X; tubercles conical rather than trans- mal ridge moderately developed; epistomal verse; other tubercles absent on mature larva ridge (fig. 36) absent; epistomal depression (see section on biology for other instars); ven- absent. Parietal bands not evident. Antennal ter of abdominal segment IX produced; X prominence (fig. 35) large, globose, occupying attached dorsally to IX; apex of X without much of anterolateral area of parietal; anten- ridges or other modifications; anus dorsal in nal disc of moderate size; antennal papillae position at apex of X, as seen in lateral view not fully differentiated from disc, projecting (figs. 25, 26). Spiracles (figs. 25, 26, 32) ex- only slightly, bearing unknown number of tremely small and inconspicuous; thoracic sensilla. Clypeus presumably short as in spiracles slightly larger than others; spiracles Sphecodosoma but upper boundary not de- without sclerites, not on tubercles; peritreme marked by epistomal ridge; clypeolabral su- apparently present; atrium perhaps project- ture almost in line with anterior mandibular ing slightly above body wall, with distinct articulations; frontoclypeal area in lateral view rim, globose; atrial wall smooth; primary tra- even less projecting than in Sphecodosoma, cheal opening with collar; subatrium (fig. 32) appearing recessed because of large antennal extremely long, consisting of many chambers prominences. Labrum (fig. 35) not projecting (at least 20 on thoracic spiracles but fewer on as far as clypeus in lateral view; labral tu- abdominal spiracles). Male with distinct but bercles virtually absent; labrum and epi- small cuticular invagination ventrally on pharynx apparently not spiculate. midline of abdominal segment IX near its Mandible (figs. 38, 39) with conspicuous posterior margin; female characters unknown subapical tooth; dorsal surface without spic- beyond the absence of male character stated ules; outer surface with only minute tubercle; above. dorsal adoral surface without teeth except for MATERIAL STUDIED: 2 postdefecating, 4 conspicuous sharp-pointed teeth that extend predefecating larvae 18 mi west of Blythe, along dorsal apical edge; ventral apical edge Riverside Co., California, April 30, May 1, without teeth. Labiomaxillary region (fig. 35) 11, 1992 (J. G. and B. L. Rozen). greatly recessed. Cardo, stipes, articulate arm of stipital sclerite not evident; maxillary palpus scarcely evident, much shorter than basal Conanthalictus conanthi diameter. Labium not divided into premen- Figures 35-39 tum and postmentum, very short, its poste- DIAGNOSIS: The numerous characters pre- rior margin in line with posterior constriction sented in the following description will easily of head capsule as seen in figure 35; labial separate this species from other known larvae palpus not evident except perhaps for vague in the subfamily. Those that are especially sensilla. Salivary lips absent; opening of sal- diagnostic are in boldface. ivary duct (fig. 36) a transverse slit. Hypo- HEAD (figs. 35, 36): As described for Sphe- pharynx nonspiculate, apparently not codosoma dicksoni except for following: Sen- grooved; hypopharyngeal groove absent. silla of head capsule so small that those of BODY: Dorsolateral tubercles with apices antennae, palpi, and parietals are scarcely if perhaps slightly roughened but sensilla not at all visible under high stereoscopic magni- evident; some areas ofbody weakly spiculate fication. Integument unpigmented except for but without fine transverse linear asperities. mandibular apices; anterior and posterior Body form (fig. 37) moderately slender; paired tentorial arms present but rest of tentorium prothoracic dorsolateral tubercles absent; ab- absent perhaps because specimen approach- dominal segment IX with small paired dor- ing ecdysis; anterior tentorial pit (fig. 36) la- solateral tubercles; venter of abdominal seg- 22 AMERICAN MUSEUM NOVITATES NO. 3066

