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 3206, 27 pp., 47 figures August 29, 1997

South American Rophitine Bees (Hymenoptera: Halictidae: Rophitinae)

JEROME G. ROZEN, JR.1

CONTENTS Abstract ...... 1 Introduction ...... 2 Acknowledgments. 2 Phylogenetic Relationships and Classification ...... 3 Morphology ...... 6 Key to Genera of South American Rophitine Bees ...... 7 Ceblurgus Urban and Moure ...... 8 Goeletapis peruensis Rozen, new genus and species ...... 8 Penapis Michener ...... 18 Key to Species of Penapis ...... 21 Penapis penai Michener ...... 22 Penapis moldenkei Bohart, Toro, and Rozen, new species ...... 22 Penapis toroi Rozen, new species ...... 26 References ...... 27

ABSTRACT

Males and females of Goeletapis peruensis, era of South American Rophitinae, as well as with new genus, new species, are described from three other rophitine genera. Morphological evidence localities in Peru. The genus is compared with suggests that the South American genera represent Penapis and Ceblurgus, the other two known gen- a single clade with possible affinities to Old World

I Curator, Department of Entomology, American Museum of Natural History.

Copyright X American Museum of Natural History 1997 ISSN 0003-0082 / Price $3.00 2 AMERICAN MUSEUM NOVITATES NO. 3206 genera rather than North American taxa. A key to used for Goeletapis, and its female characters are the South American genera of the Rophitinae is recorded for the first time. Two new species in presented. addition to P. penai Michener are recognized: P. Specimens of Goeletapis were collected as they moldenkei Bohart, Toro, and Rozen and P. toroi foraged from (and almost certainly sought mates Rozen. These three species are quite similar ana- at) flowers of Exodeconus (Solanaceae) at a num- tomically, and their distinguishing features are il- ber of localities. Data are presented concerning lustrated. A key to them is offered, and their ovaries and oocytes. known distributions, all in Chile, are mapped. Penapis Michener is redescribed in the format

INTRODUCTION Although the Rophitinae (= Dufoureinae) Research Experience for Undergraduates are well known from the Holarctic and Af- Program, American Museum of Natural His- rica, their presence in South America was tory, who contributed to the manuscript by undetected until 1965, when Penapis penai searching for taxonomic characters of Goe- was described (Michener, 1965) based on a letapis and by preparing some of the illustra- single male specimen from Chile. Other tions. specimens of Penapis have been collected Dr. John L. Neff, Central Texas Melitto- through the years, but no new finds were re- logical Institute, Austin, Texas, freely offered ported until Ceblurgus longipalpis was de- his detailed notes and illustrations comparing scribed from Brazil more than a quarter of a Penapis with specimens of Goeletapis (see century later (Urban and Moure, 1993). Remarks, below) collected by Dr. Beryl Now, yet another rophitine genus and spe- Simpson from near Paijan, Peru. Dr. Simpson cies, here named Goeletapis peruensis Ro- was the first person to collect this genus, at zen, has been discovered in Peru. With this the PaijaLn CIPA-III Experimental Station, addition, the South American fauna of the Dept. La Libertad, July 10, 1983, on native subfamily consists of three genera, all from cotton (Gossypium barbadense). When he xeric regions. Goeletapis and Ceblurgus re- examined the specimens, Dr. Neff recognized main monotypic at this time, but two new that they represented a new genus and spe- Chilean species are herein recognized in cies and thus prepared his notes. The speci- Penapis. mens were subsequently sent to the USDA Bee Biology and Systematics Laboratory, ACKNOWLEDGMENTS Utah State University, Logan, Utah. Dr. Ter- ry L. Griswold and Dr. G. E. Bohart retrieved The field trip that led to the discovery of the single male from that series and loaned Goeletapis peruensis in 1995, as well as a it to me. The cooperation of all of these in- subsequent venture there in 1996, was made dividuals leaves little doubt that this early possible through the support of Mr. Robert collection was Goeletapis peruensis and led G. Goelet. On both trips, Sefior Alfredo to the rediscovery of the species near Paijan Ugarte Pefia provided invaluable field assis- on the 1996 trip. tance and participated in collecting the type I acknowledge the kind cooperation of series. I wish to thank Dr. Gerardo Lamas, Prof. Haroldo Toro, Universidad Catolica de Museo de Historia Natural, Lima, for his Valparaiso, Valparaiso, Chile, and Dr. Bo- counsel as to localities where bees might be hart. They shared their unpublished manu- active in Peru at the time of the trips. Dr. script on Penapis, they were the first to rec- Asuncion Cano Eschevarria, Museo de His- ognize Penapis moldenkei to be distinct, and toria Natural, Lima, kindly identified the host they permitted me to publish its description plant from the first trip. Host plants from the herein. Specimens of Penapis loaned by Drs. 1996 trip were identified by Dr. Jackie Kal- Bohart, Griswold, and Neff and by Prof. lunki, New York Botanical Garden, New Toro made possible the study and descrip- York. tions of the species of that genus. The rec- I thank Mr. Seth Budick, Participant, NSF ognition of these species provided evidence 1997 ROZEN: ROPHITINE BEES 3 of character variation within that genus, tures that appear to be synapomorphies, sug- which, in turn, was helpful in determining gesting that the South American rophitines generic boundaries between Penapis and represent a single clade. These features are Goeletapis. discussed as items 1 through 7 below. Dr. Charles D. Michener, University of 1. The anterior tentorial pits of all three Kansas, Lawrence, Kansas, kindly loaned a South American genera are situated adjacent male paratype of Ceblurgus longipalpis Ur- to the lower outer edge of the antennal sock- ban and Moure, and Dr. Danutncia Urban and ets (figs. 1-3). This character seems partic- Padre Jesus Santiago Moure, both of Uni- ularly strong because in other rophitines the versidade Federal do Parana, Brazil, loaned pits are generally far removed from the sock- a male and female of the same species col- ets on or just above a well-defined epistomal lected after the type series. On a visit to the suture. The position of the pits in the Old Snow Museum at the University of Kansas, World Systropha appears somewhat inter- I was able to study the holotype of Penapis mediate, a fact that may prove important penai and examine certain rophitine taxa not when the phylogenetic relationships within present in the collections of the American the subfamily are considered. Museum of Natural History. 2. The postpalpal part of the galea (galeal Versions of this manuscript were read by blade) (fig. 6) in the three genera is elongate, Drs. Bohart, Griswold, Michener, and Neff that is, equal in length to or somewhat longer and by Prof. Toro. Their comments and sug- than the stipes, whereas in most other New gestions were carefully considered and have World rophitine genera the galeal blade is improved the paper. distinctly shorter than (often less than one- half the length of) the stipes. Dufourea no- PHYLOGENETIC RELATIONSHIPS vaeangliae and D. versatilis rubriventris are AND CLASSIFICATION exceptions, however. An elongate galea is also characteristic of the Old World genera An analysis of the Rophitinae was con- Rophites and Morawitzia (but not Systro- ducted to determine the relationships among pha), another possible consideration when the three South American genera. Represen- the phylogeny of the subfamily is investigat- tatives of all but two rophitine genera (Trilia ed. A short galeal blade is characteristic of and Morawitzella) were studied. For the both the Nomiinae and Halictinae, possibly analysis, males of the following non-South indicating that the elongate condition is de- American taxa were examined: Dufourea rived, presumably a number of times, where marginata (Cresson), D. australis (Miche- it is found in the Rophitinae. ner), D. versatilis rubriventris Michener, D. 3. In contrast to most other rophitines, the novaeangliae (Robertson), Protodufourea three South American rophitine genera, as eickworti Bohart and Griswold, Sphecodo- well as Rophites, Morawitzia, and Systropha soma dicksoni (Timberlake), Conanthalictus have unusually long, gradually tapering glos- conanthi (Cockerell), C. deserticola Timber- sae (fig. 6), which are as long as, to much lake, Micralictoides quadriceps Bohart and longer than, the stipes. Glossae of Dufourea Griswold, Michenerula beameri Bohart, Xe- are shorter to longer than the stipes (Ebmer, ralictus bicuspidariae Snelling and Stage, 1984, 1993). The other rophitine genera have Rophites trispinosus Schummel, R. quinque- short glossae, which are usually less than spinosus Spinola, Systropha planidens Gi- one-half the stipital length (somewhat more raud, Systropha sp., Morawitzia panurgoides than one-half in Sphecodosoma). Friese, and M. fuscescens Firese. Head cap- 4. The maxillary palpi (fig. 6) are shorter sules of specimens of most of the species list- than the galeal blade in Goeletapis, Penapis, ed above were cleared in an aqueous solution Ceblurgus, and Rophites, whereas in most of sodium hydroxide and placed in glycerin other genera (including Morawitzella so that sutures, internal ridges, tentoria, and [Warncke, 1979: fig. 2], Dufourea novaean- mouthparts were clearly visible. gliae, and D. versatilis rubriventris, but ap- Goeletapis peruensis, Penapis, and Ceb- parently not including D. minutissima Ebmer lurgus longipalpis share certain unusual fea- [Ebmer, 1993: fig. 15]), the palpi are longer 4 AMERICAN MUSEUM NOVITATES NO. 3206

