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

AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2786, pp. 1-33, figs. 1-33 June 7, 1984

Nesting Biology of Diphaglossine Bees (Hymenoptera, Colletidae)

JEROME G. ROZEN, JR.' CONTENTS Abstract ...... 1 Introduction ...... 2 Acknowledgments. 2 Techniques ...... 2 Original Observations ...... 3 Ptiloglossa arizonensis Timberlake ...... 3 Ptiloglossa fulvopilosa Cameron ...... 12 Ptiloglossa jonesi Timberlake ...... 16 Ptiloglossa guinnae Roberts ...... 18 Crawfordapis luctuosa Smith ...... 20 Policana albopilosa (Spinola) ...... 20 Discussion ...... 23 Profile of the Biclogy of the Diphaglossinae ...... 30 Phylogenetic Considerations ...... 31 Literature Cited ...... 31

ABSTRACT

New data on the nesting biology ofthe following script accounts on these and other taxa throughout taxa of bees belonging to the Diphaglossinae are the entire geographic range of the subfamily. Spe- presented: Ptiloglossa arizonensis, fulvopilosa, cific subjects dealt with are habitat preference; nest jonesi, guinnae, Crawfordapis luctuosa, and Poli- site requirements; social organization; nest struc- cana albopilosa. This information is summarized ture, including cell orientation, shape, and size; and analyzed in relation to literature and manu- provisioning; oviposition; development; cocoon

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

Copyright American Museum of Natural History 1984 ISSN 0003-0082 / Price $3.00 2 AMERICAN MUSEUM NOVITATES NO. 2786 construction and fecal deposition; daily adult ac- phylogenetic relationships ofthe subfamily to oth- tivity; seasonal activity; and the relationship of er colletids are briefly discussed in light of nesting cleptoparasitic bees (Epeolini) to diphaglossines. biology, as well as the possible relationships be- A formal, synoptic overview ofthe biological fea- tween the Diphaglossinae and the Oxaeidae and tures of this subfamily is presented. Finally, the Stenotritidae.

INTRODUCTION This paper summarizes and analyzes ex- study. I also extend my appreciation to Mr. isting information on the nest ecology and Vincent Roth, Resident Director of the architecture, nesting behavior, including co- Southwestern Research Station, for his per- coon construction, nest site associates, and sonal hospitality and cooperation during the other aspects of the biology of the Dipha- studies on Ptiloglossa arizonensis. Dr. Fred- glossinae, a tropical and semitropical group erick D. Bennett (West Indian Station, Com- of large-bodied, New World, ground-nesting monwealth Institute of Biological Control, bees. Curepe, Trinidad, West Indies) assisted me The initial impetus for the study was the in locating and excavating nests ofPtiloglossa discovery of the nests of Ptiloglossa arizo- fulvopilosa and lent specimens from his col- nensis in Portal, Arizona, in August 1982. lection so that their identification could be The site yielded data that then permitted an determined. Dr. Philip F. Torchio (ARS Bee interpretation of notes I had made on the Biology & Systematics Laboratory, Utah State nests of Ptiloglossa fulvopilosa years before University, Logan) kindly permitted me to on Trinidad. Dr. Mont A. Cazier, Curator read his manuscript (referred to elsewhere in Emeritus, American Museum ofNatural His- this study) on the nesting biology of Hylaeus tory, immediately offered his field notes, pho- bisinuatus. tographs, and specimens of Ptiloglossa Mr. Ian Stupakoffhelped in numerous ways jonesi, nests ofwhich he had excavated in the in the laboratory study at the American Mu- early 1960s in Portal. Dr. Radclyffe B. Rob- seum of Natural History and is responsible erts's informative paper (1971) on Ptiloglossa for the scanning electron microscope pho- guinnae as well as cocoons he lent me af- tographs of the bee cocoons and cells pre- forded an even broader understanding of the sented here. Mr. Stupakofis participation was biology of the genus. Dr. Gard W. Otis and supported by the Undergraduate/Graduate his associates as well as Dr. David W. Roubik Research Program, which is funded by the and Dr. Charles D. Michener kindly contrib- Greenwall Foundation. I thank Ms. Deborah uted their manuscripts and specimens of the Hickman, who painstakingly prepared the cocoons of Crawfordapis luctuosa so that manuscript for publication. comparisons of that species could be made The Southwestern Research Station pro- with Ptiloglossa. In the National Museum of vided laboratory space and living accom- Natural History, Smithsonian Institution, modations for the initial phase of the study Washington, I discovered the cells and co- of the biology of Ptiloglossa arizonensis. coons of Policana albopilosa excavated by The following specialists have kindly read Claude-Joseph in Chile in the early 1920s; I and contributed valuable advice on the borrowed them through the courtesy of Dr. manuscript: Dr. Radclyffe B. Roberts, De- Karl V. Krombein. Because of the coopera- partment of Entomology and Economic Zo- tion of so many people, it was possible to ology, Cook College, Rutgers, the State Uni- consider the nesting biology ofthe subfamily versity of New Jersey; Dr. Charles D. as a whole. Michener, Department of Entomology, the University of Kansas, Lawrence. ACKNOWLEDGMENTS TECHNIQUES I am grateful to the persons mentioned above for the loan of specimens, manu- Most ofthe procedures used in the present scripts, and other helpful contributions to the study are familiar to bee workers who exca- 1 984 ROZEN: DIPHAGLOSSINE BEES 3

vate ground-nesting species, and therefore I leaving the hard cast intact. Noncritical parts discuss here only two matters pertaining to of the cast, as for example the pouring area, technique. can be flattened with a file and collection data (1) Even though these bees are large and can be printed on the smooth surface. These their nest components correspondingly large casts are preserved in the collections of the and easy to observe, I consistently used a high American Museum ofNatural History as are quality stereoscopic microscope to examine samples ofadult bees, bee cells, and cocoons. cells, cell linings, food contents, cocoons, and behavior ofvarious immatures ofPtiloglossa ORIGINAL OBSERVATIONS arizonensis as soon as they were excavated. Considerable detailed information would Ptiloglossa arizonensis Timberlake have been lost without immediate exami- nation of material. LOCALITY: This species was first observed (2) Many features of the cells of bees are flying at approximately 5 A.M. Mountain taxon-specific. These include not only the di- Standard Time, on August 10, 1982, by Vin- mensions and proportions, composition and cent Roth outside a window of his house in thickness of cell linings, cell walls, and types Portal, Arizona. Attracted by their loud buzz- of closures, but also shape. Differences in ing at that early hour, he collected four fe- shape may be obvious (e.g., the shape of the males. After I noticed them later in the col- cell tops of Ptiloglossa compared with that lection ofthe Southwestern Research Station, of Policana) or subtle. When subtle, they are he directed me to the site on August 29. I difficult to analyze as cavities. In recognition found several nests during the daytime, began of this problem Stupakoff investigated ways observations on August 31, and continued of casting cells so that their shape as a solid them through September 6, 1982. Nine nests could be more accurately studied, measured, were discovered where Roth had first seen and preserved for future comparisons. After the adults, and a larger concentration ofmore consultation with the Department of Verte- than 50 nests was centered approximately 2 brate Paleontology of the American Mu- m. away on the property of James F. Rich- seum, he tested the use of latex and Smooth- ards.2 on®, substances that have been employed for DESCRIPTION OF SITE: Both nest groups oc- casting fossils, but found that they took too cupied a terrace about 2 m. above Cave Creek, long to dry and could not be used in the field an intermittent stream, and were subjected where the soil was wet. These substances also to flooding under only the most extreme storm deteriorate after about five years of storage. conditions. An open riparian forest canopy Several brands of plaster were also tested: consisting primarily of sycamore and juniper Gladstone®' was very hard, but usually pro- permitted sunlight to reach the site during duced tiny air bubbles. Although somewhat much of the day, and the sky overhead was softer, Speedrock 3 proved to be the most sat- visible from all burrow entrances except for isfactory in that it showed few air bubbles a few at the edge of a large rock. Both aggre- and took approximately an hour to harden. gations (fig. 17) were in nearly horizontal Diagrams of the shape of cells of Ptiloglossa ground, with a moderately dense covering of arizonensis were made from camera lucida tall grass at the Roth site and with a sparse illustrations of casts prepared with Speed- covering ofgrass but a thick accumulation of rock®#D. The powdered material was mixed dried leaves at the Richards site. Most of the with water on a 1 to 2 basis and a darkening nest entrances were hidden by this dense pigment was added to better reveal details on ground cover. the surface ofthe cast. The pigmented plaster The soil contained numerous, mostly water- was added to the water and stirred slowly so rounded stones varying greatly from the size as to avoid air bubbles. The liquid mixture was tapped on a hard surface to remove ad- 2 The site was revisited on a number of days between ditional air bubbles, and transferred to the August 14 and September 2, 1983, after the manuscript cell lumen with a medicine dropper. After an had been submitted for publication. New information hour the cell wall was broken down in water discovered in 1983 is added as footnotes. 4 AMERICAN MUSEUM NOVITATES NO. 2786