Figs. 35-39. Postdefecating larva of Conanthalictus conanthi. 35,36. Head, lateral and frontal views, respectively. 37. Entire larva, lateral view. 38, 39. Right mandible, adoral and ventral views, respectively. Fig. 40. Egg of parasite (probably a mutillid) found attached to postdefecating larva. Figs. 41- 45. Probable early last larval instar of Rophites trispinosus. 41, 42. Head, frontal and lateral views, respectively. 43-45. Right mandible, dorsal, adoral, ventral views, respectively. Scale refers to figures 37 and 40. ment IX weakly produced but appearing female with paired white, opaque, imaginal pronounced because of small, dorsally posi- discs visible through ventral integument of tioned abdominal segment X; X connected abdominal segments VII, VIII, and IX; those to IX ventrally by anterior extension which of IX nearly contiguous. permits tip of abdomen to be directed dor- MATERIAL STUDIED: Two postdefecating sally (fig. 37). Spiracles moderately small but larvae, 1 predefecating larva, and 2 nearly appearing obscure because of lack of pig- mature larvae, 4 mi east of Willcox, Cochise mentation; spiracles subequal in size; atrial Co., Arizona, September 1-5, 1992 ( J. G. wall perhaps faintly ringed; subatrium nor- and B. L. Rozen). mal in length, consisting of 8 or 9 chambers. REMARKS: Features ofthe head ofthe larva Male with distinct but small cuticular invag- are remarkably different from those of the ination posteriorly on ventral midline of ab- other rophitines, but at the same time such dominal segment IX as in Sphecodosoma; important synapomorphies as the reduction 1993 ROZEN: ROPHITINE BEES 23

(absence) of paired prothoracic dorsolateral istomal depression absent. Antennal disc tubercles, short labrum and clypeus, and moderate in size; antennal papilla small but modifications of the abdominal apex dem- strongly projecting, about as long as basal onstrate its affinities to other members ofthe diameter; clypeus of moderate length; posi- subfamily. It seems likely that many of the tion of clypeolabral suture indefinite relative head modifications (reduction of labiomax- to anterior mandibular articulations. Labrum illary regions, reduction in size of palpi, loss in lateral view extending beyond clypeus be- of salivary lips, and perhaps the loss of the cause of pair of rounded labral tubercles that epistomal suture) are related to the fact that arise from labral disc; lateral areas of labrum this species does not spin a cocoon. Such re- spiculate; epipharyngeal surface spiculate. ductions are found in other noncocoon-spin- Mandible (figs. 34-45) with dorsal surface ning bee larvae. However, other features (re- more densely spiculate that ofSphecodosoma duction in size of head sensilla, globose dicksoni; outer surface of mandible (figs. 43, antennal prominences, and loss ofpostoccip- 45) with pronounced seta-bearing tubercle; ital ridge) are more difficult to explain. The dorsal apical edge bearing numerous small apparent position ofthe anterior tentorial pits teeth and irregularities; ventral apical edge on the pleurostomal ridge is unusual. Since with small teeth and irregularities. Labiom- the pits are actually slightly above this ridge axillary region not greatly produced. Maxilla (fig. 36), the brief ridges connecting them to with articulating arm of stipital sclerite (fig. the pleurostomal ridges are probably the lat- 41) long, well developed; maxillary palpus eral remnants of the epistomal ridge. elongate, even more so than that of S. dicksoni. Labial palpus like maxillary palpus. Sal- ivary lips (figs. 41, 42) well developed, pro- Rophites trispinosus jecting but very narrow. Hypopharynx more Figures 22, 41-45 strongly curved than that of S. dicksoni; hy- The following information is based on a popharyngeal groove evident laterally as in- single larva (also discussed in the section on dentation along articulating arms of stipital nesting biology) that is poorly preserved, not sclerite. fully mature, and probably teneral. The