EPISTOMAL--' RIDGE

EPISTOMAL PLATE Q Q

I ~~~~~DEPRESS1ION

3 ATERIOR 0 ~~~~~~~~~~~TENTORIAL PIT

2 Q.0 ANTERIOR TENTORIAL PIT EPISTOMAL RIDGE -: . / \ EPvSTOMALRIDGE->(~SUBANTENNAL SUTURE EPISTOMAL PLATE OBLIQUE SUTURE EPISTOMAL SUTURE - Figs. 1-3. Heads of males of South American Rophitinae, frontal view. Shaded area to the left represents the smooth, glabrous area below the antenna. In figures 1 and 2, dashed lines depict the internal epistomal ridge and associated epistomal plate. 1. Goeletapis peruensis, n. sp. (type locality). 2. Penapis moldenkei, n. sp. (26 mi S Copiap6, Atacama Prov., Chile, Oct. 19, 1969 [Rozen and Pefia]). 3. Ceblurgus longipalpis Urban and Moure (paratype, Alagoinha, PE, Brasil, 23 March 1988 [I. C. S. Machado], in Cordia leucocephala). Scale (1.0 mm) refers to all figures. Labra flexed variably. than the blade. The maxillary palpi are also In Dufourea versatilis rubriventris, the blade shorter than the labial palpi in these same is perhaps somewhat less pigmented along four genera and in Systropha, which has a the outer side, but the blade beyond the base relatively short galeal blade. In Morawitzia is unusually thin, making this feature difficult the maxillary palpus is slightly shorter than to interpret. (Any study of the phylogeny of the galeal blade. the subfamily should include this species be- 5. The outer edge of the galeal blade of cause of its elongate mouthparts, unflattened Goeletapis, Penapis, and Ceblurgus is unpig- but dorsomedially membranous labial palpus, mented and unsclerotized, a condition par- and thin and perhaps membranous outer edge alleled somewhat by that of Rophites and of the galeal blade.) Morawitzia but not any other examined ro- 6. The labial palpi (fig. 6) of the South phitines (including Dufourea novaeangliae). American rophitines are flat, with the dorsal 1997 ROZEN: ROPHITINE BEES 5

surface membranous, at least for the basal in position. Systropha, unlike any other ro- three segments. In Rophites and Systropha phitines, is perplexing in that it has an ex- the basal segments are somewhat flattened, tremely long glossa and elongate labial palpi, but at least the first segment is not membra- but the galea is not elongate nor is its outer nous dorsally. It is uncertain, therefore, edge membranous. Dried specimens with the whether the flattened condition found in the proboscis extended show that the labial palpi South American genera is homologous with alone surround the glossa and thus create a that in the Old World genera. In Morawitzia channel for nectar ingestion. the first two segments are flattened and pre- The similarities in mouthpart structure and sumably membranous along the inner sur- function shared by these long-tongued ro- face, and the last two segments are not. The phitines and so-called long-tongued bees labial palpi of Dufourea novaeangliae and D. (Apidae and Megachilidae sensu Roig-Alsina versatilis rubiventris are not flattened. The and Michener, 1993) are interesting and de- dorsomedial surface of the palpus of the lat- serve further investigation. ter may be membranous, and that of D. no- Although the similarities shared by the vaeangliae is not only heavily sclerotized but three South American rophitine genera sug- bears stiff setae in the female that apparently gest that they comprise a distinct clade, the are involved with pollen gathering, as are relationships among the three genera are not similar setae on other parts of the proboscis clear. Adults of Goeletapis peruensis resem- (Eickwort et al., 1986). ble those of Penapis in size, coloration, and 7. The labial palpi (fig. 6) are extremely body form. They also share slender, elongate elongate in the South American rophitines, male genitalia (fig. 10) and a similar S7 (fig. Rophites, and Systropha: longer than the pre- 13), features that appear to be synapomor- mentum and sometimes as much as two phies and that are not characteristic of Ceb- times as long as the prementum. In other lurgus. On the other hand, Goeletapis and genera (including Morawitzella [Warncke, Ceblurgus agree in that both sexes display 1979: fig. 2]), the labial palpi are distinctly four-segmented maxillary palpi and a termi- shorter than the prementum except in Dufou- nal labial palpal segment that does not angle rea novaeangliae, D. versatilis rubiventris, from the palpal axis. Males of both genera and perhaps some Old World Dufourea (Eb- have a relatively long scape, short flagellum, mer, 1984, 1993), where they are subequal to and inner eye margins that diverge below. the prementum. Females of the two share branched scopal se- The elongate glossa, the flattened, sheath- tae on the hind tibia and basitarsus, a feature like labial palpi, and the elongate galeal uncommon in the rest of the Rophitinae. At blade of the South American rophitines and least some of these characteristics are de- Rophites probably function much as do these rived, and none is shared with Penapis. structures in long-tongued bees (Michener Finding a third distinctive rophitine entity and Brooks, 1984), as was already recog- in South America having possible synapo- nized by Urban and Moure (1993) for Pen- morphies with one or with both of the other apis and Ceblurgus. It seems likely that the two entities (genera) raises the question of membranous outer edge of the elongate galea what taxonomic level should be assigned to of these genera is also functionally involved them. I here propose that the new entity be with the labial mouthparts. These similarities accorded generic status. Although Goeletapis would seem to be derived and suggest that superficially resembles Penapis, and the sim- possible phylogenetic connections between ilarities of the genitalia and some of the male the South American and Old World rophitine sterna of Penapis and Goeletapis appear to faunas should be investigated, not just those be synapomorphies, there are striking differ- between the South American and North ences between the two genera in the anatomy American faunas. This is also suggested by of the other sterna. In addition, there are nu- the fact that no North American rophitine ex- merous other differences detailed in the de- amined in this study possessed anterior ten- scription below, which together indicate that torial pits adjacent to the antennal sockets, the phenotypic gap between these two genera but the pits of Systropha were intermediate (as well as between them and Ceblurgus) is 6 AMERICAN MUSEUM NOVITATES NO. 3206 as great as, or greater than, the gaps between nal area as found in the Andrenidae. Fur- some of the other rophitine genera. thermore, in Penapis (but not Goeletapis) the The three species of Penapis are very sim- smooth area is bordered mesally by a suture ilar to one another, and none shows any an- (fig. 2, subantennal suture) running from the atomical feature that can be interpreted as be- inner antennal socket to the epistomal suture. ing intermediate between the other two spe- On the outside, the area is bordered by the cies and Goeletapis: that is, Penapis is re- slitlike anterior tentorial pit (fig. 2). No ex- markably homogeneous and quite distinct ternal suture runs along the outer side of the from Goeletapis. It seems likely that, with smooth area, although internally the darkly further exploration of the South American pigmented anterolateral extension of the an- small-bee fauna, other species of Goeletapis terior tentorial arm curves here to fuse with and Ceblurgus will come to light. the lateral epistomal ridge. Michener (1965) All three genera could be relegated to a observed a fine suture (fig. 2, oblique suture) single genus, as separate subgenera. Such an in Penapis penai that extends obliquely arrangement would reflect the hypothesis that across the smooth area from the anterior ten- the three belong to a single South American torial pit inward to the faint but evident ep- clade. However, the distinctive differences istomal suture. He (personal commun.) inter- among these genera and the lack of knowl- preted it to be the outer suture. Neither the edge of their relationships to other taxa (such slitlike anterior tentorial pits nor the oblique as Rophites and Systropha) argue for a more suture that cuts across the area can qualify as neutral designation, albeit provisional. The homologous to the outer subantennal suture subfamily is interesting, and the appropriate of the Andrenidae because they extend only time to assign suitable rank to its components from near the tentorial pit to the epistomal is when relationships are analyzed broadly. suture and not from the antennal socket to the epistomal suture. The outer subantennal MORPHOLOGY suture of the Andrenidae extends from the outer edge of the antennal socket to the an- Michener (1965) in his description of Pen- terior tentorial pit, which is close to or on the apis noted the presence of "defined suban- epistomal suture. Confusion here has been tennal areas" but was uncertain "whether the the result of the unusually close approxima- 2 subantennal sutures [under each antennal tion of the anterior tentorial pits to the an- socket] are comparable to those of Andreni- tennal sockets. The smooth, glabrous inte- dae." He was, of course, constrained in his gumental areas should probably be consid- investigation because he had only a single ered a feature associated with the anterior specimen to study. More specimens have tentorial pits rather than with the antennal now been collected, and because of the rel- sockets because similar, though smaller, spe- evance of this problem to Goeletapis, the ce- cialized areas are associated with the anterior phalic exoskeletons of Penapis moldenkei tentorial pits in at least males of Systropha, and Goeletapis peruensis have been exam- Sphecodosoma, and Conanthalictus. ined in detail. This was undertaken in part to The oblique suture of Penapis apparently determine whether double subantennal su- is associated with the attachment of the me- tures of the Andrenidae were homologous to sal branch of the anterior tentorial arm to the those reported by Michener for Penapis and front of the head and seems to arise at the in part to understand the differences in the base of the articulatory process for the cardo, external appearance of this area between which ends just behind the face. This suture these two genera. Internal structures were ob- is not in evidence on the cleared head cap- served on head capsules cleared in an aque- sule of Goeletapis or on any of the other ro- ous solution of potassium hydroxide. Female phitines examined, suggesting that it is a sec- cephalic anatomy did not differ significantly ondary development apparently idiosyncratic from that of conspecific males. to Penapis (the cleared head capsule of Ceb- In both Penapis and Goeletapis, a rela- lurgus should be examined in this regard). tively smooth, glabrous area exists beneath Both Penapis and Goeletapis exhibit a fine each antennal socket, suggesting a subanten- but weak median epistomal suture that, in the 1997 ROZEN: ROPHITINE BEES 7 case of Goeletapis, becomes obscure in the cleared and uncleared specimens, there is a vicinity of the dorsolateral angles of the clyp- visible shallow integumental depression at the eus. In the case of Penapis, the median sec- lowest end of the dip in the epistomal ridge, tion of the epistomal suture is represented by at the point where the epistomal plate atta- a strong internal ridge that divides (or wid- ches. ens) when it approaches the antenna, with the In summary, the head capsule of Goele- dorsal branch curving toward the antennal tapis differs externally from that of Penapis suture. A fanlike extension of the anterior in the absence of the oblique suture, the ab- tentorial arm curves mesally and joins this sence of the subantennal suture, and the re- dorsal branch just before it meets the anten- duction of expression of the epistomal suture. nal suture. The subantennal suture is the ex- Internally, they are fundamentally the same, ternal evidence of the joining. The ventral primarily because of the unusual closeness of branch of the median epistomal ridge dips the anterior tentorial pits to the antennal downward and then curves toward an antero- sockets. lateral extension of the anterior tentorial arm As far as can be determined through ex- and then toward the anterior mandibular ar- amination of the uncleared head capsule of ticulation. This curving branch is accompa- Ceblurgus, this genus will probably be found nied by a fine external line that is the lateral to be similar internally to the other two gen- part of the epistomal suture. The large exter- era. Externally (fig. 3) the median transverse nal surface that the suture partly circum- section of the epistomal suture is clearly ev- scribes is the glabrous area beneath the an- ident and, just before it reaches the antennal tennal socket and has at its upper end the socket, it curves down and then up again be- anterior tentorial pit. Its integument in Pen- fore it extends laterally toward the base of apis is clear and less pigmented than the in- the mandible. The smooth, glabrous area be- tegument elsewhere, and the darkly pig- tween it, the antennal socket, and the anterior mented articulatory process for the cardo can tentorial pit is shorter than in the other two be seen just behind it on the cleared head genera, corresponding to the shorter, wider capsule. Near the bottom of the curvature, lower face of Ceblurgus. the ridge fans out into a broad, obliquely truncated, pigmented apodemelike structure, KEY TO GENERA OF SOUTH here termed the epistomal plate. AMERICAN ROPHITINE BEES In Goeletapis (fig. 1) the internal anatomy The generic key of the Rophitinae of the is somewhat similar even though the external Western Hemisphere (Michener, 1965) is no sutures are obscure in many places. The longer serviceable because of characteristics elongate anterior tentorial pit is situated close of Ceblurgus and Goeletapis.2 The three to the lower outer edge of the antennal socket, South American genera can be distinguished and the glabrous area lies ventral to it. The from North American forms on the basis of horizontal median section of the epistomal the following: forewing with three submar- ridge divides laterally, but the dorsal branch ginal cells; maxillary palpus shorter than gal- ends before it reaches the antennal socket and, eal blade; anterior tentorial pit adjacent to an- as a consequence, an external subantennal su- tennal socket (figs. 1-3); integument of ver- ture is not expressed. The dorsal branch is, tex and scutum punctate but not pebbled as however, met internally by the mesal exten- in Conanthalictus. None of the North Amer- sion of the anterior tentorial arm. The ventral ican genera possesses this combination of branch descends abruptly and then rises to characters. The key below separates the three meet the anterolateral extension of the anterior South American genera. tentorial arm. The darkly pigmented epistomal In some Conanthalictus, the anterior ten- plate is acutely pointed in this species and curves mesad. Although no external integu- 2 Persons using the keys in Michener (1965) and mental suture can be observed on an uncle- Michener et al. (1994) for identifying North Amercian ared specimen, a fine dark line is visible on rophitines should be aware that some males of Spheco- the cleared head capsule and probably is cor- dosoma have two submarginal cells and therefore will rectly called the epistomal suture. On both not key correctly. 8 AMERICAN MUSEUM NOVITATES NO. 3206 torial pits are rather close (but nowhere near basal tuft of long white hairs on dorsal sur- as close as in the South American forms) to face; T7 with conspicuous, narrow pygidial the antennal socket, about one-half socket di- plate with carinated sides and with only min- ameter away. This feature is presumably cre- ute setae; scopal setae of hind tibia with ated by the low placement of the sockets on dense, short branches (fig. 19) arising along the face rather than by upward migration of shaft. Ceblurgus the pits as in the South American genera. In other North American genera, the anterior CEBLURGUS URBAN AND MOURE tentorial pits are about one socket diameter Figures 3, 19-24 removed from the socket. Ceblurgus Urban and Moure, 1993: 102. Type- Hereafter, T = metasomal tergum and S = species Ceblurgus longipalpis Urban and Moure. metasomal sternum. Hence, T2 refers to the Monobasic. second metasomal tergum, S7 to the seventh Because this genus (and its only known metastomal sternum, etc. species) was recently described, it is not re- 1. Maxillary palpus 4-segmented; median, de- described here, but certain of its features are pressed, dorsal surface of propodeum vari- treated comparatively elsewhere in this pa- ously microscopically textured but never per. with radiating striae; labial palpus with axis of fourth segment continuous with that of Goeletapis peruensis Rozen, preceding segments; males with antennal new genus and species scape (exclusive of basal ball) longer than Figures 1, 4-17 distance between antennal sockets; males either without pygidial plate or with slender Goeletapis is monotypic as of this time; plate many times longer than basal width; its type is here designated to be G. peruensis midtibia of females with stout, curved, dark Rozen. Whether characters presented below or amber spinelike setae apically in addition are generic or specific is problematic. For to pale, thin, plumose setae; midtrochanter that reason and because a single description of female without ventral carina; scopal se- should be easier for the user, the following tae on hind tibia distinctly plumose .... 2 Maxillary palpus 6-segmented; median, de- account validates both genus and species. pressed area of dorsal surface of propodeum Characters thought to be diagnostic com- with striae radiating from base; labial palpus pared with other rophitine genera are itali- with fourth segment angled outward from cized; those features by which Goeletapis axis of preceding segments; males with an- differs from its most similar-appearing rela- tennal scape (exclusive of basal ball) shorter tive, Penapis, are in boldface type. In the de- than distance between antennal sockets; scription, extensive comparisons are made of males with large, broad, glabrous pygidial Goeletapis with Penapis and Ceblurgus. plate, basally as broad as length; midtibia of Male genitalia and terminal sterna of Ceb- females with only slender, plumose, pale setae; midtrochanter of female with ventral ca- lurgus, described by Urban and Moure rina; scopal setae mostly simple (but some (1993), were not examined for this study be- setae with one or more very short branches). cause too few specimens were available...... Penapis DIAGNOSIS: This genus can be distin- 2. Maxillary palpus tapering to fine point (fig. 7); guished from all North American and Old median depressed surface of propodeum World genera except for some Systropha by finely roughened, nonstriate; male mandible the placement of the anterior tentorial pits without basal tuft of long white hair on dor- adjacent to the outer lower edge of the an- sal surface; T7 of male covered by long se- tennal sockets (fig. 1). The postpalpal part of tae without distinct pygidial plate; scopal se- the galea of Goeletapis is longer than the stipes tae of female hind tibia with relatively few long branches arising apically (fig. 17). ... and the maxillary palpus is much shorter Goeletapis than the galeal blade (fig. 6), whereas in Sys- Maxillary palpus normal, not tapering to fine tropha the galeal blade is distinctly shorter point; median depressed area of dorsal sur- than the stipes and the maxillary palpus is face of propodeum with transverse, subpar- longer than the blade. allel striae; male mandible with conspicuous With respect to the South American Ro- 1997 ROZEN: ROPHITINE BEES