900 I 5mm

NECK

FILTER

FLOOR

4 cm

4 5

FIGs. 1-5. Ptiloglossa arizonensis. 1. Cell, cell closure and lateral tunnel, side view. 2. Cell, egg, and provisions, side view. 3. Cell, cocoon and feces, right side, cut away. 4. Nest 1, August 31, 1982, side view. 5. Nest 6, September 2, 1982, side view. Scales refer to 1-3 and 4, 5, respectively. Drawing of cell made with camera lucida from cast of cell. ofa pebble to large rocks 25 cm. in diameter. rows, and perhaps the rather unpredictable The abundance of these stones may have ac- arrangement of brood cells. Between rocks, counted for the meandering tracks ofthe main the soil was moist, very dark, and heavy, often burrows, the varying diameters of the bur- containing organic material and numerous 1984 ROZEN: DIPHAGLOSSINE BEES 5 earthworms as well as smaller rocks. Many nearly vertically. Although the entrance was small roots and a number of large ones were open in almost every case, nests containing also encountered. The soil was easily com- adult females were plugged with soil 2.5 to 6 pacted and contained a great deal of heavy, cm. (10 measurements) below the surface, at very fine clay that formed the walls of the least when the female was in the nest (i.e., cells, as discussed below. during most of the day). Three ofthese plugs The nine nests on Roth's property were ranged in length from 3 to 20 mm. They tend- scattered within an area of 1 m. in diameter, ed to be flat and compacted below, and loose and some entrances were only 5 cm. apart, and unconsolidated above. The female, be- so that their tumuli were contiguous. The large low the plug, apparently constructed it by nest concentration on the Richards property tamping loose soil upward, probably with her stretched over a triangular area 2.5 by 2.5 by clypeus, but possibly with her pygidium. 1.5 m. I did not determine the exact number The overall nest configuration ofPtiloglos- ofnests here because I did not wish to disturb sa arizonensis was never completely re- the site by removing the ground cover. At vealed, but the plan was clearly a main tunnel least 12 nests were scattered in an area of descending downward in a twisting fashion one-quarter of a square meter. through the soil and around the roots and DESCRIPTION OF NESTS: I carefully exca- stones (figs. 4, 5), and cells randomly clus- vated eight nests (figs. 4, 5) on the Roth site tered around the burrow within a radius of to reveal overall nest structure, and quickly approximately 4 to 15 cm. It was impossible dug more than five nests clumped on the to determine where and in what directions Richards site for information on food con- closed lateral tunnels branched from the main tents, immature stages, parasites, cocoons, burrow. Open lateral tunnels approaching and cell structure. open cells rose approximately 1 cm. over a Almost all nest entrances were centered in distance of 2 to 3 cm. before bending sharply copious dark tumuli. Tumuli ranged in di- downward at the cell neck into the vertical ameter from 6 to 8 cm. when not close to one cell chamber. All laterals were filled after cell another, and varied in height from less than closure. 0.5 to 2.5 cm. When close to others, tumuli The more than 30 cells both ofcurrent and were often much higher, one being 4.5 cm. previous generations excavated at the Roth Composed of claylike soil, they became con- site varied in depth from 11 to 36 cm. Cells solidated when allowed to dry, even though in a single nest gave the impression of being they appeared to consist of loose fragments. nearly equal in depth, perhaps because fe- Active nest entrances had a diameter of 8 males took advantage of pockets of soil large mm. (five measurements), but, because of enough to accommodate cells. The order of twisting and turning, the tunnels below, cir- cell construction was not revealed, but clearly cular in cross section, varied in diameter from the older cells were not the lowest in the nest, place to place from 8 to 10 mm. Presumably nor the newest ones highest, as is the case the variation was created by the female's need with Colletes. Cells seemed to be not only co- to find a path around stones and roots. The equal in depth, but also grouped into small moist tunnel and tumulus walls often bore units of two or three cells near one another. the outline and striae ofthe female's pygidial An understanding ofcell arrangement will be plate. These demonstrated that the pygidium gained when nests are found removed from was used to compact the claylike soil of the one another so that the confusion of over- wall. Burrow walls were moderately rough in lapping nest elements is avoided, and when contrast to the very even surfaces ofthe cells. cells are found in a substrate with fewer stones When tested with water droplets the burrow and other inclusions. wall, though not truly waterproof, slowly ab- The cells (fig. 1) of Ptiloglossa arizonensis sorbed water. This slow absorption rate (dis- were unusual among bee cells in that the top cussed below) probably results from the fine (neck) ofthe cell, which was continuous with claylike soil rather than from a secretion ap- the rising lateral, bent downward more than plied by the female. 90 degrees so that the central axis ofthe main Main tunnels (figs. 4, 5) initially descended part of the cell was vertical. The lower end 6 AMERICAN MUSEUM NOVITATES NO. 2786 was narrowly rounded (as opposed to broadly could be carefully removed intact from the rounded) so that the maximum diameter was cell when the cell was softened by being sub- approximately halfway between the bottom merged in water. There can be little doubt point and the floor of the neck. Cells mea- that the lining and claylike wall protect the sured as in figure 1 varied from 20.0 to 23.0 contents of the cell from acquiring water or mm. (12 measurements in length) and from desiccating, as the case may be. 10.0 to 11.5 mm. (10 measurements) in max- After foraging and egg deposition, the fe- imum diameter. The diameter of the cell at male closed the cell. All closures were marked the level of the floor of the neck was 9 mm. by a thin, circular wad (fig. 1) of various in one case and the diameter of the neck at thicknesses of cotton-like material that was various places varied from 7.4 to 8.0 mm. usually stained amberish on the inside (cell (five measurements). In a few cases, the en- side) and that was loosely attached to the end trance neck seemed slightly wider than high; of the cell lining. Although the female may in other cases it seemed circular in cross sec- have provided this fibrous material before tion. she constructed the closure of soil, this ma- The cell wall was very smooth and fine- terial may possibly also have been mold hy- grained with little coarse material projecting phae that grew from the closure and neck. into the cell lumen. The even nature of the Whatever its origin, it was present in all cells wall, its waxy texture, and the fact that it was and clearly allowed air passage. Immediately retardant to water droplets suggested that the in front of this circular, fluffy wafer was the female impregnated it with a waxy secretion cell closure proper. In five or six cases the before she lined it. However, no waxy coat closure appeared as a vague, coarse spiral of could be detected, either with solvents or by two or three rings to the radius, with the out- heating, and I finally determined that the clay ermost ring the most consolidated. The inner in the soil gave the appearance and texture area was infirmly attached to the outer ring to the cell wall. Pieces of moist soil when and had vague fenestrations that extended to rubbed with the rounded surface of a pair of the somewhat denser fill ofthe lateral tunnel. forceps developed the same smooth, even The closure was more or less concave on the surface retardant to water that was charac- inside. teristic ofthe unlined cell wall. When heated PROVISIONING: Linsley (1962) and Linsley over a Bunsen burner, cell walls became per- and Cazier (1963, 1970) have discussed the manently hard like fired clay. Hence, I con- floral relationships ofthis species and ofother cluded that the female, after roughing out the members of the Caupolicanini. The obser- cell, smoothed it, perhaps by tamping with vation site for Linsley and Cazier (1970) was her pygidial plate, thus imparting smooth- less than a kilometer from the nesting site ness to the surface. Several cells showed py- that I studied.3 gidial imprints, particularly near the closure. Provisions in the cells containing eggs or After smoothing the wall, the female ap- early instars (fig. 2) were a watery fluid, nearly plied a lining ofnearly clear, nonfibrous, cel- clear on the top half and grading into a lower lophane-like material (figs. 1, 15) over the bottom half consisting mostly of yellow pol- entire inner surface, including the neck, as far len that made the fluid opaque. All such cells as the position of the future closure. This also had oily droplets floating on the surface lining consisted of at least three similar lay- of the liquid. ers, appressed so closely that their individual Foraging activities are reported in the sec- nature could be distinguished only when the lining was torn with forceps. One layer re- the torn could then be 3 Numerous males and females of Ptiloglossa arizo- vealed along edge nensis were discovered visiting Agave Palmeri with Apis peeled from the next. Together these layers mellifera and Bombus until 6:30 in the morning from adhered like one to the cell wall, there being Portal to Paradise in 1983. This plant, generally consid- no space between layers or between the lining ered to be bat-pollinated, has sufficient nectar early in and the wall itself, as is the case with some the morning to be visited by these bees as well as by Colletes. This lining, impervious to water, moths. It may be the source of the copious nectar in the was sufficiently thick and strong so that it Ptiloglossa cells. 1 984 ROZEN: DIPHAGLOSSINE BEES 7 tion on Daily Adult Activity, but method of to touch it. It bit it a number of times with food transport is of interest here. Pollen was its apically pigmented, elongated, sharply carried in a dry state, primarily on the scopa. pointed mandibles. The egg, though not The scopa in this genus was identified on punctured, was dented. Two cases of double pollen-laden females and consisted of hairs oviposition suggest that this may not be an that tended to be very plumose; patches of uncommon phenomenon. In yet another cell the hairs occurred on the hind tibiae and fem- an intermediate stage Ptiloglossa larva (per- ora, on the ventrolateral faces of the first haps third instar) was feeding on its side, as metasomal tergum, and on the posterior side described below, and the head capsule of a of each hind trochanter. On fully laden fe- dead first instar Triepeolus was partly sub- males, pollen also adhered to adjoining body merged in the provisions. These facts raise areas, and the dorsum of the metasoma was the interesting possibility that Ptiloglossa lar- ordinarily pollen-dusted so as to appear paler vae are able to defend themselves against the than that ofnonpollen foraging females. Nec- larvae of parasitic bees such as Triepeolus tar was carried in impressive quantities in the and that the selection strategy favoring this greatly expandable crop in the anterior part type of behavior involves sibling combat re- of the metasoma. sulting from double oviposition. Such a sit- DEVELOPMENT: Whitish, dull translucent uation would be the first recorded case of eggs (fig. 2) floated on the surface film of the defense against parasite eggs or larvae by the liquid provisions, probably because of the host first instars. Subsequent Ptiloglossa in- hydrofuge nature ofthe microscopically pap- stars also had apically slender and darkly pig- illate chorion. Elongate and slightly curved, mented and sclerotized mandibles, so that six eggs measured 3.9 to 4.6 mm. long and somewhat older larvae may also be able to 0.90 to 1.05 mm. in maximum diameter. The defend themselves. Young larvae of certain anterior and posterior ends were equally other solitary, ground-nesting bees have en- rounded so that they could be distinguished larged, pigmented mandibles (e.g., Pro- from one another only after the embryo was toxaea, Rozen, 1965b), suggesting that this observed. In each offour eggs the ventral side may not be a unique phenomenon. of the embryo was on or near the convex, Young and intermediate Ptiloglossa ari- dorsal surface of the egg. The embryo was zonensis larvae were aquatic in that they either on its back (two cases) or 45 degrees floated in the nearly clear, upper layer of the from being on its back (two cases). Torchio provisions, with most of their bodies sub- (Ms) has recently discussed embryo orienta- merged, just as do the larvae of Ptiloglossa tion in colletid bees. jonesi (fig. 7). When this stage larva was not One cell contained two Ptiloglossa eggs.4 I in motion, all spiracles on the upper side of assumed this to be an abnormality, for other the body and surrounding integument were cells seemed to contain single eggs and no cell exposed to the air, but when it dipped its head was found with two larvae. However, because downward, the anterior spiracles were sub- it was late in the season, relatively few cells merged. with eggs were encountered; most cells housed Feeding apparently took place as young and large larvae. On the final day ofthe field study intermediate larvae very actively moved the the last cell opened contained one very young submerged anterior part ofthe body dorsally larva ofPtiloglossa swimming on its side next and ventrally, that is, parallel to but under to its cast chorion, and a Ptiloglossa egg float- the surface of the provisions. Larvae occa- ed nearby. This cell, then, had also contained sionally ducked their heads into the thicker two eggs of Ptiloglossa. The larva, which pollen-bearing stratum of the provisions and turned out to be a second instar, as discussed appeared to stir pollen upward into the clear- below, moved its head as do larger feeding er stratum so that with the further swinging larvae and seized the egg when it happened of the anterior part of the body dorsally and ventrally, the mouthparts encountered the suspended pollen grains. The unusual atten- 4In 1983, seven cells with eggs were excavated and uate larval thorax of Ptiloglossa (fig. 7) may one of these contained two eggs. be a modification for ducking into the pollen 8 AMERICAN MUSEUM NOVITATES NO. 2786 layer of the provisions. The sharp-pointed The larva then reoriented so that its head mandibles of these larvae seem ill-suited for extended upward, and started cocoon con- ingesting the provisions, and brief observa- struction. tions did not demonstrate their usage in feed- The cocoon is complicated, and the se- ing. Further detailed observations on feeding quence of its construction was pieced togeth- mechanisms as well as mode of ecdysis are er from observation of cocoons being spun needed. as well as of cells of previous generations. At least five larval instars were identified The larva, reaching up toward the neck ofthe on the basis of anatomical features and there cell, first spun a nearly flat disk (fig. 3) of was a suggestion of yet one more. The first salivary silk, the cocoon operculum. When stadium occurred apparently entirely in the completed, this appeared as an even, rigid, chorion, because a cast skin was removed slightly concave reddish amber plastic-like from the chorion found in the cell containing surface with numerous small but distinct an active second instar and another Ptilo- round holes (figs. 13, 24, 25) suggesting a sieve. glossa egg (see above). The skin clearly showed This disk adhered to the lining near the top body segmentation, spiracles with attached of the cell and was neither horizontal nor tracheae and denticulate mandibles. I noticed vertical, but rather uniformly angled toward a similar situation with Svastra (Rozen, 1964), the cell closure in all cases. In early construc- and referred to the cast skin as an "embryonic tion, it appeared as a white, very fine trans- cuticle." Changes in the anatomy of instars lucent gauzelike material composed of fine most noticeably involve lengthening of palpi silk. The larva moved its head back and forth and the salivary spout, forward projection of over the fine netting and applied with its pro- the labiomaxillary region, and increase in jecting labial region strand after strand ofthis head width. The changes in these structures viscous silk. These strands coalesced to such from one instar to the next were not gradual an extent that they fused, leaving only the but progressed in discrete "jumps" so that, circular holes in the completed disk. The silk for example, the palpi of the last instar were was extruded from the peculiar spoutlike sal- elongate, those of the penultimate larva ivary opening of the larva, and the down- somewhat short, and those of the antipen- curved palpi as well as the tip of the spout ultimate even shorter. Body size increased in rubbed over the surface of the fabric as the proportion to food ingestion. Additional spinning proceeded. The taxonomically dis- specimens are required in order to document tinctive features of Ptiloglossa mouthparts fully the number of larval instars and the (McGinley, 1981) thus seem correlated with differences between them. the method of cocoon construction. There was no indication that the rate of One larva traversed the closure 11 times development of the immature stages was ab- in one minute when applying silk to the disk. normal in relation to other bees. As the nest- Each traverse consisted of several short lines, ing site was excavated rather late in the flying in slightly different directions. Ten opercula, season, one would expect to find many large essentially circular, ranged in diameter from predefecating and postdefecating larvae and 7.5 to 9.0 mm. After completing the disk, the comparatively few early instars, which in- larva then began cocoon wall construction deed was the case. This contrasts with the immediately below the operculum and ap- very slow development of such bees as Col- parently concomitantly spun a woolly, filter- letes thoracicus. like cushion of considerable thickness (fig. 3) Food consumption depleted all provisions immediately below the operculum. This filter except for the oily material discussed else- consisted of an upper part of very open, where, and afterward the now very large larva sheetlike pale silk more or less parallel to the almost completely filled the cell because of operculum but actually attaching to the cen- the massive amount ofliquid in its soft body. tral part of the operculum. No oil adhered to At the end offeeding, larvae presumably had the fibers nor were nematodes encountered their heads at the bottom of the cell and the here. Separating the upper part of the filter abdomen folded back on itself so that the from a denser mass ofwoolly material below anus was somewhat higher than the head. was a thin septum of tightly woven parch- 1984 ROZEN: DIPHAGLOSSINE BEES 9 ment-like silk. Strands of the dense woolly surviving Triepeolus larvae from the site material below it were heavily coated with metamorphosed at about the same time, but oil so that amberish oily deposits could be three more had not yet done so by the start seen clearly when the filter was examined up- of June. side down under a stereoscopic microscope DAILY CYCLE OF ACTIVITY: Observations with transmitted light (fig. 12). Numerous on daily adult activity were made on August nematodes also adhered to these strands. How 31, September 1, and September 3, 1982, the lower part of the filter became impreg- when astronomical sunrise ranged from 5:53 nated with oil and infested with nematodes to 5:55 A.M., Mountain Standard Time. The while the upper remained free is not under- eastern horizon was faintly light at 5 A.M. and stood. The lower surface of the completed bees could be seen at 5:30. All three days were filter, that is the chamber ceiling, was dome- mostly clear. On August 31 preliminary ob- shaped and its plane horizontal. servations indicated the bees became active A thin oily film covered the fibers of the around 5 A.M., came and went frequently for operculum as well as the inner walls of the about an hour, and then gradually reduced completed cocoon. The larva then extended their frequency of departures and arrivals, the parchment-like cocoon wall downward. and were almost inactive at about 7:10 A.M. The cocoon wall and the clear cell lining However, females occasionally arrived and formed a single sheet and were difficult if not departed even as late as 9:45 when I revisited impossible to separate. The combined co- the site. coon and lining were semiopaque and pale On September 1, starting at 4:45 A.M., I amberish except near the operculum where carried out further observations using a red they were somewhat darker. The cocoon (figs. electric light at first. Unfortunately, the light 3, 20) did not extend to the bottom of the caused some buzzing and piping (short bursts cell, but rather stopped at a line 3 to 4 mm. of vibrations at a higher pitch than buzzing) from the bottom. After the larva finished the in the leaves on the nest surface, and several cocoon wall, it destroyed most of the trans- females flew to it. This suggested that the red parent lining at the bottom 3 to 4 mm. ofthe light disturbed their behavior, and the pro- cell, leaving only flecks of the lining intact. cedure was discontinued. I heard the first fe- The cell wall there was peppered in every case male depart at 5:06 A.M. and a total of 32 by fine incisions, indicating that the larva used females left between then and 5:1 1. Fifteen the tips of its mandibles to scrape or peck more flew away between 5:11 and 5:16, and away at the lining. The larva then placed its an additional 13, between 5:16 and 5:21. The anal region over the bottom of the cell and first female was heard to return at 5:1 1 and discharged the entire yellow meconial mass by 5:21 so many females were returning and into this area that bore no cocoon and only leaving that further aural monitoring was im- flecks of cell lining. Subsequently, the larva possible. At 5:31 the same morning one or lowered its head and laid in a slightly concave more females, whether coming or going, was floor of thick, dark, parchment-like silk that buzzing at the site more than 75 percent of appeared continuous in structure with the co- the time. coon wall. Sealed from its meconial mass, the In order to obtain a sense ofdaily activity, larva then completed its activities and was I started at 5:45 A.M. to monitor by sight the now much smaller than the predefecating number of returning females over a 1 0-min- form. The destruction of the lining of the ute period, as follows: lower end of the cell presumably permitted 5:45-5:55 A.M. 16 females water from the feces to drain into the soil. 6:15-6:25 19 females All live postdefecating larvae examined five 6:45-6:55 10 females months after I excavated were still capable 7:15-7:25 4 females of moving slowly. Hence, there seems to be no true diapause. At the end of May 1983, These figures quantify the casual observa- an adult male and female emerged from the tions of August 31. Together with the aural last remaining live immatures taken to the observations before the bees could be seen, American Museum ofNatural History. Three the data coincide closely with the temporal 10 AMERICAN MUSEUM NOVITATES NO. 2786 distribution and the relative abundance of discharged droplets of oil-like material that bees on the pollen plant as recorded by Lins- floated to the surface in the dissecting dish. ley and Cazier (1970, table 1). The quantity was small by comparison with On September 3, 1 arrived at the site short- the impressively large quantity ofnectar. This ly before 5 A.M. The first signs ofbee activity, material may be the same as that observed as before, were piping and buzzing sounds on the surface of the stored provisions in the among the dead leaves. The first female de- brood cells. However, the nature, source, parted at 5:04. Seven females were captured function, and relationships of all oily sub- at random as they returned, all before 5:30; stances associated with the adult crops, larval none carried pollen, and when three of them body surfaces, cocoons, and provisions should were later dissected, all had crops filled to be carefully scrutinized in the future. capacity with clear liquid, presumably nectar, Because crops of dissected females con- as discussed below. The remaining four fe- tained almost no pollen and therefore sug- males, although not dissected, had wet faces gested that adults do not require pollen for when removed from the cyanide jar, indi- food, I hypothesized that females captured cating that they too had been replete with from the nest after discharging nectar into nectar. cells would have depleted crops.However, two Offive more returning females captured at females recovered from their nests in mid- random between 5:50 and 6:00 A.M., three afternoon had crops about half-full with nec- had scopae abundantly packed with pollen, tar. Further, the nectar was so filled with pol- and two possessed only small quantities of len that it appeared cloudy. As there was no pollen. One dissected bearer had a crop full general exodus of females from their nests ofnectar and the other females had wet faces, after 7 A.M. until these two were captured, indicating full crops. the cloudy contents of the crops suggest that A last sample of eight returning females, females may obtain food in the form of both gathered between 6:50 and 7:00, showed three nectar and pollen from still-open cells pro- females with only limited pollen and the oth- visioned earlier. Torchio (Ms) reported such ers without pollen. Three had wet faces, and a situation for Hylaeus bisinuatus Forster. one with a dry face, when dissected, had a Because my excavations took place over full crop. three or four days during daylight hours, it is These observations suggested that large clear that females did not generally depart on quantities of nectar but no pollen were gath- foraging trips during the main part ofthe day. ered on the first short foraging run of the Unfortunately, observations were not carried morning. Pollen and nectar were collected on out during the evening when Ptiloglossa jo- later trips, and still later forays were primarily nesi occasionally forages for nectar (Linsley for nectar, either because the pollen supply and Cazier, 1970).5 was running low or nectar was the main re- No males were observed at the nesting site, quirement. indicating either that mating takes place else- After the capture of females on September where or that males had already ceased their 3, an open cell in one nest contained only seasonal activity. All females collected had clear liquid. Ifthis cell belonged to one ofthe worn wings, as did the females of Triepeolus captured females, its discovery indicated that collected at the nests.6 cells were first supplied with nectar and then NEST AsSOCIATES: The cells of this species with pollen, which later sank to the bottom, forming the opaque yellow stratum of the 5 In 1983 the site was briefly visited several times in provisions. was inactive. whose were exam- the evening between 5:30 and 7:00 P.M. and Of five females crops 6 On August 14, 1983, the site was observed between ined, all had large quantities of clear nectar 5:40 and 6:30 A.M. Although this was earlier in the season that contained only a very small quantity of than the previous year's study, only a single male was pollen. When punctured, the crops of all dis- found for brief periods at the site, hovering and chasing charged liquid miscible with, but somewhat females. No copulations were seen. Most of the mating heavier than, water. This is presumed to be activity apparently takes place at the flowers and perhaps nectar. Crops of two of these females also along the flight path. 1984 ROZEN: DIPHAGLOSSINE BEES I1I of Ptiloglossa were heavily infested with of the provisions as revealed by photographs nematodes, probably representing a new ge- of this species in cells of Ptiloglossa jonesi nus in the superfamily Aphelenchoidea. They (fig. 23). The only live first instar ofthe Trie- seemed to be most abundant in cells where peolus in the nests of P. arizonensis was en- the bee larva had died, but there were scarcely countered on the side of the cell.8 When I any cells without a moderately heavy infes- tipped the cell so as to make the provisions tation. Nematodes could be seen on the sides flow toward it, it retreated up the wall, seem- ofthe cells, swimming through the provisions ingly to avoid the liquid. This larva was very where larvae were young, and crawling over active. Its dorsal surface constantly, rapidly the surface of mature larvae. Bodies of nem- quivered. It held its head, which also quiv- atodes were visible on cocoon opercula (figs. ered, away from the cell wall and its man- 14, 24). The impact ofmature nematodes on dibles were partly spread. When touched with the cell contents is unclear. Suzanne W. T. a pin, the head swung up and down as it tried Batra examined two female Ptiloglossa and to seize the pin. Whereas the head and an- found "no nematodes in the Dufour's glands, terior body were able to leave the wall, the guts or elsewhere" (in litt.). posterior body with the lateral extensions Several large adult Triepeolus belonging to could not be elevated, suggesting that these an undescribed species were collected flying lateral extensions, in addition to serving in over the nesting site, and numerous larvae of floating, may be stabilizing devices to hold various stages as well as eggs were recovered the larva in situ on the cell wall as it battles from cells. This species is the same as the one with the Ptiloglossa. Cazier (personal commun.) found infesting Last stage larvae of six Triepeolus were all cells of Ptiloglossa jonesi and whose mature found standing rigidly on end in the vertical larva was described by Rozen (1966) as Tri- cells with the anterior end curved forward, epeolus species b. All eggs were inserted most of the abdomen straight, and the anal through the cell lining into the cell wall, ap- area resting on the cell bottom (fig. 6). All proximately at right angles to the cell wall, were in complete diapause and the fecal ma- so that the anterior end of the egg was ap- terial, discharged as an opaque yellow semi- proximately flush with the cell lining. That liquid, was smeared irregularly over most of end was circular, nearly flat, and surrounded the cell wall. The greatest quantity of feces by a narrow rimlike flange. The seven eggs appeared to be heaped above the head of the or egg chorions discovered ranged from 8 to hibernating larva so as to restrict, although 15 mm. from the bottom of the cell. One egg in most cases not completely close off, the of Triepeolus measured 2.4 mm. long, its an- passage to the cell entrance. As is character- terior end was 0.7 mm. in diameter exclusive istic of other Nomadinae, this larva did not ofthe flange, and it tapered from there to the spin a cocoon, and in contrast to the larva of narrow posterior end. No cell had more than Ptiloglossa the cell lining was not destroyed. one Triepeolus egg. Hence, how the great quantity of liquid from First instars (fig. 23) with long, slender the cell provisions was removed is not under- mandibles, pigmented head capsule, elongate stood in the case of the cuckoo bee. The yel- labral tubercules, and very thin, elongate low integument of the larva, rigid and with maxillary palpi will be described in a sub- pronounced dark spiracular atria and tuber- sequent paper. Most of the body segments cules, contrasted with the flabby, whitish lar- had a pair of tapering, elongate tubercules vae ofthe host whose spiracles, though large, extending outward at right angles to the body, were unpigmented and therefore inconspic- and the apical abdominal segment possessed uous. As discussed above three of the post- a pair of eversible pygopod-like structures. defecating Triepeolus larvae brought to the The lateral body tubercules apparently assist American Museum metamorphosed by late in stabilizing the larva floating on the surface May 1983, more or less synchronously with several host larvae. 7Kindly identified by Dr. William R. Nickle, Nema- tology Laboratory, Plant Protection Institute, United 8 Numerous other first instars were seen floating on States Department of Agriculture, Beltsville, Maryland. provisions in 1983. 12 AMERICAN MUSEUM NOVITATES NO. 2786