fol- BODY: As described for Sphecodosoma lowing features suggest that it may be the last dicksoni except for following: Integumental instar: well developed salivary lips, long slen- spicules and sculpturings unknown because der palpi, sclerotized maxillary sclerites, and larva poorly preserved and teneral. Body form spiracular development. However, the la- of mature larva unknown (early instars elon- biomaxillary region (fig. 42) projects only gate); intersegmental and intrasegmental lines moderately and the ventrolateral tubercles are impossible to evaluate because larva not fully well developed (such tubercles are not evi- fed; paired, rounded dorsolateral body tu- dent in last-stage larvae of Sphecodosoma bercles large and very conspicuous on me- dicksoni). Whatever its stage, it does provide sothorax, metathorax, and abdominal seg- enough information about the last stage for ments I-VII; prothoracic tubercles smaller the following abbreviated description. than following ones; abdominal segment VII DIAGNOSIS: This larva can be separated apparently with small dorsolateral tubercles; from those of other known rophitines (with size ofall dorsolateral tubercles oflittle com- the possible exception of Systropha, Batra parative value because specimen immature and Michener, 1966) in that the labrum bears (hence body not drawn in lateral view); ven- a pair of distinct, apically rounded labral tu- trolateral tubercles present on many seg- bercles. An illustration of the mature larva ments (fig. 22) (but these tubercles are prob- of Rhophitoides canus (Enslin, 1921: fig. 1) ably absent on fully fed larva). Spiracles small; suggests that it also has paired labral tuber- sizes relative to one another unknown; su- cles. batrium long, consisting of more than 15 HEAD (fig. 41, 42): As described for Sphe- chambers (i.e., like that of S. dicksoni, fig. codosoma dicksoni except for following: 32). Male with distinct cuticular invagination Only mandibular apices faintly pigmented on venter of abdominal segment IX (as in S. (presumably because specimen teneral). Ep- dicksoni); female sex characters unknown. 24 AMERICAN MUSEUM NOVITATES NO. 3066

MATERIAL STUDIED: 1 immature, possibly elongate, pronounced basal tubercle on outer last instar, Vercorin, Switzerland, July 4, 1964 surface; foretibia with distinct apical tubercle (J. G. Rozen). on outer surface; midtibia with smaller tubercle in same position; hind tibia without distinct apical tubercle. PUPA OF SPHECODOSOMA DICKSONI METASOMA: Integument with fine regular Figures 33, 34 spicules in many areas; setae absent. Tergum DIAGNOSIS: Because no other pupa of this I with a few small rounded tubercles forming subfamily has been described (although that transverse band near posterior margin; terga ofRhophitoides canus was pictured by Enslin, II-IV with transverse bands ofsmall rounded 1921), distinguishing features separating this tubercles near posterior margins; tergum V species from related taxa cannot be given. No with fewer such tubercles; tergum VI without doubt the size, shape, and distribution ofpu- tubercles; sterna without tubercles; apex of pal tubercles (in contrast to tubercles that metasoma rounded, not produced as elongate cover developing adult features) will be di- apical spine. agnostic. The extremely elongate maxillary MATERIAL STUDIED: 2 female pupae, 18 mi palpi relative to the short galeae (reflecting west ofBlythe, Riverside Co., California, col- the sizes of the adult structures) (fig. 33) may lected May 1, 1992, preserved as pink-eyed be important in recognizing pupae of many pupae May 7, 1992 (J. G. Rozen). Rophitinae. HEAD: Integument possibly spiculate but difficult to see, some areas with indistinct ir- DISCUSSION regularities, but without setae. Scape with To what extent do rophitine bees share bi- several integumental irregularities on outer ological and/or larval synapomorphies with surface and with large rounded apical tuber- the Halictinae and