Figs. 4-9. Goeletapis peruensis, male. 4. Antenna. 5. Labrum, frontal view. 6. Proboscis, lateral view. 7. Maxillary palpus, lateral view. 8. Mentum and surrounding mouthparts, ventral view. 9. Left mandible and malar area, lateral view. Scale = 1.0 mm for figure 6 and 0.5 mm for all other figures. phitinae, males of Goeletapis lack a glabrous segments of the labial palpus. The dense, pygidial plate defined by ridges, whereas very short branches give these setae the ap- Penapis (Michener, 1965: fig. 20) and Ceb- pearance of being unusually thick. Compa- lurgus (Urban and Moure, 1993: fig. 5) have rable setae on Penapis and Goeletapis are such plates (which are very different in the shorter, fine, almost microscopic, and non- two genera). Females of Goeletapis, like plumose, at least under stereoscopic magni- those of Ceblurgus, have tibial scopas of fication. Numerous other features separating strongly branched setae, whereas scopal setae Goeletapis from Penapis and Ceblurgus are of Penapis (fig. 18) lack branches (or have a presented in the description below. few branches that are so obscure that setae DESCRIPTION: Male. Body length 5.8-8.7 appear simple). Whereas scopal hairs of Goe- mm (holotype 6.9 mm). Length of forewing letapis have comparatively few long branch- from costal sclerite to wing tip 4.0-5.5 mm es (fig. 17), those of Ceblurgus are densely (holotype 4.5 mm). covered with short branches (fig. 19). Fe- Integumental texture, color, vestiture. In- males of Ceblurgus have inner eye margins tegument nonmetallic in contrast to faint that diverge somewhat below (but not so greenish reflections on propodeum and ver- much as in conspecific males) and a maxil- tex of Ceblurgus and to more pronounced lary palpus that does not taper to a point (as metallic reflections on some species of Con- in Urban and Moure, 1993: fig. la). In fe- anthalictus and Dufourea; integument of male Goeletapis, the inner eye margins are head and mesosoma black, shiny, mostly subparallel, and the maxillary palpi taper to sparsely and finely punctured; upper middle a point (fig. 7). A unique, conspicuous fea- part of clypeus more sparsely and more fine- ture of both sexes of Ceblurgus is the densely punctured than comparable area of Pen- ly plumose, scattered, variable-length setae apis; dorsal surface of propodeum polished, (fig. 20) on the galea, stipes, and basal two except median depressed area more or less 10 AMERICAN MUSEUM NOVITATES NO. 3206