Ptiloglossa fulvopilosa Cameron9 and had therefore been treated with an in- LOCALITY AND DESCRIPTION OF SITE: Ob- secticide. servations on the nesting habits and ontogeny The third nesting site, at Verdant Vale in of this species were made during 1964 and Arima Valley, was discovered on February 1965 at three different localities on Trinidad, 7, 1965, at which time 25 to 30 active nests the West Indies. Dr. Frederick D. Bennett, were seen in an area of 2 by 2.5 m. under a Commonwealth Institute of Biological Con- house (fig. 19). This site was in the approx- trol, Curepe, Trinidad, ably collaborated imate position of the one mentioned by Ve- in sey-FitzGerald (1939) in his discussions of gathering of data presented below. the genus. Both the second and third sites The first area, in Nariva Swamp, on the had been active for more than a generation, eastern side of Trinidad, consisted of a few as evidenced by vacated cells. burrows. Three entrances, surrounded by tu- All three areas were similar in that they muli and open on the surface, were situated were sheltered from the rain and sunlight did under a trunk of a fallen tree in a clearing. not reach them, except perhaps late in the The trunk was about 0.7 m. in diameter and afternoon. All occupied gently sloping ground 20 cm. above the ground. A single burrow that lacked vegetation. The one at Verdant was discovered in the earthen floor ofa near- Vale had an oblique half-meter rise on the by hut. ground surface, along which many but not all The second nesting site was under a house the nest entrances were found. Although in Curepe (fig. 18), where excavations were Trinidad is tropical, the microenvironment made on September 19, 1964, and again on of the sites was nearly xeric because of the February 6, 1965. Although about 10 en- shelter provided by the houses and the fallen trances were observed in an area of 4 sq. m. tree. The soil in each site was moderately on the latter date, all the bees were dead, easily excavated and contained some mois- presumably because the burrows had been ture; few hard inclusions were encountered. thought to be those of the parasol ant Atta, The soil, at least at Verdant Vale, was very pale, decomposed schistlike material, in con- 9 The correct name for this species is in doubt. Vesey- trast to the very dark, claylike soil associated FitzGerald (1939) and McGinley (1981) referred to it as fulvopilosa; Moure (personal commun.) believes that it with P. arizonensis. is lucernarum Cockerell. I compared a long series of At the second and third sites the burrow males and females from a number of localities from the openings were irregularly spaced, some being north range of Trinidad, all of which represent a single separated by only 4 or 5 cm. species. Males from this species differed from a male DESCRIPTION OF NESTS: Although most ob- Panamanian specimen offulvopilosa described by Mich- servations were made at Verdant Vale, those ener (1954) in that the pale hairs of the thorax and first from the other sites agreed. These observa- metasomal tergum were consistently more infuscated than tions were incomplete and therefore difficult those ofthe Panamanian specimen. Further, the narrow, to interpret at the time. The notes on P. ful- apical, pale bands of terga 2, 3, and 4 were somewhat vopilosa as well as the nest components stored more conspicuous than those of the Panamanian indi- American Museum now can be inter- vidual. Other aspects of the extemal anatomy were re- at the markably similar in the two samples. Trinidadian fe- preted more meaningfully because ofthe more males were compared with a single Panamanian female extensive investigations on P. arizonensis. of the lucernarum also referred to by Michener (1954). Each nest consisted of a nearly vertical, All Trinidadian females possessed narrow but distinct slightly meandering main tunnel between 9 apical bands on terga 2, 3, and 4, whereas the Pana- and 10 mm. in diameter and numerous singly manian female lacked such bands. Furthermore, the arranged vertical cells ranging from 2 to 6 ocellocular distance was distinctly shorter than the ocel- cm. from the main tunnel. Over most of its lar diameter in Trinidadian females, whereas in the Pan- surface, the main burrow seemed to have a amanian specimen it was somewhat greater. Hence, "plastered" wall of soil of variable thickness whether the Trinidadian population represents fulvopi- but 1 mm. thick. losa, lucernarum, or yet another species cannot be re- averaging approximately solved until a taxonomic revision ofthe genus is worked Whether it was constructed from soil pre- out, or at least until types are examined. Because ful- pared elsewhere or merely the result oftamp- vopilosa has been applied to the Trinidadian species in ing by the female is unknown, but it appar- print in the past, it is used here. ently did not have a counterpart in the nest 1 984 ROZEN: DIPHAGLOSSINE BEES 13