Nomiinae? This question cle; pedicel and flagellum without tubercles. has been addressed with limited success by Vertex with lateral ocellar tubercle on each Torchio et al. (1967) and Eickwort et al. side arising from irregular, raised base; me- (1986). This is due in part to the fact that dian ocellar tubercle small; genal tubercle ab- little information was available for the Ro- sent; frons with lumpy elevation on each side; phitinae, many of the features of the Rophi- clypeus without tubercles; labrum unmodi- tinae appear to be plesiomorphic, and com- fied. Mandible very strongly swollen subapi- parative data especially concerning the cally but without tubercles; other mouthparts Nomiinae have yet to be compiled. without tubercles. There is a considerable body of organized MESOSOMA: Integument finely, evenly spic- data on the nest architecture ofthe Halictinae ulate in some places; setae absent. Lateral (Sakagami and Michener, 1962; Eickwort, lobe of pronotum swollen but without tu- 1969; Eickwort and Sakagami, 1979; and ref- bercle; posterior margin of pronotum with erences therein). The nest patterns of the transverse swelling on each side; mesepister- Nomiinae have not been so thoroughly stud- num with small but distinct rounded tubercle ied although a number ofpapers (Hirashima, behind foretrochanter (fig. 33); mesoscutum 1961; Stephen et al., 1969) give useful infor- with low mound on each side near anterior mation concerning a number of taxa. The margin; axilla with low mound; mesoscutel- nest patterns of these two subfamilies are lum with moderate tubercle on each side; highly variable and, at least in the Halictinae, metanotum somewhat produced medially. of potential value in phylogenetic analysis Tegula with distinct moderate-size tubercle; (Eickwort and Sakagami, 1979). However, the wings with outer surface somewhat irregular nests of the Rophitinae fall into one basic but without tubercles. Each coxa with small primitive type that is also found in the other apicoventral tubercle; each trochanter with two subfamilies as well as in a number of larger apicoventral tubercle; fore and hind other families ofbees. At this time, gross nest femora with basal, pointed swelling; fore- and architecture seems of no value for phyloge- midtibiae each with small rounded basal netic interpretation of the Rophitinae. swelling on outer surface; hind tibia with Other aspects of nesting biology of the ro- 1993 ROZEN: ROPHITINE BEES 25 phitines appear to be specialized. If homol- these two subfamilies consume the food ogies can be found among taxa outside the masses along one side so that the loaves subfamily, conclusions about relationships quickly become asymmetrical as seen in side may be forthcoming. The three most note- view. worthy specializations ofthe Rophitinae are: When Eickwort et al. (1986) described the 1) Nonreflective, more or less water ab- larva of Dufourea, they found no synapo- sorbent cell walls. Cell walls ofthe other two morphies of the mature larvae that linked subfamilies are shiny and water repelling, a rophitines with the other two halictid sub- feature that seems to be plesiomorphic since families. The present study demonstrates that lining of cells in many families of bees are one of the rophitines, Conanthalictus, does shiny.6 The nonreflective nature ofthe lining not spin a cocoon with salivary silk. Like the should not be interpreted to mean that ro- mature larvae ofthe Nomiinae and Halictin- phitine cells are unlined. If cells were truly ae (none ofwhich spin cocoons), its larva also unlined, then the polarity of this character has a recessed labiomaxillary region, reduced might well be questioned. However, the fact palpi, missing salivary lips, and an undivided that water droplets on the cell wall are ab- labiomaxillary region. Although such simi- sorbed more slowly than those placed on the larities cannot be totally dismissed as syna- surface of the substrate suggests that a