VOSELLA

17 BASAL PROCESS-

MULTISERRATE / MEDIAN EXPANSION- ~15_ MEDIAN SCLEROTIZED APICAL PROCESS APICALA L PROCESSE"i. 16 Figs. 10-16. Male genitalia and apical sterna, Goeletapis peruensis. Membranous areas shaded. 10. Genital capsule, ventral view on left, dorsal view on right. 11. Same, lateral view. 12. Apex of vosella, ventral view. 13. S5, left side, ventral view. 14. S6, left side, ventral view. 15. S7, left side, ventral view, with microscopic sculpturing of membranous area enlarged on left. 16. S8, ventral view on left, dorsal view on right. Scale = 1.0 mm; entire scale bar refers to figures 10 and 11; right segment of scale bar refers to figures 13-16. faintly roughened but never with fine radi- tarsi more or less ochraceous); integument of ating striae as in Penapis or with transverse metasoma black to very dark brown. Body striae as in Ceblurgus; undersurface of fla- pubescence consisting of (1) inconspicuous, gellum ochraceous; legs dark to medium fine, faintly brownish, recumbent setae that brownish, except foretibia and foretarsus of- are short and apparently nonplumose, and (2) ten more or less ochraceous (in Ceblurgus all conspicuous grayish white, erect setae that 1997 ROZEN: ROPHITINE BEES are plumose and moderate in length, but present on S2; on some specimens, patches only about half as long as those of Penapis on S4 tending to coalesce, forming marginal and Ceblurgus; erect setae sparse over most hair band; setal patterns of S5-S8 described of body, generally not dense enough to ob- below; gonoforceps (figs. 10, 11) bearing scure body surface. Pubescence on ventral scattered long setae at apex. surface of mandible (fig. 9) highly plumose Structure. Head breadth only slightly and organized as fringe; mandible without greater than length, not 1.4 times length as conspicuous basal tufts of long, white hair as in Ceblurgus. Antennal sockets far below found in Ceblurgus (Urban and Moure, middle of face (fig. 1). Anterior tentorial pits 1993: figs. 3, 4); long, erect setae of maxilla adjacent to outer lower edge of antennal and labial palpus fine, nonplumose, moderate sockets, much as in Penapis and Ceblurgus; in length, inconspicuous as in Penapis (in subantennal sutures not apparent; area be- contrast, those of Ceblurgus long to very low antenna invading clypeus on each side; long, densely minutely plumose, conspicu- these areas glabrous, rather smooth, not de- ous, as in fig. 20); anterior (ventral) surface fined by external epistomal suture, but clear- of forecoxa with only short setae, none of ly identified by broad internal epistomal which are more than half as long as coxa; ridge on cleared specimen (see Morphology, pubescence on outer surfaces of mid- and above, for interpretation and comparison hind tibiae and basitarsi semirecumbent, with Penapis). Clypeus moderately short, dense, arranged in orderly, parallel fash- length about one-third distance between low- ion, moderate in length, strongly plumose, er inner orbits in frontal view, but not very partly obscuring integument (in Penapis, short as in Ceblurgus (fig. 3); median part of setae on midtibia long, semierect, weakly epistomal suture (fig. 1) evident externally as plumose, not organized in parallel but fine integumental line somewhat below level dense so as to partly obscure integument; of antennal sockets but lateral parts not de- on hind tibia similar to midtibia but less fined externally (but see Morphology, dense; in Ceblurgus, setae on midtibia short, above); frontal carina absent; facial fovea ab- semi-erect, weakly plumose, not organized in sent; malar space virtually absent, with an- parallel, sparse, not obscuring integument; on terior mandibular articulation only slightly hind tibia, setae extremely elongate, strongly farther removed from eye than posterior ar- erect, not obscuring integument); midcoxa ticulation (fig. 9); vertex produced well without inner apical tuft of white setae that above upper ends of eyes. Inner margins of are nearly as long as trochanter, as found in eyes diverging slightly below (fig. 1), not Ceblurgus; midfemur with small basal diverging as much as those of Ceblurgus patch of appressed golden setae on ante- (fig. 3). Labrum (fig. 5) broader than long, rior surface, these setae shorter and less its width about 1.5 times maximum length, plumose than longer white setae on rest of hence distinctly longer than that of Pen- segment (golden setae absent in Penapis apis and Ceblurgus; anterior surface essen- and Ceblurgus); hind trochanter without out- tially flat at base but bending at about one- er and more mesial specialized, long, thick, third way to apex so that its apical surface curved, apical setae as in Ceblurgus. Meta- in different plane; at junction, two surfaces somal pubescence not forming bands on forming low transverse ridge from which T1-T5; premarginal and marginal areas of irregular row of short simple setae arises; T2-T5 with very fine recumbent setae row of long, tapering, mostly simple setae arising from very fine uniform punctures; arising from apex. Mandible elongate, about some punctures on T4, T5 also larger; T5 as long as eye seen laterally, with preapical with tuft of elongate, plumose setae at api- inner tooth, without peculiar modifications of colateral angles; T6 with rather dense, long, Ceblurgus (Urban and Moure, 1993: figs 3, plumose apical hairs, generally covering T7; 4). Proboscis (fig. 6) elongate, in repose T7 completely pilose dorsally with apical reaching between forecoxae; glossa (fig. 6) hairs long; S3 and S4 with sublateral mar- very long; annular hairs of glossa simple, ginal patches of recumbent whitish plu- acutely pointed, elongate (like those of Pen- mose setae that are sometimes also vaguely apis, longer than those of Ceblurgus; more 12 AMERICAN MUSEUM NOVITATES NO. 3206

distal ones of Ceblurgus tapering to flat, par- Scape (fig. 4) moderately long and fla- allel-sided blade that abruptly narrows api- gellum short, so that length of scape, ex- cally to end in simple or bifid sharp-pointed clusive of basal ball, equal to or longer apex); paraglossa small, linear; mentum a than combined length of first five flagel- small, triangular sclerite (fig. 8) with basal lomeres (about as in Ceblurgus); in con- point elongate; lorum (fig. 8) membranous trast, scape of Penapis short and flagellum except for elongate sclerite on each side; fla- elongate, so that scape shorter than first bellum absent; labial palpus (fig. 6) much three flagellomeres; flagellomeres (fig. 4), longer than maxillary palpus, shorter than exceptfirst and last, broader than long, with- glossa; segment one not unusually narrowed out elongate specialized hairs; each flagello- at base; segments one and two equal in mere normal, without raised ring of sensilla length, extensively flattened on dried speci- that partly circumscribes subsegment half- men, with only ventral surface sclerotized; way as in Spheodosoma. length of segment three two-thirds length of Pre-episternal groove well developed be- segment two (equal to segment two in Ceb- low level of scrobe; mesepisterna forming lurgus), appearing not flattened because of only shallow longitudinal groove along ven- lack of sclerotization but probably actually tral midline, less pronounced than median flattened in life; segment four about one- grooves of Penapis and Ceblurgus. Dorsal half length of segment three (about three- surface of propodeum about as long as scu- fourths length of segment three in Ceblur- tellum. gus), scarcely flattened, not angled from Forewings dusky because setae more nu- other segments as is also true for Ceblur- merous and perhaps darker brown than in gus (i.e., its long axis a continuation of seg- Penapis and Ceblurgus; veins dark brown, ment three), contrasting with very short, about the same as in Penapis penai; stigma cylindrical segment four of Penapis, which with marginal viens dark, disc pale, as in P. angles outward from long axis of segment moldenki; three submarginal cells present, last three. Galea elongate, postpalpal part two together about as long as first; basal vein slightly longer than stipes, without comb on strongly curved, about as in Penapis. Legs inner surface; outer edge of galeal blade relatively unmodified, like those of Penapis membranous, with many short setae and few except where noted, not highly modified like longer ones, much as in Penapis and Ceb- those of Ceblurgus and Dufourea; legs slen- lurgus (but galea of Ceblurgus with many der; fore- and midfemur, tibia, and tarsus not more short setae than in Goeletapis and Pen- as robust as those of Penapis; forebasitarsus apis); maxillary palpus (figs. 7, 8) gradu- long, distinctly more than one-half length ally tapering to pointed apex, slightly less of entire tarsus (not distinctly less than one- than half as long as galeal blade (com- half length of tarsus); foretarsal claws ap- pared with nearly as long as blade in Pen- pearing simple but actually microscopically apis), with three sclerotized segments and cleft with rami scarcely diverging (as in one unsclerotized segment (not six sclero- male Ceblurgus), not with pronounced in- tized segments as in Penapis); first seg- ner tooth as on other legs or as on foretar- ment cylindrical, second and third seg- sus of Penapis; strigilis normally long, not ments, sclerotized only on outer surface, short as in Ceblurgus; midtibia with thormlike flattened; second segment slightly longer apical projection on outer surface; midbasi- than first, third segment subequal to second; tarsus moderately long, almost or actually fourth segment unsclerotized, one-half as long as midtibia, so that first two tar- length of third, not obvious on dried spec- someres together longer than midtibia (in imen, appearing as tapering distal end of Penapis midbasitarsus shorter, little more segment three (in contrast, maxillary palpus than one-half length of midtibia, and mid- of Ceblurgus even shorter compared with tibia equal in length to first four tarsomer- postpalpal part of galea, composed of four es); disc of basitibial plate partly obscured by segments, all of which are at least partly recumbent setae; hind basitarsus almost as sclerotized and do not taper toward palpal long as hind tibia (in Penapis, hind basi- apex). tarsus slightly less than one-half as long as 1997 ROZEN: ROPHITINE BEES tibia); tibial spurs minutely ciliate; outer es, producing reticulate pattern (fig. 15); S8 hind tibial spur about two-thirds as long as with maximum width of basal process only inner one (outer one almost as long as in- slightly wider than that of apical process and ner one in Penapis); arolia present. with median expansion wider than both (fig. Metasoma with posterior margins of terga 16); sclerotized part of apical process with moderately depressed; gradulus of T2 directed dorsal surface more or less evenly covered backward at side, above and behind spiracle; with moderately short simple setae (fig. 16); posterior margin of T6 weakly impressed; ventral surface with only scattered setae, as T7 normally hidden by T6, dorsally evenly illustrated (fig. 16), without subapical clump rounded, punctate throughout, without im- of setae as in Penapis; extreme apex of apical punctate pygidial area defined by strong process membranous or semimembranous. carina as in Penapis (Mchener, 1965: fig. Genital capsule (fig. 10) unusually elongate, 20) and Ceblurgus (Urban and Moure, even more so than that of Penapis, widest at 1993: fig. 5). S2 slightly depressed immediately behind gradulus (more broadly de- base of gonoforceps in dorsal or ventral pressed in Penapis); S3 and S4 normal con- view (in Penapis widest at base of penis vex plates, not highly modified as in Pen- valve); anterior foramen of gonobase ventral apis; S5 normally convex, not emarginate in position as in Penapis, oval in shape, apically, without sublateral downward-di- somewhat longer than wide (fig. 10), but rected processes, and with pronounced me- not about three times longer than middle dian punctate apical process (fig. 13); erect width as in Penapis (Michener, 1965: fig. setae along posterior margin of sternum mod- 25); voselia moderate in size, larger than in erately long, plumose, but setae on apical Penapis, apparently consisting of single com- two-thirds of posterior median process be- plex sclerite bearing approximately 3-5 stout, coming somewhat shorter and nonplumose; dark teeth at apex of curved distal process; S6 (fig. 14) generally triangular, with narrow gonoforceps rounded apically as seen from median apical emargination that is nar- above or below (fig. 10); penis valve (fig. rower than that of Penapis (Michener, 10) with round, partly sclerotized, subapi- 1965: fig. 21) and without transverse inci- cal, dorsal projection that may appear re- sions on each side; sternum normally con- curved because of backward-directed, vex, without sublateral mounds, becoming sclerotized element, with subapical pointed membranous toward posterior margin; S6 dorsal process on inner surface, and with with pronounced tuft of long, erect, plu- sharp-pointed, sclerotized, recurved ven- mose hairs on each side of midline directed tral projection apicad of dorsal projection posterolaterally, so that setae appear part- (contrasting with unmodified penis value of ed along midline of sternum; these setae Penapis as seen in lateral view). arising from darkly pigmented alveoli; Female. Body length 6.5-8.8 mm (allotype sternum also with pair of smaller recum- 7.7 mm). Length of forewing from costal bent setal tufts posterior to plumose tufts; sclerite to wing tip 4.2-5.1 mm (allotype 4.4 these setae shorter, nonplumose, directed posteromedially, but difficult to see on un- mm). cleared specimens because they are hidden Integumental texture, color, vestiture. In- beneath plumose tufts (in Penapis, integu- tegument as in male except for following: un- ment of S6 with sublateral moundlike pro- dersurface of flagellum brown, not ochra- cess on each side, with narrow transverse ceous; foretarsus and foretibia brown, not incision on each side [figs. 32, 38, 44], and ochraceous; on three specimens, integumental with glabrous, nonpilose region along mid- color of metasomal terga reddish infused with line); S7 (fig. 15) wishbone-shaped, without brown on anterior part of Ti and laterally and lateral lobes, as in Penapis (Michener, 1965: immediately anterior to marginal areas on oth- fig. 26), but with median sclerotized apical er terga; on other specimen, all terga generally process narrower than in Penapis; lateral medium brown. Body pubescence only slight- membranous areas with fine transverse, mul- ly shorter than that of female Penapis and tiserrate ridges that anastomose in some plac- Ceblurgus; that on ventral surface of mandi- 14 AMERICAN MUSEUM NOVITATES NO. 3206