COCOON WALL

FECES

I5mm

PROVISIONS

orR -CELL WALL

LARVA 7 1 1 FIG. 6. Triepeolus, postdefecating larva in cell of Ptiloglossa arizonensis, side view. FIG. 7. Ptiloglossa jonesi, intermediate larva, partly submerged in provisions, showing part of body above surface in white, top view of cell. FIG. 8. Ptiloglossa fulvopilosa cocoon, top part, right side cut away. FIG. 9. Ptiloglossa guinnae, bottom of cocoon and fecal chamber, right side cut away. FIG. 10. Crawfordapis luctuosa, cocoon, right side cut away. FIG. 11. Policana albopilosa, cell and cocoon, right side cut away. Scale refers to figures 6-11. of P. arizonensis. The entrance of fresh bur- which in two cases measured 4 and 8 mm. rows was marked by a conspicuous tumulus thick. Below the plug, the actively worked (fig. 19) of loose soil 3 to 4 cm. high and 7 burrows were open and approximately 28 to to 8 cm. in diameter on a flat surface. The 30 cm. deep, although one nest extended tumulus soil next to the entrance was moist, down 42 cm. Each cell was connected to the consolidated, and continuous with the bur- main burrow by a lateral that usually rose row wall. About half of the burrow entrances about 3 cm. as it neared the cell. This side at Verdant Vale were plugged, either at or tunnel, about 9 to 10 mm. in diameter, had just below ground level, with a septum of soil a plastered coating, at least in most places 14 AMERICAN MUSEUM NOVITATES NO. 2786

kv.

FIGS. 12-16. Ptiloglossa arizonensis. 12. Photograph ofcocoon filter area taken with light transmitted through the filter as seen from below. Note oily droplets. Magnification approximately 6 x. 13. Scanning electron micrograph of cocoon operculum, seen from above. Magnification approximately 8 x. 14. Scanning electron micrograph of operculum, showing several nematodes. Magnification approximately 190 x. 15. Scanning electron micrograph of cell lining attached to cell wall as seen from inside of cell, showing nonfibrous nature of lining. Magnification approximately 150 x. 16. Scanning electron micrograph of cocoon wall, adhering to cell lining and cell wall taken from inside the cell, showing fibrous nature of cocoon fabric. Magnification approximately 150 x.

and, like the main tunnel, it was smooth with a distinct neck that angled downward toward some irregularities. the lateral. They were 11.5 to 13.0 mm. in The cells, similar in shape and size to those maximum diameter (four measurements) and ofP. arizonensis, were vertical and possessed 22.0 to 23.0 mm. in length measured from 1 984 ROZEN: DIPHAGLOSSINE BEES 15

1C 1; ,- tk 0

20 FIG. 20. Ptiloglossa arizonensis, bottom ofva- cated cocoon, side view, showing dark cocoon wall and cell lining above and pale dried fecal material at very bottom. Magnification approximately 4 x.

sis, cells possessed a plastered wall that varied in thickness roughly from 1 to 4 mm. The thickest places seemed to be both around the neck and at the cell bottom. The wall surface was coated with a semitransparent, extremely smooth, waterproof lining that was perhaps thinner than but in other ways similar to the lining in P. arizonensis. This lining extended into the end of the neck as in P. arizonensis. The somewhat obscure spiral cell closure consisted of several rings of slightly consolidated soil in front of which the side tunnel FIG. 17. Ptiloglossa arizonensis, view of nest- was filled with soil. Immediately inside sev- ing site in tall grass on both sides of barbed wire eral closures in the American Museum col- fence, Portal, Arizona. lection were small wads ofwhitish or grayish FIGS. 18, 19. Ptiloglossafulvopilosa. 18. View fibrous material, either hyphae or female se- of nest area under house, Curepe, Trinidad, being cretions, as in P. arizonensis. The order of examined by Dr. Frederick D. Bennett. 19. View cell construction was apparently not sequen- of nesting area under house in Verdant Vale, Ar- tial, for older larvae were found both above ima Valley, Trinidad, showing abundant, dry tu- ones. muli, some on sloping surfaces. and below younger PROVISIONING: The liquid provisions filled a little less than halfthe cell. Semitransparent the bottom to the level of the floor of the on the surface, they were less fluid and more neck (three measurements). The diameter of opaque at the bottom because of the yellow- the neck was approximately 8.0 mm. Neck ish pollen. The mixture was stringy and ap- floors in several cases formed a sharper angle peared to have the consistency of a broken with the cell wall below than did those of P. hen's egg, though not so yellow in color. In arizonensis. All the cell bottoms were nar- this respect they differed from the fluid, non- rowly rounded. Unlike those of P. arizonen- stringy stores ofP. arizonensis but agreed with 16 AMERICAN MUSEUM NOVITATES NO. 2786 the "ropey slime" in the cells of P. guinnae The banded fabric and the very bottom of (Roberts, 197 1). the cocoon adhered closely to the cell lining The elongate white egg, 4.7 mm. long (two just as did the cocoon material above. How- measurements), floated on the surface of the ever, the lower 3 to 4 mm. of the cell lining provisions. As is the case with P. arizonensis, had been perforated, presumably by the man- young larvae fed on their sides while partly dibles of the larva before the silk had been submerged in the provisions. applied. These fine incisions penetrated into After all provisions were consumed, the the soil and were often in linear sequence and larvae closed the cells with a medium brown other times randomly arranged. A drop of cocoon operculum (figs. 26, 27) just below water placed on the inside ofthe banded fab- the elbow. Circular and 9.8-10.0 mm. in di- ric slowly penetrated the combined cocoon ameter (three measurements), the operculum and cell lining. This indicated that the fabric slanted toward the cell closure. Its upper sur- here was not waterproof and that the punc- face was flat to slightly concave, dull, fine- ture marks made in the lining, though not textured, and semiopaque. Paler than the co- destroying the lining, permitted seepage of coon operculum of P. arizonensis, its texture water from the fecal material into the soil. was more fibrous and parchment-like, and After the lower section of the cocoon had the holes were more numerous and much been produced, the larvae deposited the me- smaller.'0 The filter beneath the operculum conial mass at the very bottom ofthe cell and was restricted primarily to the periphery of then spun a floor of parchment-like silk over the operculum (fig. 8). The lower surface of the feces as was characteristic of P. arizo- the filter, though fenestrated, was somewhat nensis. more consolidated than the "dome" of P. DAILY CYCLE OF ACTIVITY: Vesey-Fitz- arizonensis and was actually composed of Gerald (1939) reported that this species was several sheets of fibrous material. truly nocturnal in that females returned to The tan cocoon wall was appressed to the their nests "at the first light of dawn." cell lining so as to form a single sheet ofsemi- SEASONAL CYCLE: A collection of adults, transparent, thin, parchment-like material. kindly lent by Dr. Bennett, contained speci- The outline ofthe larva could be easily seen. mens taken during every month of the year This material extended down to the level 4 except April, November, and December. Pre- to 5 mm. from the bottom of the cell. Below sumably, this species is active throughout that level, the fibers were no longer fine and most if not all of the year. multidirectional, but rather were now thick, NEST ASSOCIATES: Although no cuckoo bees primarily parallel to one another, and grouped were encountered as adults or immatures at into bands, much as in Ptiloglossa guinnae Verdant Vale, Bennett recovered larvae of (fig. 9). These bands were concentric with the Odyneropsis apicalis Ducke from nests in the bottom point of the cell. Now the material Nariva Swamp (Rozen, 1966). Adults reared had a "rice-paper" effect, and not only looked from these larvae had enlarged ocelli, indi- like strapping tape, but also tore along lines cating that they too were nocturnal. Other ofweakness between the bands. This banded species of Odyneropsis in the American Mu- material extended down 2 to 3 mm. and then seum collection all have large ocelli so that gradually returned to a parchment-like fab- the genus may consist solely ofnocturnal bees. ric, with strands becoming multidirectional Triepeolus associated with P. arizonensis had at the very bottom of the cell. ocelli that were scarcely if at all enlarged. Indeed, the two adults that I captured were 10 These small holes were most numerous around the flying over the nesting site at a time when periphery where the ceiling of the living chamber was there was sufficient ambient light for me to not appressed to the operculum. In addition to these fine see them. It would be interesting to determine holes, two of the three preserved opercula possessed a bees flew at the few (maximum eight) much larger holes (three or four whether these cuckoo very times the diameter of the smaller holes) scattered in the early part of the daily flight period of P. ar- periphery. izonensis when there was scarcely any light. 1 984 ROZEN: DIPHAGLOSSINE BEES 17

4 P.

FIGS. 21-23. Ptiloglossa jonesi. 21. View of habitat, Portal, Arizona. 22. Egg floating on liquid provisions, top view. Magnification approximately 4 x. 23. Cell containing first instar of Triepeolus species b. Magnification approximately 5 x. Pho- tographed by Dr. Mont A. Cazier and Mr. Martin Mortenson.

Ptiloglossa jonesi Timberlake LOCALITY: Dr. Mont A. Cazier and Mr. M. Mortenson discovered the nest site of this species on July 24, 1964, on the property of Dr. R. G. Willys in Portal, Arizona. Infor- mation concerning this nesting site comes The ground surface was level to slightly slop- from a number of sources. First, Cazier (per- ing and had few stones, except for a low-lying sonal commun.) has kindly shared consid- retaining wall. The soil was a rich loam that erable data derived from notes and memory had been disturbed. The area was mostly un- about his study. Second, he has generously shaded during the daytime except for a few supplied numerous documentary photo- trees, and the low-lying ground cover con- graphs taken at the nesting site at the time of sisted of sparse herbaceous plants. Soil con- his investigations. These photographs, some tained numerous roots but far fewer stones of them reproduced here, reveal facts about than the soil at the site of Ptiloglossa arizo- nest structure, provisions, larval feeding hab- nensis. its, and various other matters. Third, Cazier DESCRIPTION OF NESTS: Nest entrances oc- during the course of his investigations de- curred in the open, and near, next to, or under posited fragments ofcells and cocoons in the such objects on the ground as rocks and small collections ofthe American Museum ofNat- plants. Tumuli apparently surrounded the ural History. Together, these resources now entrance hole as in Ptiloglossa arizonensis, permit a substantial understanding of the except when washed away by rain. Main bur- nesting biology of this species as compared rows, generally vertical, twisted considerably with that of Ptiloglossa arizonensis. and had the approximate diameter of those DESCRIPTION OF SITE: The nesting aggre- of P. arizonensis. gation, consisting ofapproximately 92 active Cells were essentially identical in shape with nests, occurred within an elongate area 3 by those of P. arizonensis. They were 23.0-25.0 5 m., all enclosed in a fenced yard (fig. 21). mm. in length (three measurements) mea- 18 AMERICAN MUSEUM NOVITATES NO. 2786 sured from the bottom to the floor of the the cell bottom after the cell lining at the neck; and 11.0-1 1.5 mm. in maximum di- bottom had been perforated (or destroyed), ameter (three measurements). The neck on presumably by the larva's mandibles. Tiny one specimen extended about 8.0 mm. to the incisions remained in the cell bottom. Also, closure and had a diameter of 8.0 mm. The just as in P. arizonensis, the larva had spun diameter of another cell at the level of the a thick, concave floor over the feces. neck floor was 9.0 mm. There was no indi- NEST AsSOCIATES: The large Triepeolus as- cation ofa "plastered" cell wall, and the wall sociated with the nest of P. jonesi was the was very smooth as is the case with P. ari- same species that attacks the nest of P. arizonensis. The cell lining was conspicuous as zonensis. Figure 23, kindly supplied by Ca- in P. arizonensis but at least on two speci- zier, shows its first instar floating on the pro- mens it became very thin on the neck surface visions, with its tapering, lateral body so that the neck wall was less waterproofthan tubercules, and large forked pygopod-like the main cell wall. structures everted from the last abdominal PROVISIONING: Cazier (personal commun.) segment. reported that the provisions of this species Preserved cells that had contained Trie- were a slightly sweet tasting liquid and ap- peolus revealed the same fecal pattern as the peared to be fermenting. Photographs (fig. 22) cells of Ptiloglossa arizonensis attacked by ofcells with eggs and larvae show that at least the Triepeolus. during early stages provisions are stratified into a lower, more opaque layer and an upper, Ptiloglossa guinnae Roberts clear watery one. Foraging ofthis species was Roberts (1971) provided considerable in- discussed by Linsley (1962) and Linsley and formation about the nesting, mating, forag- Cazier (1963, 1970). ing, and provisioning habits of this species, DEVELOPMENT: In shape, chorion texture, and kindly supplied a fragment of a single and probably size the eggs of this species ap- cocoon (fig. 9) with the fecal chamber at- peared identical with those of P. arizonensis. tached at the bottom. Although the opercu- They (fig. 22) floated on the surface film in lum was missing on the specimen, Roberts the middle of the cell, with the convex side (1971) fortunately pointed out that the oper- upward. Cases of more than one egg to a cell culum had "colander-like" perforations. This were not observed by Cazier (personal com- and the following information indicate that mun.). Young and intermediate larvae (fig. the cocoon is not markedly different from 7) floated partly submerged in liquid provi- those of other Ptiloglossa. sions, just as described for P. arizonensis. The fabric of the cocoon wall above the Fragments of several cocoons, partly as- fecal chamber was composed of a single thin sociated with cell walls, showed a circular, layer of very pale tan silk with multidirec- nearly flat operculum, 9.0 mm. in diameter tional fibers more fused than in the cocoon (figs. 28, 29). It bore numerous holes, that of other species so that the material was were not so evenly spaced as in P. arizonen- moderately shiny and transparent. Although sis. The fabric between the holes on the upper the fabric almost certainly adhered to the cell surface was not fused and plastic-like, as in lining, observations concerning this point P. arizonensis, but was softer and parchment- were not possible because the cell lining was like. The filter area on the cocoon consisted not preserved. The cell wall just above the of fine silk strands similar to those of P. ar- fecal chamber had silk strands that were izonensis, but further description and inter- somewhat more apparent than those farther pretation ofthe filter was impossible because above, and these were irregularly pigmented of damage. with brown, perhaps indicating a thicker ap- The cocoon wall did not extend to the bot- plication of silk. Just at the level ofthe floor, tom of the cell, at least in the two preserved the fibers of the cell wall became parallel to specimens. One cocoon terminated 3.5 mm. one another and concentric with the bottom from the cell bottom, much as in P. arizo- point of the cell, all as described for Ptilo- nensis, and the fecal material was packed into glossa fulvopilosa. Some of these parallel fi- 1984 ROZEN: DIPHAGLOSSINE BEES 19

'as > z * s* U

-J! .. 'O , -

FIGS. 24-29. Scanning electron micrographs ofdiphaglossine cocoon opercula, top view; pictures on left magnified approximately 30 x, on right, approximately 150 x. 24, 25. Ptiloglossa arizonensis. 26, 27. Ptiloglossa fulvopilosa. 28, 29. Ptiloglossa jonesi. bers were darker than others, so that the par- fiber pattern at the cocoon bottom. I could allel streaking of the cocoon of the fecal not determine whether the cocoon wall ofthe chamber was quite apparent, and the ten- fecal chamber was impervious to liquids, nor dency ofthe fabric to tear along parallel lines whether the cell wall had been perforated by of weakness was noted. The "rice-paper" ef- the larva. However, the similarity between fect extended downward for about 3 mm. and the fabric ofthe fecal chambers ofPtiloglossa gradually changed back to a multidirectional guinnae and fulvopilosa was so striking that 20 AMERICAN MUSEUM NOVITATES NO. 2786

I would not be surprised if the chamber and very large, reflecting the large size of the cell. its lining permitted drainage of fecal liquid Three measured 32 to 33 mm. long, from the into the substrate. center of the operculum to the bottom of the The fecal chamber (fig. 9), located at the cocoon, (i.e., bottom of the cell), and two rounded bottom ofthe vertical cell as typical measured 14.0 mm. in greatest diameter. for the subfamily, was 4.5 mm. at maximum They were similar in shape to those of Ptilo- depth and approximately 8.5 mm. in diam- glossa, that is, elongate and rounded on the eter at the top. The living chamber floor above bottom. The operculum angled obliquely in it was slightly convex, rather than concave, relation to the long axis of the cocoon. as are those of other known Diphaglossinae. The cocoon wall of semitransparent, tan, However, this shape may have resulted from parchment-like material resembled in texture the material being preserved in alcohol. that of Ptiloglossa and also was closely ap- Drying of the meconial mass in fecal cham- pressed to the cell lining. Unlike those of bers may cause the concavities observed in Ptiloglossa studied here, the walls extended other species. The floor was tan, smooth silk to the bottom of the cells and the nature of and turned up at the edges to meet the cell the wall fabric was the same throughout. From wall. this and from the large size ofthe fecal cham- (Smith) ber (fig. 10), I conclude that the larva did not Crawfordapis luctuosa perforate or destroy the cell lining, and that Several nesting sites of this species were the feces, very liquid in nature (Michener, studied by Otis et al. (1983) in Costa Rica personal commun.; field notes), were depos- and by Roubik and Michener (in press) in ited without any mechanism to reduce the Panama. Both teams of researchers kindly liquid content of the stored meconial mass. supplied specimens of cocoons for the fol- One can only assume that the soil, saturated lowing description. Roubik and Michener also by the moisture of the elfin forest environ- gave me a copy of their notes, as well as a ment, would not have absorbed such liquid manuscript which provided information on even if there had been no waterproof lining fecal material that could not have been de- at the bottom ofthe cell. Michener and Rou- termined by examining the preserved speci- bik (field notes) commented that the feces mens. were often distinctly stratified into an upper The upper surface of the operculum (figs. clear liquid layer and a lower equally thick, 30, 31) of this species was similar to that of opaque white or yellow layer (grading to Ptiloglossa arizonensis except it was medium brown above) of firmer material (pollen tan in color, 11.5 to 13.0 mm. (four mea- grains). The floor ofthe living chamber above surements) in diameter and the sieve holes consisted of a thick, concave layer of dark were somewhat smaller. Whereas the rigid, brown leathery silk, obviously applied by the plastic-like, slightly concave opercular sur- larva after defecation. It was similar to the face and its circular outline closely resembled cocoon floors found in Ptiloglossa. the same features in P. arizonensis, the filter Policana albopilosa (Spinola) area beneath was thinner as seen from the side than that of P. arizonensis, but thicker Claude-Joseph (1926)" treated the biology than that of P. fulvopilosa. In cross section of this species (as Policana herbsti Friese) in (fig. 10) it was composed ofoblique sheets of considerable detail. Presumably as a result of open netting, some of which attached to the this study he preserved larvae, pupae, and a undersurface of the operculum. The lower small clump ofearth containing four cells, all surface of the filter (i.e., the ceiling of the from Correo Nufnoa, Chile. The samples were living chamber) was a single dome-shaped deposited in the National Museum ofNatural sheet of denser, more opaque, parchment- History, and the larvae were described by like material that contained small openings Michener (1953) and McGinley (1981). The that were finer than those of the operculum. The presence of oily material was not de- " Claude-Joseph, an ecclesiastical name used by H. tected in any ofthe cocoons, all ofwhich have Janvier for his 1926 publication, was abandoned on his been preserved in alcohol. The cocoon was subsequent papers. 1984 ROZEN: DIPHAGLOSSINE BEES 21

i FIGS. 30-33. Scanning electron micrographs ofdiphaglossine cocoon opercula, top view; pictures on left, magnified approximately 30 x, on right, approximately 150 x. 30, 31. Crawfordapis luctuosa. 32, 33. Policana albopilosa. following account of the cells, cocoons, and One cell was clearly constructed partly within fecal material was possible because ofthe pre- another that had been previously constructed served cells. The soil was very fine-grained and lined. A wall of fine soil 0.3 mm. thick with no pebbles. separated the bottom of the one from the DESCRIPTION OF NESTS: The vertical cells other. There was no way of determining were elongate, rounded on the bottom, and, whether the first cell was from a previous in general, shaped as in figure 1 1. They ranged generation or whether the female for some in maximum diameter from 7.0 to 8.0 mm. reason abandoned the first cell and started a (four measurements) and the cell height in new one in the same place. one case was roughly 18 mm., as measured The side tunnel, 4.5 and 5.0 mm. in di- from the bottom of the cell to the approxi- ameter at the cell entrance (two measure- mate level of the closure. The cell wall was ments), curved away from the cell toward the very smooth except embossing from the fe- main tunnel (Claude-Joseph, 1926), but its male's pygidial plate was evident here and curvature was gentle (illustrated at about a there over the surface. The cocoon height in 45-degree angle by Claude-Joseph) as was also two cases was approximately 14.5 and 15.0 revealed on one exit tunnel partly preserved. mm., measured from cell bottom to the cen- This curvature was in sharp contrast to the ter of the operculum. Cells were lined with a more than 90-degree angle between the long transparent layer of waterproof, shiny ma- axis of the cell and the lateral of Ptiloglossa terial that extended to the cell closure, as and perhaps Crawfordapis. Claude-Joseph's characteristic of other known diphaglossines description ofthe spiral closure seems to cor- (that is, well above the cocoon operculum). respond closely with that of Ptiloglossa ari- 22 AMERICAN MUSEUM NOVITATES NO. 2786 zonensis, and indeed, the hardened outer spi- sisted of dull parchment-like, circular sheets -ral of the closure was still clearly evident in 5.5 and 6.5 mm. in diameter (two measure- two samples. ments). In addition to threads of silk embed- DEVELOPMENT: There appears to be a ma- ded in the matrix, there were small, extremely jor discrepancy between Claude-Joseph's ob- thin spots that glistened in reflected light, ap- servations on this species and mine on Ptilo- parently representing ribbon-like strands of glossa arizonensis (and for that matter, my silk as shown in figure 33. The mature larva interpretations of other species described of this species is peculiar for the Diphaglos- here). He attributed the construction of the sinae in that the salivary lips are flattened to operculum ("la membrane de fermeture"- form a small transverse slit that presumably the closing membrane) to the female's con- is capable offorming both threadlike and rib- structing this structure rather than to the bon-like silk (Michener, 1953; McGinley, spinning of the larva, as is unquestionably 1981). One operculum bore irregularly scat- the case in Ptiloglossa arizonensis and tered fenestrations ofvarious sizes over most jonesi. Indeed, he never discussed the pres- of the surface; the other had these holes ab- ence of the cocoon at all. Although the sim- sent in the center, which was about 2.0 by ilarities of the cocoon of Policana albopilosa 2.5 mm. Differences in the filters ofthese two and that of Ptiloglossa arizonensis are too cocoons, as described below, accounted for great for me to assume that they are products this dissimilarity. In both cocoons, the num- ofdifferent life stages in the two genera, some ber of holes was much fewer than in those of of his data are so detailed that there may be other taxa treated in this paper (see fig. 32). some basis in fact. The details are as follows: The opercula slanted somewhat in the direc- He describes how the female, after oviposition of the cell entrances. tion, positions herselfat the cell entrance and, The filter of one cocoon consisted of sev- with the lobes of the tongue closed together, eral well-formed sheets of netting that ex- constructs strands that support a viscous ma- tended partway across the upper end of the terial that forms "the membrane." He further cocoon and were loosely tied with silk strands asserts that during construction of the cell to the other sheets, to the operculum, or to closure, bits ofsoil drop onto the membrane. the dome-shaped lower surface of the filter. How he made these rather startling obser- The parchment-like lower surface bore nu- vations is not clear. He also gives details about merous randomly scattered small holes that how the female rubs the metasoma against tended to be smaller than the openings ofthe the cell wall in order to smooth the surface operculum. The filter on the other cocoon before she uses her tongue to disburse a vis- had the lower surface arched upward in the cous secretion over the wall. This is another center, so that it adhered to the operculum. remarkable observation that seems to cor- Where these two surfaces were contiguous, respond correctly to how the cell lining is there were no holes. Elsewhere the holes were applied in Colletes, as reported by Batra similar to those described for the other co- (1980). Additional observations need to be coons. The sheetlike netting was somewhat made on the cell lining and cocoon construc- similar to that of the other cocoon, except, tion in Policana albopilosa and, at that time, of course, it was not present in the central we can determine what role, ifany, the female area. has in the construction of the cocoon oper- The cocoon wall, like that of other di- culum. phaglossines, was transparent, parchment- The preserved cocoon operculum (figs. 32, like, brown, with numerous strands of silk, 33) was a nearly flat surface, slightly beveled all closely adhering to the cell lining. The wall downward just at the cell wall. A flange of (fig. I 1) extended downward, but terminated cocoon material extended up the cell wall before reaching the cell bottom. The line of from the attachment of the operculum and a termination seemed to be 2.0 to 3.5 mm. number of strands of silk were strung diag- from the bottom (three measurements), but onally upward from the operculum surface the exact line was hard to determine because either to the cell wall or to the flange. The the silk floor of the living chamber was dark, two opercula examined were tan and con- opaque, deeply concave, and overlapped the 1984 ROZEN: DIPHAGLOSSINE BEES 23 end of the cocoon wall almost 1 mm. so that et al., 1983; Roubik and Michener, in it was difficult to determine the lower ter- press minus of the wall. Most of the cell lining of the fecal chamber was clearly perforated by Tribe Diphaglossini: numerous tiny incisions, presumably made Diphaglossa gayi'2 Spinola - Janvier, by the larva's mandibles, although the lining 1933 in this area was not destroyed to the extent Cadeguala occidentalis12 (Haliday) - of that of Ptiloglossa arizonensis. Feces, yel- Claude-Joseph, 1926, as Policana oc- lowish in color, were deposited beneath the cidentalis floor. They contained some fibrous material Policana albopilosa12 (Spinola)-Claude- on their lower surface that may have been Joseph, 1926, as Policana herbsti hyphae rather than silk strands. Friese ECOLOGY: The Diphaglossinae are restricted to the New World. Their habitats range DISCUSSION from desert to near desert conditions on the The information presented above, when one hand and to wet tropical rain forest on incorporated with the literature accounts of the other. Some species (Crawfordapis luc- diphaglossine bees, demonstrates that the tuosa, Ptiloglossa guinnae, and apparently P. nesting biology of the subfamily is homoge- mexicana (Cresson) (Roubik and Michener, neous and contrasts to a considerable extent in press) have been found nesting under cool, with that ofother colletid subfamilies. In spite wet conditions of montane rain forests in ofsimilarities, differences in nesting behavior Central America. Ptiloglossa arizonensis and among taxa within the subfamily are also ev- P. jonesi, though in a desert, are active during ident. These similarities and differences, to- the summer rainy period and the claylike soil gether with my interpretation ofresearch areas of the site of P. arizonensis was moist. Ptilo- needing further study, are presented below. glossa fulvopilosa, from tropical Trinidad, Unless otherwise stated, literature citations nests in situations that are sheltered from the to species are as follows: rain so that the sites themselves are actually quite dry. Ecological information is not avail- Tribe Caupolicanini: able on the South American taxa discussed Caupolicana albiventris Friese-Jan- by Claude-Joseph (1926) and Janvier (1933, vier, 1955 1955). Caupolicana funebris Smith-Claude- All Diphaglossinae are ground nesting. Joseph, 1926; Janvier, 1955 Edaphic requirements for nesting seem var- Caupolicana gayi (Spinola)- Claude-Jo- ied, with some species apparently preferring seph, 1926 sandy situations, others claylike material, and Caupolicana gaullei Vachal-Janvier, no particular pattern of specificity emerges 1955 from limited data on hand. There may, how- Caupolicana ocellata Michener-Mich- ever, be some preference for slope ofthe nest- ener, 1966 ing surface. Cadeguala occidentalis, Policana Caupolicana pubescens Smith-Janvier, albopilosa, and some Caupolicana, appear to 1933 prefer either vertical banks or at least sloping Ptiloglossa arizonensis Timberlake- surfaces, whereas Ptiloglossa has been found present paper nesting primarily in horizontal surfaces. Ptiloglossa fulvopilosa Cameron-pres- Ground cover at nesting sites varies from ent paper; Vesey-FitzGerald, 1939 scattered herbs to trees; some sites seem to Ptiloglossa guinnae Roberts-Roberts, be exposed throughout the day to the sun, 1971; present paper whereas others (e.g., Ptiloglossa fulvopilosa) Ptiloglossa jonesi Timberlake-present are completely shaded for most of the day. paper SOCIAL ORGANIZATION: There is a distinct Crawfordapis luctuosa12 (Smith)-Otis tendency for individuals to nest gregariously in irregular groups of a few to many nests. 12 These genera are monotypic. Roberts (1971) counted 1348 nests at the site 24 AMERICAN MUSEUM NOVITATES NO. 2786 of Ptiloglossa guinnae, and Otis et al. (1983) Nests consist of a single entrance burrow found a maximum of400 nests at one site of without multiple entrances and without ves- Crawfordapis luctuosa, although the number tibules. Although it is difficult at times to fluctuated from season to season and year to determine the number of cells to a particular year. The Cadeguala occidentalis site had 200 nest because of nests grouped close to one burrows, as reported by Claude-Joseph another, all species have more than one cell (1926), who also found colonies of Caupo- and probably many cells to a nest. The nests licana funebris consisting of "hundreds of in horizontal surfaces are generally deep to galleries." Such groupings in diphaglossine very deep. Diphaglossa gayi, with cells rang- bees often persist for more than one gener- ing from 10 to 25 cm. in depth and Ptilo- ation, as evidenced by old, vacated cells glossa arizonensis, from 11 to 36 cm., appear among the cells containing immature bees, to be among the most shallowly nesting reported here for Ptiloglossa arizonensis and species, whereas Crawfordapis luctuosa has fulvopilosa. One of the sites of Ptiloglossa burrows descending as far as 120 cm. Nests fulvopilosa may have been the one studied by entering sloping or vertical surfaces may in Vesey-FitzGerald in 1939, in which case it some cases be shallower than those in hori- would have been in existence for 26 years. zontal surfaces; the works of Claude-Joseph Roubik and Michener's (in press) site of (1926) and Janvier (1933, 1955) are difficult Crawfordapis had been present for at least 25 to interpret in this regard, but may reflect the years, and Claude-Joseph (1926) claimed that situation. Otis et al. (1983) diagrammed the aggregations of Cadeguala occidentalis and main burrows of Crawfordapis luctuosa en- Caupolicanafunebris persisted for a number tering vertical surfaces as being short, and of years. It would not be surprising if long- those entering horizontal surfaces as being term nesting aggregations were characteristic long. of the entire subfamily. MAIN BURROWS AND LATERALS: Main gal- Evidence reported here and in other liter- leries are large in diameter, no doubt reflect- ature indicates that the Diphaglossinae are ing thle large to very large body size ofalmost solitary bees (nonsocial), and that there is but all species in the subfamily. These burrows a single female to a nest. However, Otis et were open in all reported cases, although a al. (1983) noted a strong tendency for marked plug of soil slightly below the ground surface females of Crawfordapis luctuosa to visit a has been characteristically noticed for Ptilo- number of nests. On the other hand Roubik glossa arizonensis, guinnae, and fulvopilosa, and Michener (in press), upon excavating when the females are in the burrows. Roubik nests ofthe same species during the rain when and Michener note that dead females often most bees would be in their nests, never re- stoppered the main burrows of Crawfordapis covered more than a single female in a bur- luctuosa, but found no other obstructions. row. Visitations to more than one nest by a Burrows are unlined by female secretion, and single female may in some way be related to only Ptiloglossa fulvopilosa apparently con- double ovipositions found in several cells of structs a wall of plaster-like soil in the main Ptiloglossa arizonensis. These matters war- burrows. Burrows of most taxa descend in an rant further investigation. irregular, meandering path that becomes ex- NEST STRUCTURE: A few generalities re- tremely sinuous where there are roots and garding overall nest structure of the Dipha- rocks in the substrate. Burrows appear to be glossinae can be presented. A copious tu- unbranched in all species of Ptiloglossa. Al- mulus, presumably always concentric, is though the same may be true for Crawford- reported for both Ptiloglossa and Crawfor- apis luctuosa as suggested by Roubik and dapis in situations where the species nest in Michener (Ms), the diagrams of Otis et al. horizontal surfaces, and such tumuli will (1983) suggest that either the burrow branched probably be found characteristic of the or the laterals branched, in which case the subfamily. Tumuli obviously erode rapidly distinction between main burrow and laterals under certain weather conditions such as the would have to be defined by whether the tun- high rainfall at the sites of Crawfordapis luc- nel is filled with soil after oviposition. tuosa and Ptiloglossa guinnae. In all taxa, laterals radiate from the main 1 984 ROZEN: DIPHAGLOSSINE BEES 25 burrow in various directions and each ends regarded the curved upper part of the cell as in a single cell. They are filled with soil after the downturned end ofthe lateral connecting oviposition. Apparent exceptions are Di- to the cell. Roberts (1971) studying Ptilo- phaglossa gayi, in which the laterals appar- glossa guinnae has regarded the cell as only ently are not filled according to Janvier (1933) the chamber just below the curved neck, and Cadeguala occidentalis, for which probably, at least in part, because he was un- Claude-Joseph (1926) diagrammed a nest able to define a spiral earthen cell closure. having a grapelike cluster ofcells with laterals However, such a closure is present in cells of connected to cell bottoms in some cases. Ad- Ptiloglossa arizonensis and fulvopilosa. The ditional studies ofthe nests ofthese two taxa spiral nature ofthe closure in these two species are required to permit an interpretation of is so indistinct that it was only on comparison their structures in relation to nests of other with a series that a consistent spiral shape diphaglossines. was revealed. The identity of the cell top of Laterals, in some but not all cases, rise be- the Ptiloglossa that I studied, as well as that fore connecting to cells. This is characteristic ofPolicana, is further confirmed by the pres- ofPtiloglossa in general and perhaps of Cau- ence of the cell lining extending up to, or policana pubescens but apparently not of nearly to, the earthen closure. Roberts dia- many of the other taxa. Roberts (197 1) sug- grammed such a lining for Ptiloglossa guin- gested that such a rise might serve to exclude nae. rain water from entering open laterals and Another confusion regarding the upper part cells. One might expect that such "sink-traps" of diphaglossine cells has been created be- would be found in nests of Crawfordapis luc- cause Claude-Joseph (1926) and Janvier tuosa because of the heavy rains in the high (1933, 1955) completely misinterpreted the rain forests, but the diagrams of Otis et al. cocoon operculum. He believed that in var- (1983) were ambiguous and observations by ious genera of South American diphaglos- Roubik and Michener (in press) were mixed sines the upper part of the cell was a silken on this matter. septum deposited by the mother bee after CELLS: Cell orientation for Ptiloglossa, ovipositioning and before constructing the Crawfordapis, Cadeguala, Diphaglossa, and earthen closure. He regarded the space be- Policana is vertical, as has also been recorded tween the operculum and the closure as an for all species of Caupolicana, with the single air chamber, large or small depending upon known exception of C. albiventris. According the species. Roberts and Roubik and Mich- to Janvier (1955) its cells are not kept per- ener have correctly recognized the operculum fectly vertical, but lean, the first to the left as the upper part of the cocoon. Not only do and the next to the right, so that no two con- my observations confirm theirs, but I was secutive cells have their axes parallel. able to observe mature larvae of Ptiloglossa So far as can be determined, the cells ofall arizonensis constructing opercula, and it was taxa are large to very large (corresponding to clear to me that opercula were not present in the large size of the bees), elongate, with a cells containing eggs, young larvae, or live diameter somewhat greater than the diameter Triepeolus larvae of any age. Claude-Joseph of the burrow, and circular in cross section. also confused the cocoon wall spun by the Those of Ptiloglossa are unique among bees larva with the cell lining provided by the fe- in that the top part of the cell bends 90 de- male; he never mentioned that any dipha- grees or more, which may also be true of glossine bees spun a cocoon. Crawfordapis luctuosa, as discussed above. The earthen cell walls ofPtiloglossa, Craw- The upper parts of cells of Caupolicana, Po- fordapis, Cadeguala, and Policana are very licana albopilosa, and Cadeguala occidental- smooth, and this will probably be character- is are bent, but at a lesser angle than those of istic of the subfamily in general. Among all Ptiloglossa. Curved to strongly curved cell taxa, only Ptiloglossa fulvopilosa is believed tops, then, seem to be a characteristic of the to construct a cell wall using mortar. The subfamily. meaning of this is unclear, but where it was The literature regarding this matter is con- noted, the substrate consisted of decom- fusing, however, because some authors have posed, schistlike material with a distinct grain, 26 AMERICAN MUSEUM NOVITATES NO. 2786 so that the plaster nature of the wall con- vide a secreted closure to their cells. Certain- trasted with the grain and was not therefore ly, the cell lining is not folded shut at the easily misinterpreted. mouth of the cells as it is in Colletes. In all known species, the female bee lines Cell closures appear to be more or less the cell wall with a clear to semitransparent poorly consolidated earthen spirals, at least lining that covers the entire cell surface. This in Ptiloglossa arizonensis, fulvopilosa, Cau- lining has sometimes been described as col- policana gaullei, albiventris, Cadeguala oc- orless, but Roubik and Michener (in press) cidentalis, and Policana albopilosa. Ptilo- referred to it as whitish in Crawfordapis luc- glossa guinnae and Crawfordapis luctuosa are tuosa and Roberts (1971), as tan in Ptilo- both reported not to have a special spiral glossa guinnae. This lining, probably in all earthen closure, although laterals leading to cases, adheres closely to the cell wall so that cells are filled with soil. Janvier (1933) stated there is no air space between it and the wall. that Diphaglossa gayi has neither an earthen Filmlike, it is multilayered, at least in Ptilo- closure nor a filled lateral. According to giossa arizonensis, and is a laminate of a Claude-Joseph (1926) and Janvier (1933), the number ofidentical films, which on fresh cells closures of Caupolicana gayi and pubescens can be teased apart only with effort. With consist of a series of flat plates, a description some species the total lining is sufficiently that is difficult to relate to other diphaglos- strong so that it can be removed intact while sines. still holding the liquid provisions. With most Roberts (1971) broached the subject ofgas species, however, it seems to be weaker, and exchange in the cells of Ptiloglossa guinnae, I found that it could best be examined intact especially with respect to carbon dioxide pro- by placing cell and substrate in a dish ofwater duced by yeasts in the provisions. He pointed and washing away the soil. The lining is out that the lack of a cell closure may have waterproof, an apparent necessity consider- evolved as a means of facilitating gas ex- ing the liquid nature of the larval provisions. change. Even in those Ptiloglossa where I de- This lining seems to be homologous with tected a cell closure, the middle part of the that of the cells of other colletid subfamilies, spiral is poorly formed, unconsolidated, and and probably is applied with the specialized sufficiently fenestrated that its gas exchange glossae characteristic of all colletid bees. capability probably would not be dimin- However, there are no air spaces between the ished. Gas exchange is also discussed below lining and the cell wall nor are its layers sep- in relation to the cocoon structure. arated by air spaces and fibrous strands, as Order of cell construction has been dis- seems to be typical of Colletes, Scrapter, Hy- cussed by several authors. Janvier (1933) laeinae, and Xeromelissinae. The lining of stated that the most recently constructed cells diphaglossine cells therefore lacks the glis- are the lowest in the nests ofDiphaglossa gayi tening, reflective appearance of these other and Caupolicana pubescens. Diagrams of the colletids and much more closely resembles nests of Policana albopilosa, Caupolicana the "varnished" cell surface of other families gayi, and funebris presented by Claude-Jo- with a conspicuous, nonwaxlike lining. A few seph (1926) indicate the same, as do illustra- examples of such linings in the American tions of the nests of Caupolicana gaullei and Museum collection include Protoxaea, An- albiventris provided by Janvier (1955). Otis drena, and Nomia. One wonders then what et al. (1983) indicated that the oldest cells of are the chemical differences and similarities Crawfordapis luctuosa are closest to the sur- of the cell linings of the Diphaglossinae, of face and that still open cells are farthest away. the rest of the Colletidae, and of those other On the other hand, Roberts (197 1) stated that families with nonwaxy linings. And what are "lateral branches are not constructed in any the phylogenetic implications? obvious vertical sequence" in Ptiloglossa With the possible exception of the circular guinnae, and my observations on Ptiloglossa wad of cotton-like material attached to the arizonensis and fulvopilosa suggest that the cell lining immediately in front ofthe earthen order is indeterminate, with cells of various closure of Ptiloglossa arizonensis, there is no ages at different levels in no apparent se- suggestion that female diphaglossines pro- quence. Hence, Ptiloglossa may differ from 1984 ROZEN: DIPHAGLOSSINE BEES 27 some of the other genera in this regard. Dia- of Caupolicana funebris, eggs are found on grams presented by Claude-Joseph (1926) the surface of the provisions near the center show a confusing picture of cells in tandem of the cell. Certainly in Ptiloglossa and in in Cadeguala occidentalis, old larvae being Caupolicana gayi they float on the provi- both higher and lower than young ones in a sions, but according to Claude-Joseph (1926) single series. and Janvier (1933, 1955) with some other PROVISIONS: A consistent feature of the species of Caupolicana, as well as with Ca- nesting biology of the Diphaglossinae is the deguala occidentalis and Policana albopilosa, liquid nature ofthe provisions, which occupy they rest on pollen that floats on or emerges approximately the lower half of the vertical from the surface of the provisions. Janvier cell. This has been noted with respect to all (1933) stated that the egg of Caupolicanafu- species ofPtiloglossa, Crawfordapis luctuosa, nebris is suspended from the "membranous many species of Caupolicana, Diphaglossa wall by a short filament," but the meaning of gayi, Cadeguala occidentalis, and Policana this is unclear. albopilosa. At least in Ptiloglossa, but pos- DEVELOPMENT: Aside from what little is sibly also in other taxa, the provisions are known of the development of Ptiloglossa ar- stratified initially into a nearly clear liquid izonensis presented here, almost nothing has (nectar) on top and an opaque mass of wet been reported elsewhere on the development pollen at the bottom ofthe cell. There is some or feeding activities of the larvae of dipha- indication in Ptiloglossa arizonensis that the glossine bees except that young larvae ofsome nectar may be disgorged by the female first, float on the surface of the provisions. This and then the pollen, dropped into the cell will probably be found to be characteristic of later, settles to the bottom. Floating fine this subfamily in general, because ofthe very droplets of oily material were observed in liquid nature of the early provisions. Ptiloglossa arizonensis but have not been re- COCOON AND FECES: Disregarding the con- ported for other taxa. The provisions change fusion in the works of Claude-Joseph (1926) over a period of time, presumably through and Janvier (1933, 1955) with respect to the fermentation in the case of Ptiloglossa guin- cell lining and cocoon, all diphaglossines spin nae. It is not clear whether the "soupy," "ro- cocoons. To the extent that the drawings and pey," "viscous," or "slimy" nature of the diagrams of Claude-Joseph (Janvier) can be provisions of some Ptiloglossa and of Craw- interpreted in light of the known cocoons of fordapis luctuosa are the result ofaging ofthe other diphaglossine taxa, shape and structure stored food (for example, through fermen- of cocoons of the subfamily are unique for tation) or whether this consistency is char- any group ofbees. No other colletid is known acteristic of newly stored provisions as well. to spin a cocoon, a fact reflected in larval Claude-Joseph (1926) and Janvier (1933, anatomy, in that only the Diphaglossinae have 1955) stated that with Caupolicana gaullei, salivary openings produced into projecting pubescens, and apparently albiventris, the lips (McGinley, 1981). pollen forms a pyramidal mass whose base The cocoon has a number of distinctive occupies the bottom of the cell while the top parts. The cocoon top, here called the oper- emerges like an islet from the liquid. The culum, is a nearly flat surface that spans the projecting pollen then becomes the oviposi- upper part ofthe cell. The nature ofthe fabric tion site. In Cadeguala occidentalis, Caupo- of the operculum varies from species to licana funebris, and apparently Policana al- species, as indicated in the first part of this bopilosa he reported that some pollen sinks paper. It is of some interest that the opercula to the bottom of the cell and the rest floats of Ptiloglossa arizonensis, Crawfordapis luc- on the surface serving as the oviposition site. tuosa, and Caupolicana gayi (Claude-Joseph, OVIPOSITION: Eggs of all species of dipha- 1926) are very similar, consisting of fused glossines, so far as we know, are elongate and, threads of silk forming a rather rigid, plastic- at least in the case of Ptiloglossa arizonensis, like fabric, whereas the opercula of other fulvopilosa, andjonesi are slightly curved, with Ptiloglossa are somewhat softer and more the anterior and posterior ends nearly iden- fabric-like. The opercula of all cocoons ex- tical in shape. With the reported exception amined by me, as well as of Caupolicana gayi, 28 AMERICAN MUSEUM NOVITATES NO. 2786 are clearly fenestrated with large or small the presence of the operculum, the filter be- holes. The opercula ofPtiloglossa, Crawford- neath it, and the cocoon walls, remains enig- apis, and Policana are tilted somewhat to- matic. A general presumption has been that ward the cell closure, so that they are not at cocoon-spinning is a primitive (plesiomorph- right angles to the long axis of the main part ic) feature because ofthe ability ofnumerous of the cell. Claude-Joseph (Janvier) implies wasps, a sister group to bees, also to spin that the opercula are horizontal in some cocoons. Why has group after group of bees species that he examined. lost the ability to spin such overwintering The cocoon immediately below the oper- chambers? One can only suppose that the culum, here termed the filter, is of variable operculum might function as a barrier to the thickness and composition, depending upon oviposition of mutillids and other parasites, the species and perhaps the genus (see above), that the filter beneath might also exclude such but often consists of vague sheets of silk ex- parasites, and that the thin but leathery co- tending from the walls of the cocoon toward coon wall has a similar function with respect the middle part of the operculum. Beneath to predators and parasites in the soil. How- this is a denser layer ofsoft woven fabric with ever, evidence bearing upon this matter is not obvious fenestrations. With some species, in- available, and indeed, the multiple origins of cluding Ptiloglossa arizonensis, there is yet a cocoon-less state in the phylogeny of bees another woolly layer beneath that. The bot- seems to suggest that cocoons are simply not tom of the filter is dome-shaped or at least that important. curved, and forms the ceiling ofthe chamber The cocoon wall ofdiphaglossines is a thin, containing the developing bee. Claude-Jo- pale to dark tan, single-layered structure that seph (1926) and Janvier (1933, 1955) de- adheres so closely to the cell lining that the scribed the filter for a number of species of two are nearly indistinguishable upon exca- Caupolicana as well as for Diphaglossa gayi. vation. Whereas the nature ofthe cocoon wall Although his descriptions are difficult to in- is consistent throughout the known dipha- terpret, they imply that interesting species- glossines, the lower termination of the wall specific features may be found among these is another matter that seems to have some taxa. phylogenetic and ecological significance, all Both Roberts (1971) and Claude-Joseph involved with fecal deposition. (1926) broached the matter of gas exchange In all diphaglossines so far studied, feces in the cells of diphaglossine bees. Roberts are deposited at the bottom of the cell: all pointed out that the openings in the oper- Ptiloglossa, Crawfordapis luctuosa, Caupo- culum may be important for gas exchange in licana funebris and pubescens, and Policana the living chamber of the cocoon. I am in- albopilosa. In all species whose cocoons I have clined to agree with this hypothesis in view examined (situation not known for Ptilo- of the consistent presence of fenestrations in glossa guinnae), with the single exception of the opercula and filters of all the species that Crawfordapis luctuosa, the larva perforates I have examined. It seems reasonable that the cell lining in the bottom 3 to 5 mm. of the cell wall (sometimes, as in Ptiloglossa ar- the cell before defecating with the result that izonensis, in claylike soil), the strong cell lin- the lining near the bottom is no longer water- ing applied by the mother, and the cocoon proof. Some species then construct the co- itselfmay provide a substantial barrier to gas coon wall only down to that level: Ptiloglossa exchange. Hence, the fenestrated cocoon top, arizonensis, Ptiloglossa jonesi, and Policana in conjunction with the loose fill ofthe lateral albopilosa. This seems to be the case for either and the imperfect or absent earthen cell clo- Caupolicana funebris or pubescens, or both, sure may be the only passage for gas. Clearly from photographs presented by Janvier the physics of this situation needs to be ad- (1933). Other species spin a parchment-like dressed. cocoon down to that level, but below it mod- Whereas the holes in the operculum and ify the texture of the cocoon fabric, making underlying filter can be discussed in terms of it permeable to liquid: Ptiloglossafulvopilosa their adaptive significance for gas exchange, and Ptiloglossa guinnae. Only in Crawford- the presence of the cocoon itself, including apis luctuosa is the cocoon wall homogeneous 1 984 ROZEN: DIPHAGLOSSINE BEES 29 from below the operculum to the very bottom has been recorded concerning times of mat- of the cell. In C. luctuosa the lower part of ing except for Roberts' observations (1971) the cocoon that houses the fecal material is that Ptiloglossa guinnae mates only in the much larger than that in the other species, morning. and the fecal material is more copious and SEASONAL ACTIVITY: There is little infor- semiliquid to liquid, in contrast to the dry mation that presents a meaningful pattern nature of the fecal material in the other di- regarding the seasonal activity of diphaglos- phaglossines that have been studied in this sine bees. So far as is known, Ptiloglossa ar- regard. In these other species, destruction of izonensis has but a single generation a year, the lower cell lining almost certainly allows and passes the winter as a postdefecating lar- for the drainage of the copious liquid asso- va that remains slightly active. Other di- ciated with the fecal material and therefore phaglossine bees in the northern and south- the fecal chamber is small. Hence, in the Di- ern temperate regions may also be univoltine. phaglossinae the absence or at least incom- In the tropics, forms, such as Ptiloglossa plete cocoon bottom seems to be the most guinnae and Crawfordapis luctuosa seem to widespread character state, and yet the com- fly throughout the year, although the activity plete cocoon bottom found in Crawfordapis ofnesting sites may wax and wane seasonally. luctuosa seems to compare with the complete Although most bees are generally assumed to cocoons found elsewhere among those bees have at least one generation a year, Janvier that spin cocoons. Which character state is (1933) reported that Caupolicana funebris specialized cannot be determined at this time. takes two years to develop, passing the first DAILY ACTIVITY: The Diphaglossinae are winter as a larva and the second as an adult mostly crepuscular or nocturnal bees, and that emerges the following summer. He also represent the only subfamily ofbees with that stated that Caupolicana pubescens usually has predominant tendency. Some authors, par- a two-year life cycle, with only those indi- ticularly Linsley (1962; see his bibliography viduals that are in the shallower cells emerg- for other such citations), Linsley and Cazier ing in a single year. These in the deeper cells (1970), Michener (1966), and Roberts (1971) pass the second winter in the adult stage and have delineated times of flight, both in the emerge in the spring. Claude-Joseph (1926) morning and in the evening, for various taxa. indicated that Cadeguala occidentalis over- Vesey-FitzGerald (1939), Schrottky (1907), winters in the adult stage and Policana al- Claude-Joseph (1926) and Janvier (1933, bopilosa, in the larval stage. 1955) have presented other flight data that, CLEPTOPARASITIC BEES: The only parasitic although not as precise as those of the above bees associated with the Diphaglossinae are authors, suggest nocturnal, early morning, and in the nomadine tribe Epeolini. The follow- late evening activity for South American ing are the associations: forms. Probably the large body size characteristic of the subfamily is functionally cor- Diphaglossine Parasitic Bee related with the ability of individuals to fly Ptiloglossa Triepeolus species b in cool morning hours. arizonensis (present paper) Caupolicana notabilis (Smith) from the Dominican Republic was collected near mid- Ptiloglossa Odyneropsis apicalis day (Michener, 1966) and therefore may be fulvopilosa Ducke (present paper) a diurnal species. Crawfordapis luctuosa (Otis Ptiloglossa Triepoelus species b et al., 1983, and Roubik and Michener, in jonesi (Rozen, 1966; press) is clearly active during the main part present paper) of the day. The cool, moist condition of its Ptiloglossa Odyneropsis gertschi high montane habitat probably approaches guinnae Michener (Roberts, the environmental factors associated with 1971) other diphaglossine bees that fly very early Caupolicana Doeringiella gigas in the morning or late in the evening. gayi (Spinola) (Claude- Although various authors have treated for- Joseph, 1926, as aging periods of these bees, almost nothing Epeolus gigas) 30 AMERICAN MUSEUM NOVITATES NO. 2786

PROFILE OF THE BIOLOGY OF tending evenly from bottom ofcell to closure; THE DIPHAGLOSSINAE foreign substances not brought into nest or used in cell construction. Cell closure usually The following summary, similar in format a spiral, often poorly formed in the middle, to that of fideliine bees (Rozen, 1977), is an sometimes apparently a series ofoverlapping attempt to provide a formal synoptic over- plates, other times apparently without special view of the biological features characterizing shape or consistency (Crawfordapis luctuosa, this subfamily. It is based primarily on in- Ptiloglossa guinnae). formation discussed above, but in addition PROVISIONING: Provisioning habits as yet incorporates data regarding flower prefer- poorly understood, but many taxa broadly ence, mating, and diurnal activity. (e.g., Crawfordapis luctuosa) or narrowly po- NESTING: New World, ground-nesting lylectic; stored provisions liquid, often ex- species inhabiting warm, temperate, semiarid tremely fluid, in some cases stratified into to moist tropical situations. Body size large upper clear layer and lower opaque, pollen- to very large, corresponding to large to very laden layer; provisions apparently changing large size of burrows and cells. Individuals during ontogeny of larva, in some cases nesting in loose, large, irregular aggregations; through fermentation, into soupy mixture, but nesting sites reoccupied over extended pe- nature and sequence of modification poorly riods ofyears, at least in some cases. Nesting understood. Pollen transported in dry state surface horizontal to vertical, in a variety of on female scopa. substrates. Females excavating own nests DEVELOPMENT: Moderately elongate, rather than using burrows of other insects, somewhat curved egg, one or occasionally two etc., but perhaps in some cases reusing main to a cell, floating on surface of provisions or, burrows from previous generations; appar- according to Claude-Joseph (1926) and Jan- ently only one female to a nest, but some vier (1933, 1955), resting on small islet of indication of females visiting more than one pollen emerging from or floating on provi- nest; nests, at least in horizontal surfaces, deep sions. Young larvae feeding while on side, to very deep. Tumuli dry, loose, concentric with most of body surface submerged; older on horizontal surfaces; turrets absent. Main larvae curled in cells while finishing food. burrow circular in cross section, unlined by After feeding but before defecation, larvae obvious secreted substances, and usually spinning elaborate cocoon consisting of up- without special built-in walls (exception, per flat plate (operculum) somewhat below Ptiloglossa fulvopilosa); vestibules and mul- the cell closure subtended by more or less tiple entrances absent; burrow path moder- thick woolly filter-like area; cocoon wall ex- ately meandering to extremely sinuous; bur- tending from operculum nearly to or to bot- rows open but in some species with plug of tom of cell, depending on taxon. Larvae of soil near surface; laterals, at least with most most taxa piercing cell lining at very bottom species, radiating from main burrow in var- ofcell, thereby allowing drainage of fecal liq- ious directions; their walls like those ofmain uid into soil (exception, Crawfordapis luc- burrows; each terminating in single cell, ex- tuosa); this destruction apparently taking cept perhaps for Cadeguala occidentalis; lat- place during cocoon construction. Cocoon erals not ending blindly; laterals filled after extending down approximately to top of oviposition except for Cadeguala occiden- pierced lining or, if extending to bottom of talis. Cells vertical or nearly so, elongate, nar- cell, nature of cocoon fabric at bottom ofcell rowly rounded below. Upper part of cells differing from cocoon wall above pierced lin- more (e.g., Ptiloglossa, Crawfordapis luctuo- ing, presumably allowing for drainage offecal sa) or less (Policana albopilosa) curved above; liquid. Feces deposited in bottom of cell fol- cell walls very smooth, usually without ma- lowed by spinning of floor to chamber con- soned surface (except Ptiloglossafulvopilosa); taining larva so as to separate living chamber cell lining conspicuous, cellophane-like; this from fecal chamber. Postdefecating larvae re- waterproof substance secreted by female bee; maining slightly active during hibernating lining adhering closely to cell wall and ex- period, at least in one species (Ptiloglossa ar- 1984 ROZEN: DIPHAGLOSSINE BEES 31 izonensis). Cocoon fabric, composed solely might be considered relatives of the Dipha- of silk (no foreign material), adhering inti- glossinae-the Oxaeidae (Rozen, 1965b) from mately to cell lining along sides of cell. the New World, and the Stenotritidae ADULT ACTIVITY: Bees, predominantly (McGinley, 1980) from Australia. Both fam- crepuscular (especially matinal) or nocturnal; ilies are composed ofrelatively few large-sized only Crawfordapis luctuosa certainly diurnal. bees that resemble, at least superficially, the Mating observed over nesting site in some Diphaglossinae. Biological information does species, in other species on flowers; territorial not seem to align either family with the di- behavior of males suggested in some species phaglossines. Roberts (1973) provided the but not in others. first major study of the nesting biology of a CLEPTOPARASITIC BEES: Only epeoline gen- member of the Oxaeidae, Oxaea flavescens era known nest parasites of some Diphaglos- (Klug). From his studies and also from my sinae; some species of Triepeolus, Doerin- own unpublished data on Protoxaea gloriosa giella, and Odyneropsis definitely associate Fox, the only possible synapomorphies be- with certain diphaglossine species. tween this family and the diphaglossines are the vertical cell and the fact that both groups PHYLOGENETIC CONSIDERATIONS are parasitized by epeolines. There are substantial differences between the two with re- Dr. Radclyffe B. Roberts (in litt.) raised spect to food consistency, cell shape, proba- several interesting questions after reviewing bly cell lining, and larval behavior not to this manuscript: (1) what evidence presented mention anatomical differences in both adults here suggests that the Diphaglossinae and and larvae. Houston (1975) presented an in- other colletids are monophyletic and (2) what teresting account ofthe biology of Stenotritus biological information points to a relation- pubescens (Smith), as he did more recently ship between the Diphaglossinae and major for the related Ctenocolletes (Houston, MS), groups of short-tongued bees other than the and Houston and Thorp (Ms), for Stenotritus Colletidae. greavesi. Numerous aspects of the nest ar- Relationship ofthe Diphaglossinae to Oth- chitecture, provisions, and larval behavior er Colletids: The inclusion of the diphaglos- reveal no synapomorphies with the Dipha- sines in the family must continue to be based glossinae. However, Stenotritus pubescens on adult and larval anatomy. Iftrue colletids, flies one-half hour before sun-up for about then the diphaglossines would appear to be three hours; hence, it is a matinal bee not a sister group ofthe rest ofthe family because unlike most diphaglossines. At least most of their cocoon-spinning habits, unique for species of Ctenocolletes are truly diurnal, as the Colletidae but a primitive feature in higher is Stenotritus greavesi. In sum, the relation- Hymenoptera. The distinctive and conspic- ships of the Stenotritidae and Oxaeidae to uous cell lining ofall colletids, including that other bees including the Diphaglossinae re- of the Diphaglossinae, is the main biological main obscure. synapomorphy uniting the group. This lining is associated with the characteristic glossae ofthe adults. However, there are obvious dif- LITERATURE CITED ferences with respect to the number oflayers in the lining, the structure ofthese layers, and Batra, Suzanne W. T. the cell closure involving the lining, from one 1980. Ecology, behavior, pheromones, para- subfamily to another. We do not know yet sites and management of the sympatric the nature and vernal bees Colletes inaequalis, C. tho- therefore the similarities and racicus and C. validus. Jour. Kansas Ent. differences of the chemical composition of Soc., 53(3), pp. 509-538, 14 figs. the linings of most of the colletid taxa. Claude-Joseph, F. Relationship of the Diphaglossinae to 1926. Recherches biologiques sur les Hym- Other Bee Taxa: The higher classification of enopteres du Chili (Melliferes). Ann. des the Apoidea has long been vexing. Most re- Sci. Nat., Zool., series 10, vol. 9, pp. cently, two families have been recognized that 113-268, 97 figs. 32 AMERICAN MUSEUM NOVITATES NO. 2786

Houston, Terry F. 1954. Bees of Panama. Bull. Amer. Mus. Nat. 1975. Nests, behavior and larvae of the bee Hist., vol. 104, pp. 1-176, 155 figs., 3 Stenotritus pubescens (Smith) and be- tables. havior of some related species (Hym- 1966. The classification ofthe Diphaglossinae enoptera: Apoidea: Stenotritinae). Aus- and North American species of the ge- tralian Ent. Soc., vol. 14, pp. 145-154, nus Caupolicana (Hymenoptera, Col- 15 figs. letidae). Univ. Kans. Sci. Bull., vol. [MS] Biological observations of bees in the 46(20), pp. 717-751. genus Ctenocolletes (Hymenoptera: Otis, Gard W., Ronald J. McGinley, Lyn Garling, Stenotritidae). 22 pp., 4 figs, 1 table. and Luis Malaret Houston, Terry F., and Robbin W. Thorp 1983. Biology and systematics ofthe bee genus [MS] Bionomics of the bee Stenotritus grea- Crawfordapis (Colletidae, Diphaglossi- vesi and ethological characteristics of nae), Psyche, vol. 89(3-4), (1982), pp. Stenotritidae (Hymenoptera). 279-296, 15 figs., 3 tables. Janvier, Hippolyte Roberts, Radclyffe B. 1933. Etude biologique de quelques Hym- 1971. Biology of the crepuscular bee Ptilo- enopteres du Chili. Ann. des Sci. Nat., glossa guinnae n. sp. with notes on as- Zool., series 10, vol. 16, pp. 209-346, sociated bees, mites, and yeasts. Jour. 57 figs. Kansas Ent. Soc., vol. 44(3), pp. 283- 1955. Le nid et la nidification chez quelques 294, 7 figs. abeilles des Andes tropicales. Ann. des 1973. Nest architecture and immature stages Sci. Nat., Zool., series 11, vol. 17, pp. ofthe bee Oxaeaflavescens and the sta- 311-349, 22 figs. tus of the Oxaeidae (Hymenoptera). Linsley, E. Gorton Ibid., vol. 46(4), pp. 437-446, 11 figs. 1962. The Colletid Ptiloglossa arizonensis Roubik, David W., and Charles D. Michener Timberlake, a matinal pollinator of [In press] Nesting biology of Crawfordapis in Solanum. Pan-Pacific Ent., vol. 38(2), Panama. Jour. Kansas Ent. Soc. pp. 75-82, 1 table. Rozen, Jerome G., Jr. Linsley, E. Gorton, and Mont A. Cazier 1964. The biology of Svastra obliqua obliqua 1963. Further observations on bees which take (Say), with a taxonomic description of pollen from plants of the genus Sola- its larvae (Apoidea, Anthophoridae). num (Hymenoptera: Apoidea). Pan-Pa- Amer. Mus. Novitates, no. 2170, pp. 1- cific Ent., vol. 39(1), pp. 1-18, 6 figs., 2 13, 15 figs. tables. 1965a. Phylogenetic-taxonomic significance of 1970. Some competitive relationships among last instar of Protoxaea gloriosa Fox, matinal and late afternoon foraging ac- with descriptions offirst and last instars tivities of Caupolicanine bees in south- (Hymenoptera: Apoidea). Jour. New eastern Arizona (Hymenoptera, Colle- York Ent. Soc., vol. 72, (1964), pp. 223- tidae). Jour. Kansas Ent. Soc., vol. 43(3), 230, 12 figs. pp. 251-261, 3 tables. 1965b. The biology and immature stages of McGinley, Ronald J. Melitturga clavicornis (Latreille) and of 1980. Glossal morphology of the Colletidae Sphecodes albilabris (Kirby) and the and recognition of the Stenotritidae at recognition of the Oxaeidae at the fam- the family level (Hymenoptera: Apoi- ily level (Hymenoptera, Apoidea). dea). Jour. Kansas Ent. Soc., vol. 53(3), Amer. Mus. Novitates, no. 2224, pp. 1- pp. 539-552, 31 figs. 18, 22 figs., 2 tables. 1981. Systematics of the Colletidae based on 1966. The larvae of the Anthophoridae (Hy- mature larvae with phenetic analysis of menoptera, Apoidea). Part 2. The No- Apoid larvae (Hymenoptera: Apoidea). madinae. Ibid., no. 2244, pp. 1-38, 83 Univ. ofCalif. Press, vol. 91, pp. 1-307, figs. 144 figs., 2 tables. 1977. The ethology and systematic relation- Michener, Charles D. ships of fideliine bees, including a de- 1953. Comparative morphological and sys- scription of the mature larva of Para- tematic studies of bee larvae with a key fidelia (Hymenoptera, Apoidea). Ibid., to the families of hymenopterous lar- no. 2637, pp. 1-15, 22 figs., 1 table. vae. Univ. Kans. Sci. Bull., vol. 53, pp. Schrottky, C. 987-1102, figs. 1-287. 1907. A contribution to the knowledge ofsome 1984 ROZEN: DIPHAGLOSSINE BEES 33

South American Hymenoptera, chiefly atus Forster and development ofits im- from Paraguay. Smithsonian Misc. Coll., mature forms (Hymenoptera: Colleti- vol. 48, pp. 259-274. dae). (For publication Jour. Kansas Ent. Timberlake, P. H. Soc.) 27 figs. 1946. Two new species ofPtiloglossa from Ar- Vesey-FitzGerald, D. izona (Hymenoptera, Apoidea). Pan- 1939. Observations on bees (Hymenoptera: Pacific Ent., vol. 22(4), pp. 156-158. Apoidea) in Trinidad, B.W.I. Proc. Torchio, Philip F. Royal Ent. Soc. of London, Series A, [MS] The nesting biology of Hylaeus bisinu- Gen. Ent., vol. 14, pp. 107-110.