non- pomorphies, such reductions have occurred shiny lining of some sort is normally applied one or more times in the Colletidae and Mel- (but perhaps not in Conanthalictus). A lining ittidae and numerous times in the Antho- was detected in the case of Sphecodosoma, phoridae, all in conjunction with loss of co- reported here. Furthermore, the very smooth coon spinning. It seems likely that loss of nature of the rophitine cell surfaces seems to cocoon spinning with the anatomical reduc- suggest the same; bee cells that lack a lining tions has taken place de novo within the ro- (e.g., Hesperapis trochanterata Snelling, Roz- phitine clade that is defined by the larval syn- en, 1987; Parafidelia pallidula Cockerell, apomorphies of 1) paired conical dorsolateral Rozen, 1977) tend to have rough walls. tubercles on most body segments and 2) tu- 2) Early pollen-nectar loads shaped into bercles of prothorax reduced (compare with small spheres to which are added subsequent those ofrest ofthorax) or absent. For the time loads. With most bees, early loads of provi- being then, one ofthe conclusions ofthis pa- sions are normally dumped as irregular mass- per, like a conclusion in Eickwort et al. (1986), es into open cells; shaping of the food loaf is that the placement ofthe Rophitinae in the takes place only after the final load is brought Halictidae continues to rest on adult mor- into the cell. phology (e.g., Michener and Greenberg, 1985) 3) Young larvae ambulatory as they feed rather than on larval anatomy and nesting around food mass and consequently food biology. mass remaining spherical (except toward end Eickwort et al. (1986) compared in detail of consumption) as it decreases in size. This the biologies and larvae of the Rophitinae feature is associated with pronounced, pos- with those ofthe Melittidae and the andrenid teriorly directed ventrolateral tubercles on subfamilies Andreninae and Panurginae to most ofthe body segments ofthe early instars seek evidence of phylogenetic relationships. as well as with an abdominal segment X used The present study does not add sufficient ev- as a dorsally directed pygopod. This behavior idence to make a phylogenetic analysis of and the associated anatomical features are these similarities worthwhile at this time, but apparently unknown in either the Halictinae certain additional features have now come to or Nomiinae and are uncommon or absent light. These, listed and evaluated as follows, in other families of bees. Young larvae in should be added to the characters mentioned by Eickwort et al. (1986) in preparation for 6 In his review ofthis manuscript, Geroge C. Eickwort such an analysis. (in litt.) pointed out that the conspicuous shiny cell lin- 1) Early pollen-nectar supplies shaped into ings of Halictinae + Nomiinae may not be plesiomor- small spheres. This appears to be the case for phic because lactone contents suggest a separate origin the rophitines where data have been collected in the two subfamilies. (although for Dufourea novaeangliae, only 26 AMERICAN MUSEUM NOVITATES NO. 3066 intermediate loads were so shaped, Eickwort (Rozen and McGinley, 1974). Its function et al., 1986). Shaping of early loads is a fea- may be for locomotion, that is, a pygopod, ture that has been reported both for certain but, as mentioned above, it may also be in- Panurginae (Rozen, 1989) and for Hesperapis volved with applying a hardening (and wa- (Melittidae) (Rozen and McGinley, 1991), al- terproofing?) substance to the cell wall. Be- ways in association with spherical food mass- cause of its structural complexity, it would es. Because this behavioral feature at least in seem to be a strong synapomorphy (although the Panurginae is restricted to tribes thought anatomical details of this modification have to be specialized on the basis of adult and not been fully explored to determine homol- other behavioral features, this similarity be- ogies). However, there is a suggestion that the