not all, species of Conanthalictus, and contrasting with predominantly simple scopal setae of Penapis, Xeralictus, Protodufourea, Sphecodosoma, Rophites, and Dufourea; at least some scopal (discal) setae of S2 with long, conspicuous branches (those of Ceb- lurgus with short dense branches); sternal scopal setae arising from raised, darkly pigmented sockets on posterior parts of S2, S3, and S4; integument anterior to band of sockets on each of these segments somewhat dull 17 18 20 g due to tessellate surface and presence of very fine recumbent setae that are paler than erect brownish scopal setae; prepygidial fimbria dense with even denser mass of shorter reddish plumose setae at midline. Structure. Position of antennal sockets, an- Figs. 17-20. Setae of South American Rophi- terior tentorial pits, and median section of ep- tinae. 17. Goeletapis peruensis, scopal seta. 18. Penapis penai, scopal seta. 19. Ceblurgus longi- istomal suture as in male; subantennal sutures palpis, scopal seta. 20. Ceblurgus longipalpis, and epistomal ridges also as in male; frontal long seta from labial palpus of female. Scale carina absent; facial fovea not clearly present, (=0.5 mm) refers to all figures. but faint depression immediately mesad of inner orbit and slightly modified integument between there and antennal socket possibly fo- ble somewhat less plumose than that of male, vea (this area would not have been so iden- variable in length, and not therefore appearing tified if more pronounced, concave fovea had fringelike; forecoxa with plumose setae less not been identified in female Penapis; fovea than one-half length of coxa; long dorsal hairs not evident in Ceblurgus); anterior and pos- offoretarsus longer than hairs on ven- terior mandibular articulations tral brush, not curved at tips, with short but equidistant from eye. Inner margins of eyes subparallel. clearly visible branches; outer apex of mid- Labrum as described for male without tibia with dark, sharp, curved, thick, non- except plumose, spinelike setae (sometimes diffi- transverse row of simple setae about two- cult to see on dried specimen because of thirds way to apex, but with very short setae dense, pale, semirecumbent, plumose setae, scattered over apical one-third. Mandible nor- but conspicuous on cleared specimen in mal in length, with rounded dorsal subapical glycerin; similar setae also present in Ceblur- tooth. Other mouthparts as described for male. gus); midbasitarsus obscured dorsally by Scape exclusive of basal ball moderately dense, tan, plumose setae; scopal setae pres- long, somewhat longer than distance be- ent on hind basitarsus, femur, tibia, and ap- tween antennal sockets. parently trochanter and coxa, and also appar- Mesosomal and wing characters as de- ently on apical areas of basal metasomal ster- scribed for male. All tarsal claws simple as na (but not on sides of metasoma); scopal se- in Ceblurgus, not bifid as in female Pen- tae on anterior surface of basitarsus less apis; midtrochanter like that of Ceblurgus, than one-half as long as basitarsus; these without ventral carina; hind basitarsus giv- setae tan, semirecumbent, highly plumose, ing rise to second tarsomere at apex; tibial and dense, obscuring basitarsus; scopal se- spurs and arolia as described for male. tae of hind tibia moderately tan, long, Metasoma with posterior margins of terga erect, highly plumose with long, conspicu- only moderately depressed. Terga unremark- ous branches arising distally (fig. 17); these able; T2 with lateral blackish areas in posi- setae similar to those of Ceblurgus except lat- tion of foveae (as in Ceblurgus), but these ter with denser, shorter branches (fig. 19); areas neither concave nor with modified se- these setae also similar to those of most, but tae or integumental sculpturing; T5 not di- 1997 ROZEN: ROPHITINE BEES 15 vided posteriorly along midline by cleft or by specialized area of fine punctures and short setae; exposed sterna not modified (except for integumental and setal characters above); internal apical structures not examined. Pygidial plate normally hidden; surface curved, sides strongly carinate, coming to acute apex with apex itself narrowly subtruncate, with narrow rim. TYPE MATERIAL: PERU: Lima Dept.: holotype male, allotype, 8 male, 3 female paratypes, 15 km WSW Sayan, VI-26-28-1995 (J. G. Rozen, A. Ugarte), on Exodeconus, probably maritimus; 57 male, 24 female paratypes, same except VI-27-1995; 2 male, 2 female paratypes, same, preserved in Kahle's solution; 23 male, 2 female paratypes, same locality and collectors, V-1 1-1996, no plant data; 31 male, 22 female paratypes, same except V-12-1996, on Exodeconus prostratus; 13 male, 2 female paratypes, 2 km NE Sta. Rosa de Quives, near Yangas, V-27-1996 (J. G. Rozen, A. Ugarte), on Exodeconus probably prostratus; 11 male, 6 female paratypes, Sta. Rosa de Quives, near Yangas, V-28- 1996 (J. G. Rozen, A. Ugarte), on Exode- conus probably prostratus; 46 male, 27 female paratypes, same except V-28-29-1996; 4 male, 3 female paratypes, same except V-29-1996, preserved in Kahle's solution. La Libertad Dept.: 3 male, 3 female para- La 7.5 NNW V-20- types, Gloria, Paijain, Fig. 21. Map of northwestern Peru from Lima 1996 (J. G. Rozen, A. Ugarte), on Exode- Dept. northward, showing collection localities conus prostratus; 1 male, 6 female paratypes, (stars) of Goeletapis peruensis. same except V-22-1996; 31 male, 14 female paratypes, same except V-23-1996. The holotype and allotype are in the collections of Robert G. Goelet, Chairman Emeritus of the the American Museum of Natural History. Board of Trustees of the American Museum Paratypes are deposited there and in the fol- of Natural History, in recognition of his long lowing: Museo de Historia Natural, Lima, interest in enhancing the collections of this Peru; Central Texas Melittological Institute, institution. The trivial name refers to the Austin, Texas; Snow Entomological Divi- country in which the specimens were col- sion, Natural History Museum, University of lected. Pronunciation: g6-l1t-a-pfs. Kansas, Lawrence, Kansas; Universidade REMARKS: The single male collected by Dr. Federal do Parana, Brazil; USDA Bee Biol- Simpson from near Paijan (metasomal apex ogy and Systematics Laboratory, Utah State lost, labrum hidden from view) that is avail- University, Logan, Utah; collection of Prof. able to me differs from the type series as Toro, Universidad Catolica de Valparaiso, follows: Body smaller and somewhat more Valparaiso, Chile. slender (wing length 3.7 mm) than even the As indicated in figure 21, the distribution smallest male in the type series; midtibia of this species is restricted to the arid costal with basal patch of short recumbent setae region of Peru west of the Andes. nearly white, concolorous with longer setae ETYMOLOGY: The generic name honors Mr. elsewhere on segment. Female specimens 16 AMERICAN MUSEUM NOVITATES NO. 3206 collected by Dr. Simpson near Paij'an agreed hairs suggests it is not likely to be a cotton in that they too possessed branched scopal (or any big pollen) specialist." Discovery of setae on their hind legs, according to notes this species in the Paijain area visiting Exo- by Dr. Neff. Neff's illustrations of the male deconus prostratus bears out this prediction. genitalia and terminal sterna and his notes At all collection localities, males flew rap- describing mouthparts agree with the de- idly from flower to flower, almost certainly scription above. This information and the in search of mates. However, no copulations fact that some of the paratypes were collect- were observed, perhaps because they occur ed from the Paijain area in 1996 leave little within the throat of the flower (fig. 23) where doubt that Simpson's 1983 collection was they would be difficult to detect. Obviously this species. The unusually small male from males and females did not fly in copula, be- that series was atypical, as apparently was cause, had then done so, mating pairs would the association of the series with native cot- have been seen, as were mating pairs of the ton (see Biology, below). much less common Callonychium (Parany- BIOLOGY: Host plants were low-growing chium). Males frequently landed on the flow- members of the Solanaceae. In 1995, they ers and usually straddled the throat. This were identified as Exodeconus, probably stance is interpreted to be part of mate maritimus (Benth.) D'Arcy, but possibly E. searching. Occasionally, however, males en- prostratus (L'Herit.) Def., and in 1996 as E. tered the throat head first, presumably to prostratus. At the site near Sayan, they grew feed. in sandy, more or less horizontal areas in a Females, always less common than males, boulder-strewn ravine (figs. 22, 23). The Sta. also flew swiftly and, on landing on flowers, Rosa de Quives site was mostly along the immediately entered head first to gather pol- lower edge of a steep hillside from which len and nectar. Many captured females ex- corn (maize) had recently been harvested and hibited dry, lavender pollen of the host plant which was also an archaeological site. At La on their mesosomal venters, midcoxae, and Gloria, near Paija6n, the plant grew at the midtrochanters and on all segments of their edge of, and in open areas in, an abandoned hind legs, as well as ventrally on the basal olive grove where the sandy ground surface segments of their metasomas. Females did was nearly horizontal (fig. 24). All three sites not vibrate the anthers of Exodeconus; be- were sandy and exposed to the sun where the cause these anthers are not porous, the flow- host plants grew. The flowers of the host ers do no require buzz pollination as do those plant were funnelform, approximately 3 cm of Solanum. across, with the limb (outer part) nearly Although considerable effort was expend- white and the basal (inner) part purple. At ed in attempting to find nests, none was least at Sta. Rosa de Quives and La Gloria, found at any of the localities. Hence, nests flowers opened around noon and closed of none of the South American rophitines again around 3 p.m. Although occasional have been discovered to date, and their im- males were observed before noon, almost all mature stages are unknown. No cleptopar- diurnal activity by males and females oc- asitic bees were discovered at localities curred during the 3-hour period when flowers where Goeletapis was observed. were open. At this time these bees were abun- Dissection of one female preserved in dant, the most common insect visiting the Kahle's solution yielded information con- plants. Other bees foraging from, and seeking cerning its ovaries and oocytes, apparently mates at, these plants included Spinoliella the first such data reported for any rophitine. (Spinoliella) at Sayan (in 1995 and 1996) and Each ovary consisted of three ovarioles (3: La Gloria and Callonychium (Paranychium) 3). A single elongate mature oocyte (chorion at all three sites. visible through follicular wall) was present The specimens Simpson collected near (mature oocytes/ovary = 0.5); it length was Paijan were taken on native cotton, Gossyp- 1.85 mm. The egg index (E/M) (Iwata and ium barbadense. Neff (personal commun.) Sakagami, 1966) was 0.96. The oocyte was commented, "Although collected on cotton, unremarkable, like the eggs and ooctyes of the relatively dense scopae with branched most other short-tongued bees. It was 1997 ROZEN: ROPHITINE BEES 17

. 4k- *uL-'

j."Allabb - - ./" .-, . "64. :7.- . - I

w ., 'Id. js,&Q. 'N4-1

Figs. 22, 23. 22. Habitat of Goeletapis peruensis at 15 km WSW Sayan, Lima Dept., Peru. Figure near middle is standing next to mat of host plant, Exodeconus. 23. Same locality, closeup of host plant, with eyeglass case in foreground. 18 AMERICAN MUSEUM NOVITATES NO. 3206

Fit Fig. 24. Habitat of Goeletapis peruensis at La Gloria, 7.5 km NNW Paijan, La Libertad Dept., Peru, showing author collecting on Exodeconus prostratus.

curved, widest near its anterior end, and gradually tapered posteriorly. Its chonon was smooth, semitransparent, and without retic- ulations or other sculpturing. These data agree with what is known about the other halictid subfamilies (Iwata and Sakagami, 1966, and references therein). PENAPIS MICHENER Figures 2, 18, 25-47 Penapis Michener, 1965: 324. Type-species: Pen- apis penai Michener. Monobasic. DIAGNOSIS: In addition to the features identified in the treatment of Goeletapis, males of all three species of Penapis are more coarsely punctate, hairier with longer setae, and in general less shiny than those of Goe- letapis. Figs. 25-28. 25, 26. Heads of Penapis mol- DESCRIPrION: Male. Body length 6.0-8.8 denkei and P. toroi respectively, lateral view. 27, mm. Length of forewing from costal sclerite 28. Midtibial spurs of same species. 27. Straight to wing tip 4.3-4.8. spur (left), recurved spur (right). Upper scale (= Integumental texture, color, vestiture. In- 0.5 mm) refers to figures 25, 26; lower scale (= tegument nonmetallic, black to very dark 0.1 mm) refers to figures 27, 28. brown; integument of head and mesosoma 1997 ROZEN: ROPHITINE BEES 19 moderately punctate, more coarsely so than equal to length. Antennal sockets far below that of Goeletapis; integument of upper part middle of face. Anterior tentorial pits close of head microscopically uneven though to outer lower edge of antennal sockets; an- somewhat shiny; upper middle part of clyp- tennal suture present but weak; integument eus coarsely, irregularly punctured; dorsal between antennal socket and lateral section surface of propodeum with smooth areas less of epistomal suture smooth, impunctate, and extensive than in Goeletapis because of more with more or less evident suture extending extensive sculpturing and punctation, hence obliquely downward from anterior tentorial not appearing polished; undersurface of fla- pit to epistomal suture, as discussed in Mor- gellum brown to dark brown; legs entirely phology. Clypeus moderately long, length dark brown to almost black. Body setae tend- distinctly more than one-third distance being to be long, disorganized, often partly ob- tween lower inner orbits in front view, its scuring integument. Pubescence on ventral length and protuberance varying according to surface of mandible long to moderately species; median section of epistomal suture short, not arranged as organized fringe as in (fig. 2) weak but evident; lateral section ev- Goeletapis; mandible without conspicuous ident as it extends downward from junction basal tuft of long, white hairs as in Ceblurgus of antennal suture and then runs laterally (see (Urban and Moure, 1993: figs. 3, 4); setae of description in Morphology); frontal carina maxilla and labial palpus fine, nonplumose, absent except for small longitudinal tubercle short, inconspicuous, not long and plumose about halfway between level of antennal as in Ceblurgus; anterior surface of forecoxa sockets and median ocellus; facial fovea ab- with setae yellowish to white, some long, as sent; malar space nearly absent, with anterior long or nearly as long as coxa; pubescence mandibular articulation slightly farther re- on outer surface of mid- and hind tibiae and moved from eye than is posterior articula- basitarsi more or less erect, plumose but with tion; vertex produced well above upper ends short branches, moderately dense so as to of eyes. Inner margins of eyes subparallel to partly obscure integument, not organized in slightly converging below. Labrum much parallel as in Goeletapis; midfemur without broader than long, width more than 2 times basal patch of short, golden setae on anterior length along midline; anterior surface evenly surface as in Goeletapis, all setae there long, curved, without transvers ridge; row of long, plumose, and similar to those on venter of tapering, simple setae arising from apex. mesosoma; hind trochanter without apical, Mandible elongate as seen in maximum pro- specialized, long, curved setae as in Ceblur- file but somewhat shorter than eye in lateral gus. Metasomal pubescence not forming view, with preapical inner tooth. Proboscis bands on Tl-T5; dorsal surface of T2-T4 long, in repose reaching between forecoxae; with very fine, recumbent setae arising from glossa elongate; annular hairs of glossa sim- fine punctures at base of discs, elsewhere ple, acutely pointed, elongate; paraglossae with scattered, coarser punctures and larger, small, linear; mentum semimembranous, pale, erect setae; margins of T1-T6 glabrous somewhat pigmented; lorum membranous dorsally; lateral surfaces of metasomal terga except for elongate sclerite on each side; fla- with scattered, long, plumose, erect setae and bellum absent; labial palpus much longer some with finer, recumbent setae as well; py- than maxillary palpus, somewhat shorter gidial plate of T7 completely glabrous; Si- than glossa; segment one not unusually nar- S3 with short recumbent setae and sublateral rowed at base; segments one and two sub- vague patches of long, erect setae, conspic- equal in length, extensively flattened on dried uousness of which varies specifically; setae specimens, with only ventral surface sclero- of S4-S8 described below under Structure; tized; length of segment three one-third to apex of gonoforceps nonsetose, in contrast to nearly one-half as long as two, hence rela- that of Goeletapis, but gonoforceps with tively shorter than in Goeletapis, clearly flat- scattered, moderately long setae subapically tened; segment four a little more than one- on ventral surface and scattered, short setae third to less than one-quarter length of three, subapically on dorsal surface. scarcely flattened, angled from axis of other Structure. Head breadth approximately segments. Galea elongate, postpalpal part 20 AMERICAN MUSEUM NOVITATES NO. 3206 slightly longer than stipes, without comb on depressed, somewhat more so than in Goe- inner surface; outer edge of galeal blade letapis; gradulus of T2 directed backward at membranous, with many short setae and a side, above and behind spiracle; posterior few longer ones; maxillary palpus elongate, margin of T6 well impressed and well dif- only slightly shorter than galeal blade, six- ferentiated from disc; T7 exposed, with well- segmented; all segments sclerotized, normal- defined, large, glabrous pygidial plate with ly successively thinner, but palpus not com- carinated edges (Michener, 1965: fig. 20); ing to point as in Goeletapis; inner surface S2-S4 with disc of postgradular area con- of segment two and perhaps of segment three cave; posterior margin of S4 (figs. 29, 35, 41) membranous; segment two approximately with uniform comb of stout, simple setae on 1.5 times length of segment one; segments each side; in front of comb, disc with cluster one and three subequal in length; segments of long, outwardly recurved, plumose setae four, five, and six subequal in length, togeth- on each side medially in addition to long and er as long as segment two. short, straight setae elsewhere; S5 (figs. 30, Scape very short, length exclusive of basal 36, 42) deeply emarginate from behind me- ball approximately twice maximum diameter, dially, with each side of emargination pro- shorter than distance between antennal sock- duced as pronounced, sharp, flattened pro- ets, and shorter than first three flagellomeres; cess directed downward and backward; shape middle flagellomeres tending to be about as of process varying according to species; mid- long as broad; flagellum without elongate dle of emargination produced posteriorly as specialized setae; flagellomeres without rings median, pointed process, size and shape of of specialized sensilla as in Sphecodosoma. which depending upon species; S6 (figs. 32, Pre-episternal groove well developed be- 38, 44) bearing narrow, transverse incision low level of scrobe; ventral median longitu- on each side (apparently permitting flexing dinal pro- and mesepisternal groove some- of apical part of sternum), bearing sublateral, what more deeply impressed than in Goele- moundlike process on each side, and with tapis. Dorsal surface of propodeum about as fully sclerotized apex moderately broadly long as scutellum. emarginate medially; S7 without paired api- Forewings hyaline, with fewer setae than cal lobes; S8 (figs. 34, 40, 46) with maxi- in Goeletapis; veins medium brown to dark mum width of basal process much wider than brown; stigma uniformly brown or paler with maximum width of apical process; median dark border; three submarginal cells present, expansion slightly narrower to slightly wider last two together slightly longer than first; than maximum width of basal process. Gen- basal vein strongly curved, about as in Goe- ital capsule elongate (Michener, 1965: fig. letapis. Legs little modified; fore- and mid- 25); anterior foramen of gonobase very elon- femur, tibia, and tarsus moderately robust, gate, somewhat dumbbell shaped in that an- somewhat more so than those of Goeletapis; terior and posterior ends wider than midlength of forebasitarsus somewhat less than width, more than three times longer than one-half length of tarsus; foretarsal claw with midwidth. pronounced inner tooth as large as those of Female. Body length approximately 6.5- midtarsus and hind tarsus; strigilus normally 8.8 mm. Length of forewing from costal long; midtibia with thornlike apical projec- sclerite to wing tip 4.7-5.2 mm. tion on outer surface; midbasitarsus moder- Integumental texture, color, vestiture. As ately short, not much more than one-half described for male except for following: length of midtibia; midtibia equal in length Erect setae of frons and scutum somewhat to at least first two tarsomeres and sometimes longer than those of Goeletapis, but in gen- to first four tarsomeres; disc of basitibial eral these setae tending to be shorter than plate partly obscured by recumbent setae; those of male; forecoxa with some setae hind basitarsus slightly less than one-half nearly as long as coxa; long dorsal hairs of length of hind tibia; tibial spurs minutely cil- foretarsus appearing nonplumose (but actu- iate; outer hind tibial spur more than two- ally some with submicroscopic branches api- thirds as long as inner spur; arolia present. cally), longer than hairs on ventral brush of Metasoma with posterior margins of terga tarsus, not curved at tips; outer apex of mid- 1997 ROZEN: ROPHITINE BEES 21 tibia lacking sharp, curved, stout, nonplu- carina; midbasitarsus about equal in com- mose, spinelike setae as found in female bined length to remaining four tarsomeres Goeletapis and Ceblurgus; midbasitarsus (exclusive of pretarsus). only partly hidden by setae that are longer, Metasoma with posterior margins of terga paler, more erect, but less dense than those moderately depressed. Terga unremarkable; of female Goeletapis; scopal setae apparently T2 with foveae not indicated by pigmenta- present on hind basitarsus, femur, tibia, tro- tion, sculpturing, or setal patterns; T5 not di- chanter, coxa, and apparently on apical areas vided posteriorly along midline by cleft or of basal metasomal sterna, as in Goeletapis, by specialized area of fine punctures and as judged by presence of long, erect setae; at short setae. Pygidial plate normally hidden; least some scopal setae on outer surface of at least in some species, surface curved, sides hind basitarsus longer than one-half length of strongly carinate, coming to acute apex, with basitarsus; these setae nonplumose, pale apex broadly subtruncate. cream color, and more erect and less dense REMARKS: The three species of Penapis, all than comparable setae of Goeletapis, so that known only from Chile (fig. 21), are remark- basitarsus clearly visible through them; al- ably similar. The differences separating most all scopal setae of hind tibia very long, males are consistent and are identified in the nonplumose (fig. 18) (some setae near base key below and in the following descriptions. of tibia with one or two fine branches on Most of these characters are sex-limited. some specimens); scopal setae of S2 mostly However, the female characters also seem to nonplumose; some sternal scopal setae else- be consistent although subtle and difficult to where with branches; pale sternal scopal se- quantify. tae arising from pronounced sockets that are In the lists of specimens examined, some not more darkly pigmented than surrounding of the provincial assignments have been al- integument, which is dark brown; integument tered from those given on labels to corre- anterior to band of sockets on S2-S4 rather spond to currently recognized political shiny, less tessellated than in Goeletapis and boundaries in Chile. almost devoid of fine, pale, recumbent setae; prepygidial fimbria dense with even denser KEY TO SPECIES OF PENAPIS mass of shorter, reddish setae at midline, as in Goeletapis. 1. S6 of male with sublateral process backward- Structure. As described for male except pointed at approximately 450 in strict lateral for following: Head breadth slightly greater view (fig. 45); S5 of male with median pro- than length. Clypeal length and protuberance cess slightly longer than sublateral processes not varying according to species; frontal ca- in ventral and lateral views (figs. 42, 43); S4 rina facial fovea represented of male with median clump of approximate- absent; by ly 8-12 long, plumose, outwardly recurved slightly dull, vaguely concave area immedi- setae on each side (fig. 41); pterostigma of ately mesad of eye, somewhat more pro- both sexes with veined margins somewhat nounced than in female Goeletapis; on darker than disc; midtibial spur of both sexes cleared specimen, integument of fovea more gently curved to base (fig. 28); scopa strong- darkly pigmented than that of rest of frons. ly amber colored; midfemur of female with Inner margins of eyes subparallel or slightly ventral half of anterior surface shallowly but diverging below. Mandible normal in length broadly concave .... Penapis toroi Rozen with rounded dorsal subapical tooth. S6 of male with sublateral process evenly Scape exclusive of basal ball short, but not rounded in lateral view (figs. 33, 39); S5 of as short as that of male, about 3 times max- male with median process slightly shorter or imum diameter, slightly shorter than, or sub- much shorter than sublateral processes in ventral or lateral views (figs. 30, 31, 36, 37); equal to, distance between antennal sockets. S4 of male with median clump of 5 or fewer Middle flagellomere broader than long. plumose, outwardly recurved setae on each Length of forebasitarsus about one-half side (figs. 29, 35); these setae tending to be length of tarsus; tarsal claws bifid, in contrast somewhat shorter than those of P. toroi; to simple tarsal claws of female Goeletapis pterostigma of both sexes either uniformly and Ceblurgus; midtrochanter with ventral dark brown (P. penai) or with pale disc 22 AMERICAN MUSEUM NOVITATES NO. 3206

much paler than veined border (P. molden- projecting but never as much as in P. mol- kei); midtibial spur of both sexes straight or denkei (this character often difficult to see be- recurved at midlength (fig. 27); scopa gray- cause of tuft of dense, plumose setae along ish white (P. moldenkei) or grayish amber posterior of S6 with (P. penai); midfemur of female convex over edge process). sublateral, most of anterior surface, with only narrow, moundlike process evenly rounded in lateral inconspicuous, shallow concavity along ex- view (fig. 33) (about as in P. moldenkei); this treme lower edge) ...... 2 process laterally compressed so that in ventral 2. S6 with sides of sublateral process compressed view (fig. 32) appearing more linear than that so that process more linear in ventral view of P. moldenkei. S8 as figured (fig. 34). (fig. 32); S5 with median process longer, Female. As described for male (exclusive nearly as long as sublateral processes (fig. of sex-limited male characters) except for 30); pterostigma of both sexes uniformly dark following: Midfemur convex over most of brown, disc concolorous with veined mar- anterior surface but with inconspicuous shal- gins; scopa and metasomal setae (exclusive low concavity along extreme lower of orange pseudopygidial tuft) tending to be edge; more amber colored ..... Penapis penai scopal hairs tinged grayish amber. Michener MATERIAL EXAMINED: CHILE: Elqui S6 with sides of sublateral process not com- Prov.: 1 female, Llano de la Higuera, N of pressed so that process more oval in ventral El Tofo, X-14-1971 (Rozen and Penia); 2 view (fig. 38); S5 with median process short, males on plants no. 41722, 41724, 6 females poorly developed, distinctly shorter than on plants no. 41723, 41726, 41736, 41910, sublateral process (fig. 36); pterostigma of 41911, 41912, Carretera Pan-Am al norte de both sexes with veined margins dark, disc La Serena, Int. Biol. Program, 1970-1972 paler; scopa and metasoma setae (exclusive (A. R. Moldenke); 9 males, 1 female, El of orange pseudopygidial tuft) tending to be Tofo, IX-23-1980 M. more grayish white ... Penapis moldenkei (V. Cabezas, Rojas, H. Bohart, Toro, and Rozen Toro); 2 males, same except X-1972 (L. Ruz, V. Cabezas) on Argylia radiata. Huasco Prov.: 1 male (holotype), Chafiaral de Ace- Penapis penai Michener ituno, X-23-25-1957 (L. Penia). Figures 18, 29-34, 47 At present, the distribution of this species Penapis penai Michener, 1965. (fig. 47) appears to be allopatric with that of the other two species and is the most south- DIAGNOSIS: See key and description below. ern. The type locality is near the coast, west DESCRIPTION: Male. Lower part of face of Domeyko, and should not be confused (Michener, 1965: fig. 27) only moderately with Chaniaral, Chafiaral Prov. drawn out, about as in Penapis moldenkei (fig. BIOLOGY: So far as is known, the only host 25), less than in P. toroi (fig. 26). Pterostigma plant of this species is Argylia radiata (Big- uniformly dark brown, disc almost same color noniaceae). as marginal veins. Midtibial spur (as in fig. 27) in lateral view straight or slightly re- Penapis moldenkei curved at midlength. S4 (fig. 29) with median Bohart, Toro, and Rozen, new species cluster of 3-5 long, plumose, outwardly re- Figures 2, 25, 27, 35-40, 47 curved setae on each side (and sometimes with similar but shorter setae laterad of main DIAGNOSIS: See key and description below. cluster); simple setae of apical comb only DESCRIPTION: Male. Lower part of face faintly amberish, numerous, approximately 20 (fig. 25) only moderately projecting, about as on each side. S5 (fig. 30) with median apical in Penapis penai, less than in P. toroi (fig. process of sternum moderately long, shorter 26). Pterostigma with pale disc bordered by than that of P. toroi but longer than that of dark brown marginal veins. Midtibial spur P. moldenkei; apically this process somewhat (fig. 27) in lateral view straight or slightly flattened but not as strongly keellike as in P. recurved at midlength. S4 (fig. 35) with me- toroi; sublateral process in strict lateral view dian cluster of 3-6 long, plumose, outwardly (fig. 31) usually without projection along pos- recurved setae on each side; simple setae of terior margin although sometimes slightly apical comb only faintly amberish, number- 1997 ROZEN: ROPHITINE BEES 23

Figs. 29-34. Penapis penai, male. 29. S4, ventral view, setae shown only on right. 30, 31. S5, ventral (setae on right) and lateral views, respectively. 32, 33. S6, ventral (setae on right) and lateral views, respectively. 34. S8, ventral view, setae shown only on right. Scale (=0.5 mm) refers to all figures. ing approximately 12-15 on each side. S5 of sex-limited male characters) except for (figs. 36, 37) with median apical process of following: Midfemsur convex over most of sternum short compared with that of either anterior surface but sometimes with shallow P. penai or P. toroi; apically this process not concavity along extreme lower edge; scopal strongly flattened, not forming keellike ridge; hairs tending to be more grayish white in sublateral process in strict lateral view (fig. contrast to more amberish scopas of Penapis 37) with pronounced, angled projection penai and P. toroi. along posterior margin. S6 with sublateral, TYPE MATERIAL: CHILE: Copiapo Prov.: moundlike process evenly rounded in lateral holotype male, allotype, 13 male, 13 female view (fig. 39), about as in P. penai; this pro- paratypes, 85 km S Copiap6, X-23-1991 (J. cess not laterally compressed so that in ven- G. Rozen, L. Penia), on Argylia radiata; 12 tral view (fig. 30) appearing oval, not as lin- male paratypes, 40 km S Copiapo, IX-19- ear as that of P. penai. S8 as figured (fig. 1972, 1610-1700 hours (J. L. Neff) on Ar- 40). gylia puberula; 9 male, 1 female paratypes, Female. As described for male (exclusive same except IX-20-1972, 0800-0930 hours; 24 AMERICAN MUSEUM NOVITATES NO. 3206

APEX OF MEDIAN PROCESS (HIDDEN) ' ~-)--.)-SUBLATERAL PROCESS Figs. 35-40. Penapis moldenkei, male. 35. S4, ventral view, setae shown only on right. 36, 37. S5, ventral (setae on right) and lateral views, respectively. 38. Posterior part of S6, ventral view, setae shown only on right. 39. S6, lateral view. 40. S8, ventral view, setae shown only on right. Scale (=0.5 mm) refers to all figures.

4 male paratype, 50 km S Copiapo, IX-29- IX-1971 (L. Ruz); 2 female paratypes, same 1992, 1100 hours (J. L. Neff) on Argylia locality, X-1972 (L. Ruz), on Argylia radia- puberula; 1 male paratype, same except on ta; 2 female paratypes, 10-40 km SE Cal- Calandrinia; 3 male paratypes, 26 mi S Co- dera, X-17-1969 (Rozen and Penia); 1 female piap6, X-19-1969 (Rozen & Pefia); 2 male, paratype, Mineral Atacama, NW Copiap6, 1 female paratypes, 44 km S Copiap6, X-21- 1100 m, X-4-1980 (L. E. Penia). Huasco 1971 (Rozen & Penia); 2 male, 4 female para- Prov.: 1 male paratype, 90 km S Copiapo, types, Travesia, X-1 1-1977 (L. Ruz, H. Flo- X-19-1971 (Rozen & Penia); 12 male, 6 fe- res, J. C. Magunacelaya); 10 male, 4 female male paratypes, Algarrobal, X-1 1-1977 (V. paratypes, same locality, X-1969 (L. Ruz, H. Cabezas, H. Flores, H. Toro); 1 male para- Toro); 27 male, 11 female paratypes, same type, S of Canto del Agua, X-23-1980 (L. E. locality, IX-29-1983 (R. Aldunate, B. Dyer, Penia). Cha-naral Prov.: 1 male paratype, F. Rodriguez, C. Massad, C. Tobar, H. Toro, Chafiaral, X-1991 (H. Toro); 1 female para- J. Vial); 1 male paratype, same locality, type, same except X-1987 (De La Hoz); 18 1997 ROZEN: ROPHITINE BEES 25

Figs. 41-46. Penapis toroi, male. 41. S4, ventral view, setae shown only on right. 42, 43. S5, ventral (setae on right) and lateral views, respectively. 44, 45. S6, ventral (setae on right) and lateral views, respectively. 46. S8, ventral view, setae shown only on right. Scale (=0.5 mm) refers to all figures. male paratypes, 30 km S Chafnaral, X- 1- La Hoz); 1 male paratype, same except 1972, 0930 hours (J. L. Neff) most males IX-15-1994 (V. Ruiz). sleeping in flowers (Argylia puberula) but The holotype and allotype are deposited in some patrolling; 1 male paratype, 40 km S the American Museum of Natural History. Chaniaral, X-12-1983 (J. L. Neff); 6 female Paratypes are there and in the following: col- paratypes, same except on Argylia; 3 female lection of Prof. Toro, Universidad Catolica paratypes, same except on Argylia radiata. del Valparaiso, Valparaiso, Chile; Snow En- El Loa Prov.: 1 male paratype, Chiu-chiu, tomological Division, Natural History Mu- IX-27-1983 (De La Hoz). Antofagasta seum, University of Kansas, Lawrence, Kan- Prov.: 1 male, 2 female paratypes, Paposo, sas; Central Texas Melittological Institute, X-1-1993 (E. Chiappa). Iquique Prov.: 1 Austin, Texas; Universidade Federal do Pa- male paratype, La Tirana, IX-29-1983 (De rana, Brazil; and USDA Bee Biology and 26 AMERICAN MUSEUM NOVITATES NO. 3206

Systematics Laboratory, Utah State Univer- sity, Logan, Utah. The distribution of this species (fig. 47) broadly overlaps that of Penapis toroi, and the two have been collected at Chaniaral, Chafnaral Prov., Paposo, Antofagasta Prov., and La Tirana, Iquique Prov. ETYMOLOGY: This species is named in hon- or of Dr. Andrew R. Moldenke of Oregon State University, Corvallis. BIOLOGY: The recorded host plants of this species apparently are Argylia radiata and A. puberula (Bignoniaceae). According to Dr. Jackie Kallunki, these are the same species, the former being the correct name. This plant species is also the larval food source of Pen- apis penai. Calandrinia (Portulacaceae) may also host P. moldenkei. When queried about possible host plants for this species and P. toroi, Toro (personal commun.) stated that he "used to collect males [of Penapis (without reference to species)] "sleeping in flowers of Argylia and both males and females visiting Calandrinia."

Penapis toroi Rozen, new species Figures 26, 28, 41-47 DIAGNOSIS: See key and description below. DESCRIPTION: Male. Lower part of face strongly drawn out in lateral view (fig. 26), more so than in Penapis penai and P. moldenkei (fig. 25). Veins of forewing brown, about as in P. moldenkei; pterostigma with disc medium brown, bordered by dark brown marginal veins, but contrast of pale disc with dark border not so pronounced as in P. moldenkei. Midtibial spur in lateral view (fig. 28) evenly curved its entire length. S4 (fig. 41) with median cluster of approximately 12 long, plumose, outwardly recurved setae on Fig. 47. Map of Chile from Limari Prov. each side; simple setae of apical comb amber northward, showing known distributions of the colored, numbering approximately 12 on three species of Penapis. each side. S5 (figs. 42, 43) with median apical process of sternum long, longer than that of either P. penai or P. moldenkei and in ven- about 450 in lateral view (fig. 45); this pro- tral view extending beyond sublateral process laterally compressed so that in ventral cesses; apically this process flattened, form- view (fig. 44) appearing linear. S8 as figured ing keellike median ridge; sublateral process (fig. 46). in strict lateral view (fig. 43) with angled Female. As described for male (exclusive projection along posterior margin. S6 with of sex-limited male characters) except for sublateral, moundlike process with back- following: Pterostigma with disc medium ward-projecting angle, almost always of brown, veined border slightly darker, but 1997 ROZEN: ROPHITINE BEES 27 contrast between disc and border even less Other paratypes are in the American Muse- pronounced than in male. Midfemur with an- um of Natural History; Central Texas Mel- terior surface broadly concave along lower ittological Institute, Austin, Texas; Snow En- half; scopal hairs tinged with grayish amber, tomological Division, Natural History Mu- about as in Penapis penai, contrasting with seum, University of Kansas, Lawrence, Kan- whitish scopal setae of P. moldenkei. sas; and USDA Bee Biology and Systematics TYPE MATERIAL: Antofagasta Prov.: ho- Laboratory, Utah State University, Logan, lotype male, 10 male paratypes, Paposo, X-1- Utah. 1983 (H. Toro); allotype, 15 male paratypes, The distribution of this species is mapped same except collector (E Rodriguez); 48 (fig. 47). male, 1 female paratypes, same except col- ETYMOLOGY: It is a pleasure to name this lectors (E. Chiappa, B. Dyer, De La Hoz, 0. species in honor of Prof. Haroldo Toro in Martinez, M. Rojas, C. Tobar). Cha-naral recognition of his distinguished contributions Prov.: 1 male, 1 female paratype, Chaniaral, to our understanding of the bees of Chile. X-1987 (De La Hoz). Iquique Prov.: 9 male BIOLOGY: No host plant of this species has paratypes, La Tirana, IX-29-1983 (B. Dyer, been identified. However, the fact that the 0. Martinez, F Rodriguez, M. Rojas). bee has been collected at three localities with The holotype, allotype, and paratypes are specimens of Penapis moldenkei suggests in the collection of Prof. Toro, Universidad that it too may visit Argylia. See Biology, P. Catolica de Valparaiso, Valparaiso, Chile. moldenkei, above.

REFERENCES Ebmer, A. W. Michener, C. D., and R. W. Brooks 1984. Die westpalaarktischen Arten der Gat- 1984. Comparative study of the glossae of tung Dufourea Lepeletier 1841 mit il- bees (Apoidea). Contrib. Am. Entomol. lustrierten Bestimmungstabellen (In- Inst. 22: 73 pp. secta: Hymenoptera: Apoidea: Halicti- Michener, C. D., R. J. McGinley, and B. N. Dan- dae: Dufoureinae). Senckenb. Biol. 64: forth 313-379. 1994. The bee genera of North and Central 1993. Ibid. Dritter Nachtrag. Linzer Biol. America (Hymenoptera: Apoidea). Beitr. 25: 15-42. Washington, D.C.: Smithson. Inst. Eickwort, G. C., P. F Kukuk, and F R. Wesley Press. 1986. The nesting biology of Dufourea no- Roig-Alsina, A., and C. D. Michener vaeangliae (Hymenoptera: Halictidae) 1993. Studies of the phylogeny and classifi- and the position of the Du- cation of long-tongued bees (Hyme- systematic noptera: Apoidea). Univ. Kansas Sci. foureinae based on behavior and devel- Bull. 55: 123-173. opment. J. Kansas Entomol. Soc. 59: Urban, D., and J. S. Moure 103-120. 1993. Ceblurgus longipalpis gen. e sp. n. Pri- Iwata, K., and S. F Sakagami meiro representante de Dufoureinae do 1966. Gigantism and dwarfism in bee eggs in Brasil (Hymenoptera, Halictidae). An. relation to the modes of life, with notes Acad. Bras. Cienc. 65: 101-106. on the number of ovarioles. Jpn. J. Warncke, K. Ecol. 16: 4-16. 1979. Beitrage zur Bienenfauna des Iran: 3. Michener, C. D. Die Gattung Rhophites Spin., mit einer 1965. Generic review of the Dufoureinae of Revision der Westpalaartischen Arten the Western Hemisphere (Hymenop- der Bienengattung Rophites Spin. Boll. tera: Halictidae). Ann. Entomol. Soc. Mus. Civ. Stor. Nat. Venezia 30: 111- Am. 58: 321-326. 155. Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates and Bulletin published during the last five years are available at World Wide Web site http://nimidi.amnh.org. Or address mail orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024. TEL: (212) 769-5545. FAX: (212) 769-5009. E-MAIL: [email protected]

( This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).