tween the panurgines and rophitines is similarities may actually be convergences re- thought to be the result of evolutionary con- sulting from the fact that the larvae overwin- vergence and not an indication of relation- ter in cells that have not been completely ship. waterproofed by the female bee. Noteworthy 2) Noncocoon-spinning larvae. The larva here is the fact that mature larvae of Neo- of Conanthalictus conanthi, as reported here, pasites (Rozen, 1966), Biastes (present study), does not spin a cocoon, a feature shared with and Townsendiella (Rozen and McGinley, all of the Andrenidae as well as with many 1991), nomadine cleptoparasites associated ofthe Melittidae. As stated above, this is not with these two groups, also have produced considered to be the result of sister-group re- venters of abdominal segment IX. Similarly, lationships of these taxa at this time, and mature larvae ofNeolarra (Rozen 1966) from associated larval similarities (loss of salivary Perdita cells (also unlined) have this modi- lips, reduced palpi, recessed labiomaxillary fication of the abdominal segment IX. Other region) likewise are convergences. larval Nomadinae do not have a ventrally 3) Paired tubercles arising from labral disc protruding abdominal segment IX. There is of mature larvae. This feature is a synapo- a possibility that this feature in the clepto- morphy ofthe Panurginae (McGinley, 1987) parasites is a synapomorphy of a restricted and is also found in Rophites trispinosus, pre- group rather than a character state that has sumably the result of convergence. arisen a number of times. Nonetheless, the 4) Abdominal segment X of larvae posi- correlation of larval host anatomy, larval tioned dorsally on segment X, anus more or parasite anatomy, and brood cell features is less dorsal on X, and abdominal segment IX suggestive that larvae ofvery different groups produced ventrally. This complex suite of have independently evolved similarities in characters is found in the Rophitinae and also response to the same environmental factors. in Hesperapis and Capicola of the Melittidae

REFERENCES Batra, S. W. T., and C. D. Michener ocorynura inflaticeps. Biotropica 11: 28- 1966. The nest and description of a new bee, 37. Systropha punjabensis from India (Hy- Eickwort, G. C., P. F. Kukuk, F. R. Wesley menoptera: Halictidae). J. Kansas En- 1986. The nesting biology of Dufourea no- tomol. Soc. 39: 650-558. vaeangliae (Hymenoptera: Halictidae) Eickwort, G. C. and the systematic position of the Du- 1969. Tribal positions of Western Hemi- foureinae based on behavior and de- sphere green sweat bees, with comments velopment. J. Kansas Entomol. Soc. 59: on their nest architecture (Hymenop- 103-120. tera: Halictidae). Ann. Entomol. Soc. Enslin, E. Am. 62: 652-660. 1921. Beitriige zur Kenntnis der Hymenop- Eickwort, G. C., and S. F. Sakagami teren. 1 Biologie von Rhophites canus 1979. A classification of nest architecture of Evers. Deutsche Entomol. Z. 1921: 59- bees in the tribe Augochlorini (Hyme- 65. noptera: Halictidae; Halictinae), with Hirashima, Y. descriptions ofa Brazilian nest ofRhin- 1961. Monographic study of the subfamily 1993 ROZEN: ROPHITINE BEES 27

Nomiinae of Japan (Hymenoptera, 1977. The ethology and systematic relation- Apoidea). Acta Hymenopterol. 1: 241- ships of fideliine bees, including a de- 303. scription of the mature larva of Paraf- Hurd, P. D., Jr. idelia (Hymenoptera, Apoidea). Am. 1979. In K. V. Krombein et al., Catalog of Mus. Novitates 2637: 15 pp. Hymenoptera in America North of 1984a. Nesting biology of diphaglossine bees Mexico, 2: 1741-2209. Washington, (Hymenoptera, Colletidae). Am. Mus. D.C.: Smithsonian Institution Press. Novitates 2786: 33 pp. Kukuk, P. F., G. C. Eickwort, and F. R. Wesley 1984b. Comparative nesting biology of the bee 1985. Mate-seeking behavior of Dufourea no- tribe Exomalopsini (Apoidea, Antho- vaeangliae (Hymenoptera: Halictidae: phoridae). Am. Mus. Novitates 2798: Dufoureinae): the effect of resource dis- 37 pp. tribution. J. Kansas Entomol. Soc. 58: 1987. Nesting biology and immature stages of 142-150. a new species in the bee genus Hesper- Malyshev, S. I. apis (Hymenoptera: Apoidea: Melitti- 1925a. The nesting habits ofspiral-horned bees dae: Dasypodinae). Am. Mus. Novi- of the genus Systropha Ltr. (Hymenop- tates 2887: 20 pp. tera, Apoidea). Rev. Russe Entomol. 19: 1989. Life history studies of the "primitive" 21-26 [in Russian, English summary]. panurgine bees (Hymenoptera: Andren- 1925b. The nesting habits of Rhophites Spin. idae: Panurginae). Am. Mus. Novitates (Hymenoptera, Apoidea). Rev. Russe 2962: 27 pp. Entomol. 19: 105-110 [in Russian, En- Rozen, J. G., Jr., and N. R. Jacobson glish summary]. 1980. Biology and immature stages ofMacro- 1935. The nesting habits of solitary bees. A pis nuda, including comparisons to re- comparative study. Eos, Riv. Espaniola lated bees (Apoidea: Melittidae). Am. Entomol. 11: 201-309. Mus. Novitates 2720: 11 pp. McGinley, R. J. Rozen, J. G., Jr., and R. J. McGinley 1981. Systematics of the Colletidae based on 1974. Phylogeny and systematics of Melitti- mature larvae with phenetic analysis of dae based on the mature larvae (Insecta, apoid larvae (Hymenoptera: Apoidea). Hymenoptera, Apoidea). Am. Mus. Univ. California Pub. Entomol. 91: 307 Novitates 2545: 31 pp. pp- 1976. Biology ofthe bee genus Conanthalictus 1987. In F. W. Stehr. Immature insects. Du- (Halictidae, Dufoureinae). Am. Mus. buque, Iowa: Kendall/Hunt Co., pp. Novitates 2602: 6 pp. 689-704. 1991. Biology and larvae ofthe cleptoparasitic Michener, C. D. bee Townsendiella pulchra and nesting 1979. Biogeography ofthe bees. Ann. Missou- biology of its host Hesperapis larreae ri Bot. Gard. 66: 277-347. (Hymenoptera: Apoidea). Am. Mus. Michener, C. D., and L. Greenberg Novitates 3005: 11 pp. 1985. The fate of the lacinia in the Halictidae Rozen, J. G., Jr., and C. D. Michener and Oxaeidae (Hymenoptera-Apo- 1988. Nests and immature stages of the bee idea). J. Kansas Entomol. Soc. 58: 137- Paratetrapedia swainsonae (Hymenop- 141. tera: Anthophoridae). Am. Mus. Nov- Popov, V. B. itates 2909, 13 pp. 1933. Notes on the parasitic bees allied to the Sakagami, S. F., and C. D. Michener genus Biastes Panz. (Hymenoptera, No- 1962. The nest architecture of the sweat bees madidae). Proc. Zool. Inst., USSR Acad. (Halictinae). A comparative study. Sci. 2: 51-75 [in Russian]. Lawrence: Univ. Kansas Press, 135 pp. Roig-Alsina, A. Snelling, R. R., and G. I. Stage 1991. Cladistic analysis of the Nomadinae s. MS. Systematics and biology of the bee ge- str. with description ofa new genus (Hy- nus Xeralictus (Hymenoptera: Halicti- menoptera: Anthophoridae). J. Kansas dae: Dufoureinae). Entomol. Soc. 64: 23-37. Stockhert, E. Rozen, J. G., Jr. 1922. Uber die Lebenweise von Rhophites 1966. The larvae of the Anthophoridae (Hy- 5-spinosa Spin. (Hym. Apid.). Deutsche menoptera, Apoidea). Part 2. The No- Entomol. Z. 1922: 381-392. madinae. Am. Mus. Novitates 2244: 38 Stephen, W. P., G. E. Bohart, and P. F. Torchio PP. 1969. The biology and external morphology 28 AMERICAN MUSEUM NOVITATES NO. 3066

ofbees, with a synopsis ofthe genera of parasite, Neopasites (Hymenoptera: northwestern America. Agric. Exp. Sta., Apoidea). J. New York Entomol. Soc. Oregon State Univ., Corvallis. 140 pp. 75: 132-146. Torchio, P. F. Torchio, P. F., G. E. Trostle, and D. J. Burdick 1989. Biology, immature development, and 1988. The nesting biology of Colletes kincaidii adaptive behavior of Stelis montana, a Cockerell (Hymenoptera: Colletidae) cleptoparasite of Osmia (Hymenoptera: and development ofits immature forms. Megachilidae). Ann. Entomol. Soc. Am. Ann. Entomol. Soc. Am. 81: 605-625. 82: 616-632. Torchio, P. F., J. G. Rozen, Jr., G. E. Bohart, and M. S. Favreau 1967. Biology of Dufourea and of its clepto-

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024.

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER.