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

Home , Biastini, Neopasites

AMERICAN MUSEUM Norntates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3005, 11 pp., 13 figures May 22, 1991

Biology and Larvae of the Cleptoparasitic Bee Townsendiella pulchra and Nesting Biology of Its Host Hesperapis larreae (Hymenoptera: Apoidea)

JEROME G. ROZEN, JR.1 AND RONALD J. McGINLEY2

ABSTRACT Townsendiella (Townsendiella) pulchra Craw- The first instar ofthe cleptoparasite, with most of ford (Anthophoridae: Nomadinae) is a cleptopar- the typical features of first-stage larvae of the No- asite of Hesperapis (Panurgomia) larreae Cock- madinae, was recovered from the host nest, and erell (Melittidae: Dasypodinae), the first confirmed its chorion was extracted from the host cell wall. host association of a Townsendiella species. Hes- A late-stage larva shares features with last instars perapis larreae is a ground-nesting, solitary bee. ofother nomadine taxa, especially Neopasites and Details of its nest structure, provisioning, and de- Neolarra. Two dissected females of T. pulchra re- velopment are presented and compared with vealed 7 ovarioles per ovary, with totals of 5 and known information regarding other Dasypodinae. 4 mature oocytes per individual.

INTRODUCTION On a brief field trip to Arizona, we en- perapis (Panurgomia) larreae Cockerell (Me- countered adults ofTownsendiella (Townsen- littidae: Dasypodinae). This paper (1) de- diella) pulchra Crawford (Anthophoridae: scribes and discusses the nesting biology of Nomadinae) in abundance as they searched the host in relation to other dasypodines, (2) the ground for host nests. Investigating the presents information on the mode of para- area further, we identified the host as Hes- sitism of T. pulchra, and (3) describes the

' Curator, Department of Entomology, American Museum of Natural History 2Chairman and Curator, Department of Entomology, National Museum of Natural History, Smithsonian Insti- tution, Washington, D.C.

Copyright © American Museum of Natural History 1991 ISSN 0003-0082 / Price $1.70 2 AMERICAN MUSEUM NOVITATES NO. 3005 first- and late-stage larvae of the cleptopar- tified at this locality within a week, and four asite. Little has been recorded about the be- were traced to cells. havior, ecology, and ontogeny ofthe two gen- Although both sites contained a large num- era to which these species belong. ber ofplant species characteristic ofthe xeric Michener (1936) found Townsendiella (Er- Southwest (Prosopis, Cercidium, Sphaeral- emopasites) californica Michener "flying close cea), creosote bush, Larrea tridentata, dom- over the ground in a colony of' Hesperapis inated the region. It was the pollen source for (Hesperapis) rufipes (Ashmead), and R. R. Hesperapis larreae, as well as for Centris, Snelling (personal commun.) believes that Trachusa (Heteranthidium), Ancylandrena Conanthalictus may be the host of the third larreae (Timberlake), and Colletes. Among known species of the genus, Townsendiella cleptoparasites, several species ofEpeolus and (Xeropasites) ruJiventris Linsley. The associ- one each of Triepeolus and Sphecodes were ation of T. pulchra with H. larreae is the first collected while they flew in search of host host association of Townsendiella confirmed nests. Several specimens of the rare Hexe- by the recovery of immature cleptoparasites peolus rhodogyne Linsley and Michener were from host nests. netted at both sites. Of all cleptoparasites, Immatures and nest components associ- Townsendiella was the most common when ated with this study are in the American Mu- nests were found. seum of Natural History, and adults (both in The Larrea plants were in maximum bloom fluid and dried) are there and in the Smith- at both sites at first discovery. At last visit sonian Institution. on May 11, 1990, the Larrea at 1 1 mi southwest of Congress was almost totally past ACKNOWLEDGMENTS bloom and at 8 mi northwest of Wickenburg We express our appreciation to Barbara L. it was declining rapidly. We were unable to Rozen for her assistance in searching for nests discover Townsendiella adults at either site of Hesperapis larreae. We thank G. I. Stage at that late time. and R. R. Snelling for permitting us to read Both sites occupied gently sloping ground the biological section oftheir revision ofHes- in the predominantly creosote bush desert. perapis. Information therein added substan- All Hesperapis nests were on surfaces that tially to our understanding of the behavior sloped less than 10°. Although creosote bush- and ecology of a number of species in the es were widely distributed at both sites, genus. Snelling, Los Angeles County Muse- searching Townsendiella tended to be con- um of Natural History, also confirmed the centrated, i.e., the cuckoo bees and presum- identification of this Hesperapis. Richard H. ably host nests were not found throughout Kruzansky provided the analyses ofthe nest- the entire area occupied by creosote bushes. ing soils. At the Congress site most Townsendiella were The encountered within a radius of 30 m, and at manuscript has been critically re- the Wickenburg site, within a radius of ap- viewed by Roy R. Snelling and Byron Al- proximately 50 m. At the Wickenburg site, exander. nests of Hesperapis larreae, though few in number, were widely distributed within the DESCRIPTION OF SITES 50 m radius. The soil at both sites was a sandy loam We first discovered Townsendiella pulchra derived from degraded igneous rock. At the flying in association with Hesperapis larreae Congress site it consisted of 62% sand, 29% at 11 mi southwest ofCongress, Yavapai Co., silt, and 9% clay (coarse fragments >2 mm Arizona, on April 29, 1990. The single nest = 10.8%); at the Wickenburg site, 70% sand, of H. larreae discovered there (fig. 1) was 21% silt, and 9% clay (coarse fragments >2 excavated on the following day. A second site mm = 8.0%). (fig. 2), approximately 8 mi away, was iden- The soil appeared dry at the surface and at tified May 3, 1990, at a place in Yavapai Co. cell level. The surfaces generally consisted of 8 mi northwest ofWickenburg, Maricopa Co. loose material, and the subsurface was irreg- Six nest entrances of Hesperapis were iden- ularly consolidated. Large rocks were absent, 1991 ROZEN AND McGINLEY: CLEPTOPARASITIC BEE AND ITS HOST 3

i~*' : - - I\- AW;¶ A_

Rkr~~~~~~~~-

-.. I. 4 - ..2 AA -

Figs. 1, 2. Nesting sites of Hesperapis larreae. Top. 11 mi southwest of Congress, Yavapai Co., Arizona. Bottom. 8 mi northwest of Wickenburg, Maricopa Co., Arizona. 4 AMERICAN MUSEUM NOVITATES NO. 3005 and plant roots were uncommon. Nests were scended in a meandering and irregular path on nearly barren surfaces between creosote (figs. 4-6). The descent of one seemed ver- bushes except one which was at the base of tical, but with all others the path was at about a bush. a 450 angle. Burrow walls were unlined and irregular. BIOLOGY OF HESPERAPIS LARREAE Main burrows were open, at least for the most part, and often revealed side branches The nest entrances of Hesperapis larreae, (some soil-filled), particularly near the sur- although usually hidden in small clumps of face. We are uncertain whether some ofthese low (less than 8 cm high), dry herbaceous side tunnels may have been created by other plants, were not extensively shaded by other insects, but other side tunnels almost cer- vegetation including creosote bushes. tainly were the work of Hesperapis, perhaps Finding nest entrances was difficult be- aborted attempts offemales to find their way cause they tended to be hidden under low- through consolidated soil. growing, dried plants and they did not pos- Although all traced side tunnels ended sess obvious tumuli. We identified only two blindly (i.e., did not connect to cells), unde- of the seven nest entrances by observing re- tected filled side tunnels, probably branching turning Hesperapis females entering them. We deep in the ground, led to cells. Six nests discovered all other nests because we ob- contained two sets of cells in linear series, served one or more Townsendiella females and one nest contained three such sets. In all perched near a potential nest entrance or re- five nests, the older cells were closer to the turning repeatedly to one spot. Townsendiella surface than the newer ones. Within a single females, when searching for nests, custom- series the newest cell was closest to the tunnel arily investigated these low clumps of dried leading from it; the oldest cell was at the rear vegetation, which suggested that most nests of the series. were secreted this way. We saw one returning Of the 11 separate series encountered in Hesperapis female enter its nest swiftly, with- five nests, 1 consisted offour cells, 2 ofthree out slowing and weaving. This observation cells, 6 of two cells, and 2 of one cell (one of and our inability to discover more than two these was a complete series; the other rep- females entering or leaving nests during a resented the end cell ofa series to which other week of study by two persons suggest that cells presumably would have been added). comings and goings may always be rapid, per- The depths (cm) of the series in each of five haps a behavioral adaptation that lessens the nests were: 21, 23, 24; 22, 26; 24, 28; 26, 29; chance ofnest identification by predators and and 27, 32. Hence, in these five nests the parasites. within-nest differences between the depth of Hesperapis and Townsendiella were active cell series ranged from only 1 to 5 cm. during the middle part of the day (9:00 a.m. Cells in series (fig. 3) were virtually end to to 4:00 p.m.), and from collection records end, there being no filled or unfilled burrow their seasonal activity seems restricted most- separating them. In one typical case the cen- ly to the spring blooming of Larrea. How- ter of the cell closure was 1.0 mm from the ever, Hurd and Linsley (1975) reported a fe- rear ofthe cell in front ofit. Nearly horizontal male ofHesperapis larreae collected on Larrea cells tilted toward the rear 10 to 20°. on September 9 in Otero Co., New Mexico, Cells were symmetrical around their long and a female in the American Museum of axis, that is, the cell ceiling was not vaulted Natural History was captured on Larrea on more than the floor. Their maximum di- August 17 in Cochise Co., Arizona. ameter ranged from 5.0 to 5.3 mm (N = 5), No nest was occupied by more than one and the length from rear to center of closure female, and males were not found in nests. was 6.5 to 7.1 mm (N = 3). Their large di- NEST ARCHITECTURE: Main burrows were ameter in relation to their length made them irregular in diameter, probably because ofthe appear unusually globular compared with cells uneven consolidation of the substrate, and of many other bees. The diameter of a still ranged in diameter from 3.0 to 4.0 mm (N open cell entrance was only 2.7 mm, which = 6). Entering the ground obliquely, they de- accentuated the robust appearance ofthis cell. 1991 ROZEN AND McGINLEY: CLEPTOPARASITIC BEE AND ITS HOST 5

CELL CLOSURE CELL WALL

a,.,_/ 4

aG, G 5 3

SIDE TUNNEL

GO 6

0OC)>--

Figs. 3-6. Nest elements of Hesperapis larreae. 3. Cell series consisting of a closed cell and an open one, side view. 4-6. Examples of nests, side view. Scale lines refer to figures 3 (= 5 mm) and 4-6 (= 10 cm), respectively. Walls (approximately 0.5 to 1.0 mm thick) often be carefully teased from the substrate of recently constructed cells, though fragile, for microscopic viewing. However, they could tended to be slightly more consolidated than not be preserved unless extracted with sub- the substrate. Small sections of walls could strate attached and then impregnated with a 6 AMERICAN MUSEUM NOVITATES NO. 3005

mixture ofwhite (casein) glue and water. Fresh One preliminary provision sphere in an open walls were not waterproof and immediately cell measured 2.3 mm in diameter. Fully absorbed water droplets. They were smooth, formed provisions were approximately dull on the surface, and gave no hint as to spherical, uncoated, uniformly moist, and what substance (if any) accounted for their measured 3.1 to 3.5 mm in diameter (N = slightly greater strength than the substrate. 10). These spheres rested on the cell floors. The soil of the cell wall was uniformly fine- The curved eggs ofHesperapis larreae were grained in sharp contrast to the irregular par- placed on the top of the provisions in the ticle size of the surrounding substrate. Hes- median longitudinal cell plane; their more perapis females are apparently capable of rounded anterior ends faced the cell closures sorting out fine particles to construct the thin and the tapering posterior ends pointed to- walls. ward the rear of the cell. The chorion was The cell walls from which mature (or near- smooth, shiny, and transparent over the gray- ly mature) larvae were removed were mark- ish-white egg. One egg measured 1.75 mm edly different from the walls (and closure) of long and 0.45 mm in maximum diameter. newly constructed cells. The altered cell walls Young larvae apparently crawl over the sur- are described below under Provisioning and face of the food sphere as they feed, for they Development. possessed paired ventrolateral tubercles on Cell closures exhibited a number of un- the thoracic and most abdominal segments. usual features. Like the inner surface of cell Although older larvae cradled depleted food walls, they consist ofuniformly fine material. masses with their somewhat elongate bodies, They were deeply concave on the inside, and other observations on larval behavior are re- the concavity was cone-shaped (fig. 3). The quired to understand feeding activities surface of the cone was nearly continuous throughout the larval stage. Several larvae with the cell wall, and the two surfaces shared defecated soon after feeding and one survived the same texture. The center point ofthe cone to become a white, quiescent postdefecating was almost pitlike when viewed from inside form within a week after capture, indicating the cell. In contrast to the cell closures ofmost that the species probably overwinters in this ground-nesting bees, the inner surface of the state. closure was nearly smooth, i.e., there was no Either after feeding or perhaps shortly be- distinct spiral contouring of this surface. fore finishing, large larvae produced a sub- However, by observing these closures in stance (source unknown but perhaps anal or oblique light through a microscope, a vague salivary) that impregnated the cell wall and spiral structure could be detected in some closure so that these structures became strong, cases. Closures ofnewly constructed cells were took on a dark "wet" appearance, glistened fragile, consisting ofpoorly consolidated main places, and became water-retardant. This terial throughout, but, as with cell walls, clo- substance possessed no pollen grains (at least sures of cells containing older larvae were at first) and a section ofimpregnated cell wall impregnated with some substance as is de- did not "dissolve" when submerged in water scribed below. Closures of three cells ranged for several hours. Because of their new from 3.2 to 3.5 mm in maximum diameter strength, cell walls and closures in this con- on the inside. The depth ofthe concave inner dition were extricated intact from the subsurface of four cells ranged from 1.5 to 1.8 strate. mm. One larva had started to apply fecal pellets PROVISIONING AND DEVELOPMENT: Fe- over the inner surface of the cell wall before males transported Larrea pollen and nectar we collected it. Larvae did not produce co- as large moist masses on their hind legs. Early coons (unless the glistening substance de- loads were apparently shaped and deposited scribed above represents a modified salivary as small spheres in the cells, as was also re- secretion). Certainly no silk threads were in ported for Hesperapis trochanterata (Rozen, evidence in cells, and the recessed labiomax- 1974; but see data regarding early provision illary region of larval Hesperapis character- load of Dasypoda plumipes in Lind, 1968). izes noncocoon-spinning larvae. 1991 ROZEN AND McGINLEY: CLEPTOPARASITIC BEE AND ITS HOST 7

BIOLOGY OF pasites; see description of late-stage larva); TOWNSENDIELLA PULCHRA long paired labral tubercles present; antennae nonprotuberant; mandibles elongate, curved, Although the two Hesperapis larreae nest- sharply pointed; maxillary palpi conspicu- ing sites were separated by only 8 mi, the ous; labial palpi indistinguishable; most ab- Townsendiella pulchra population at 1 1 mi dominal segments (but not thoracic seg- southwest of Congress was further advanced ments) with paired ventrolateral tubercles, than the one at 8 mi northwest of Wicken- and terminal abdominal segment not dis- burg. At the Congress locality no males of tinctly forked but with single median elongate Townsendiella were encountered, and the projection and with two less elongate lateral outer edge of the anterior wings of collected projections. A detailed description of this females were nearly totally worn away. At the specimen will be prepared later in conjunc- Wickenburg site males were not uncommon, tion with a proposed comparative examina- and nearly half of the females still retained tion of the first instars of the Nomadinae by most of the outer wing margins. JGR. At both localities females flew moderately The two other Townsendiella larvae were slow, meandering paths close to the ground probably second instars and in one cell the and halted or slowed near clumps of dried cast skin and head capsule of the first instar herbs as they sought Hesperapis nests. At the were recovered. In the second instar, the head Wickenburg site, the flight of males could be capsule was no longer pigmented, mandibles identified because it was faster and less me- were short, and the projecting venter of ab- andering though close to the ground. We ob- dominal segment IX had developed. The la- served no matings. bral tubercles, though shorter, were still in As stated above, we were able to observe evidence as were the ventrolateral abdominal Townsendiella females hovering around one tubercles. point on the ground or perching near it. Often A single large Townsendiella larva was re- a female would wait several centimeters from covered from the fourth cell which still con- an entrance, and then once in a while fly to tained a large provision mass. The larva was the entrance, inspect it briefly, and then re- superficially so similar to host larvae of sim- turn to her perch. Two or three females be- ilar size that JGR misidentified it at first. haved this way simultaneously at one spot Only after he saw its short mandibles and without interacting with (e.g., chasing) one distinct, if small, labral tubercles did he cor- another. When we had the opportunity later, rectly recognize it. The specimen is a penul- we were able to identify Hesperapis nest en- timate instar or a young last instar. Which- trances at such places. In one case we cap- ever instar it may be, its features suggest tured both a female Townsendiella and a Hes- relationships between the Townsendiellini perapis female as they emerged from a single and some other nomadine tribes. For this nest. reason it is described, illustrated (figs. 7-12), Although Townsendiella was moderately and compared with last-stage larvae of cer- abundant, particularly at the Wickenburg site, tain other Nomadinae at the end of this pa- we recovered Townsendiella immatures from per. only four cells. However, a good many ofthe Also in one ofthe parasitized cells, the egg other cells were damaged in excavation so insertion hole (0.175 mm in diameter) of a that immatures of neither host nor parasite Townsendiella was discovered in the cell wall. were discovered, and a few cells were open When the section of the wall containing this or freshly closed. hole was submerged in water, the cast chorion One ofthe Townsendiella larvae was a first was retrieved. The transparent colorless cho- instar with the following characteristics: head rion with the hole at one end was 1.0 mm capsule elongate, sclerotized, moderately pig- long. mented, semiprognathous; vertex projecting Two freshly killed females of Townsen- immediately in front of sharply constricted diella were dissected so that their ovaries posterior margin (also characteristic of Neo- could be examined. Each had seven ovarioles 8 AMERICAN MUSEUM NOVITATES NO. 3005 per ovary (i.e., 7:7). One had a total of 5 Although available nesting information is mature oocytes,3 the other 4 mature oocytes. too scanty to provide an overview ofthe nest- Egg index (based on the definition of mature ing biology of the Dasypodinae, certain be- oocyte in footnote 3) (i.e., egg length oflargest havioral features of Hesperapis larreae are mature oocyte divided by the mean meso- noteworthy. Linear cell arrangement has not somal width, Alexander and Rozen, 1987) been reported for any other Hesperapis, all was 0.69. Oocytes (fig. 13) measured 1.05- of which construct single cells at the ends of 1.18 mm (N = 2) and were slender, rounded branches. Cell series, however, were noted for anteriorly, and somewhat tapering posteri- Capicola braunsiana. The large cell diameter orly. They exhibited no operculum or other relative to length is not characteristic ofmost unusual structures except that each bore a bees but has also been reported for H. pel- small raised area, nipplelike and not hook- lucida and perhaps other species in the genus. shaped, at the anterior end. The chorion was Walls of fresh cells of Hesperapis larreae smooth, nonreticulate, and shiny. are smooth, dull, and composed ofsoil much finer and more evenly grained than the sub- DISCUSSION OF BIOLOGY strate. They contrast sharply with the rough Nesting biologies of the following species unlined cells walls ofH. trochanterata. Stage of Dasypodinae have been studied: and Snelling (Ms) noted that the walls of H. Dasypodini: pellucida "did not seem to be particularly Dasypoda plumipes Muller (1884) smoothed or compacted." They reported that Panzer those of H. ilicifoliae seemed to be "lined Malyshev (1927) with fine soil," perhaps smoothed with the Lind (1968) tongue and "impregnated with saliva which Hesperapis (H.) rufipes Stage and Snelling (MS) made it harder and darker than the adjacent (Ashmead) matrix ... it readily absorbed drops ofwater H. (H.) trochanterata Rozen (1987) though not as quickly as did the adjacent ma- Snelling trix." At one site, "the cell walls were so strong H. (H.) unnamed spe- Stage and Snelling (MS) cies that it was possible to remove some of them H. (Amblyapis) ilicifoli- Stage and Snelling (MS) intact." Stage and Snelling stated that the cell ae (Cockerell) walls of H. rufipes were like those of H. iHi- H. (Panurgomia) lar- present paper cifoliae and were apparently smoothed in the reae Cockerell same way. Hence in the genus some species H. (P.) nitidula Cocker- Stage and Snelling (MS) (trochanterata and pellucida) have rough cell ell walls not coated with fine soil, whereas others H. (P.) pellucida Cock- Stage and Snelling (MS) (larreae, ilicifoliae, and rufipes) have smooth, erell fine-textured walls. The hardness and water- H. (P.) regularis (Cres- Burdick and Torchio absorbing qualities of the smooth walls will son) (1959) MacSwain et al. (1973) have to be restudied in light of the fact that Stage and Snelling (MS) we now know water absorption, strength, and H. (P.) unnamed species Stage and Snelling (MS) even color in a cell ofH. larreae are modified Capicola braunsiana Rozen (1974) by actions of the mature or nearly mature Friese larva. Consequently, wall characteristics re- Sambini: ported for other species ofHesperapis as well Haplomelitta ogilviei Rozen (1974) as for Capicola braunsiana and Haplomelitta (Cockerell) ogilviei may not have been the result of the adult activity. Cell walls of Dasypoda plu-

3Mature oocytes as here defined were those that had taken on their final size and shape and had developed a visible, slightly shiny chorion. They were quite distinctive and presumably are Iwata's (1955) Category A oocytes. Slightly larger and also smaller oocytes without nutritive cells but also without chorions were also present in Town- sendiella. These oocytes were a less well-defined group and probably conform to Iwata's Category B. Alexander and Rozen (1987), in their study of oocytes of parasitic bees, grouped Categories A and B together because they were difficult to distinguish. This was not a problem with Townsendiella, perhaps because specimens were fresh. 1991 ROZEN AND McGINLEY: CLEPTOPARASITIC BEE AND ITS HOST 9

mipes are like the newly constructed walls of it bears such a strong resemblance to the last- H. larreae, i.e., dull, smooth, and absorptive. stage larvae of Neopasites cressoni Crawford Cell closures of Hesperapis larreae were and Neolarra pruinosa Ashmead (Rozen, deeply concave on the inside and were es- 1966) that comparisons with these taxa may sentially smooth, i.e., without a distinct spiral be valid and potentially revealing of rela- structure. Similar closures have been report- tionships. ed for Capicola braunsiana and Haplomelitta DIAGNosIs: Like mature or nearly mature ogilviei. Closures ofDasypodaplumpies, while larvae of most other Nomadinae, this larva deeply concave, exhibit a distinct spiral. possesses small but distinct paired labral tu- The behavior of Townsendiella pulchra bercles, recessed labiomaxillary region (indi- conforms with that of other Nomadinae. Fe- cating absence of cocoon spinning), greatly males search for host nests and almost cer- reduced or lost posterior thickening of the tainly enter them when host females are head capsule, posterior tentorial pits that are away-as indicated by Townsendiella fe- not structurally tied into the posterior thick- males waiting or repeatedly returning to nest ening or to the hypostomal ridge, mandibles entrances. We are uncertain whether waiting that are so short that their apices do not meet cleptoparasites are an indication that the host when closed, and labial palpi that are so re- female is in the nest. Possibly a cleptoparasite duced that each is represented only by a single cannot detect the presence of a host female sensillum. in the nest unless she sees the host depart. Neopasites cressoni, Neolarra pruinosa, and The single chorion embedded in the cell Townsendiella pulchra, as mature or nearly wall confirms that, like other Nomadinae, mature larvae, are the only Nomadinae which Townsendiella inserts its egg there, presum- have a frontoclypeal area that is enlarged and ably while the host female is away foraging. overhangs a reduced labrum and which have The Townsendiella first instar is hospicidal, a protruding venter to abdominal segment X. as are those ofall known Nomadinae. It pos- Townsendiella pulchra can apparently be dis- sesses the major characteristics of the others tinguished from the other two taxa because with the interesting exceptions ofa differently it alone has its parietals so enlarged that the modified abdominal apex and the presence head capsule appears to have a median de- of paired ventrolateral abdominal tuber- pression when viewed from the front (fig. 8). cles-both features presumably associated TOTAL LENGTH: Approximately 4.0 mm. with locomotion. Neither of these features HEAD (figs. 8, 9): Integument of head cap- have been reported for other Nomadinae, but sule with scattered, minute, nonsetiform sen- first instars ofthe subfamily are poorly known. silla; maxilla spiculate, especially strongly so Specifically, first instars of Neopasites and on side; epipharynx and hypopharynx non- Neolarra have yet to be examined. Consid- spiculate. Integument unpigmented except for ering the similarities in the following descrip- anterior part of hypostomal ridge and man- tion between the late-stage larva of Town- dibular bases. sendiella pulchra and last instars of Head size normal in relation to rest ofbody; representatives of Neopasites and Neolarra, head capsule much wider than length mea- first instars ofthe latter two taxa might clarify sured from top of vertex to level of lower the phylogenetic relationships of the Town- margin of clypeus as seen in frontal view. sendiellini, Biastini, and Neolarrini. Inter- Tentorium thin but complete except appar- estingly, early instars of the host also possess ently for anterior arms; anterior tentorial pits paired ventral tubercles on the abdominal small, low on face; posterior tentorial pit segments as well as on the three thoracic seg- small, in line with hypostomal ridge and ap- ments. parently lying in unpigmented thin posterior LATE-STAGE LARVA OF extension of ridge; posterior pit near poste- TO WNSENDIELLA PULCHRA rior margin of head but posterior margin of Figures 7-12 head without ridge and therefore difficult to define; hypostomal ridge moderately devel- Although we are uncertain as to whether oped anteriorly but almost disappearing near the specimen described here is a last instar, posterior tentorial pit; ridge without dorsal 10 AMERICAN MUSEUM NOVITATES NO. 3005

Figs. 7-13. Townsendiella pulchra. 7. Last-stage larva, lateral view. 8, 9. Head of same, frontal and lateral views. 10-12. Right mandible of same, dorsal, inner, and ventral views; sensilla and spicules not shown. 13. Mature oocyte, side view, anterior end at top. Scale lines (both = 1 mm) refer to figures 7 and 13, respectively. ramus; pleurostomal ridge weakly developed; labral disc; labrum perhaps faintly emargin- epistomal ridge weakly developed laterad of ate apically; epipharynx not produced ven- anterior tentorial pits, absent between pits; trally. external epistomal depression not evident. Mandibles (figs. 10-12) short so as not to Parietal bands faint but evident. Antennal meet apically, stout at base, tapering gradu- prominences not developed; antennal disc ally to apices, and with outer surface spicu- moderate in size; antennal papilla small but late; mandible normally curved, not nearly pronounced, nearly as long as basal diameter; as straight as in Neopasites; adoral surface each papilla bearing three sensilla. Vertex perhaps somewhat short compared with out- rounded as seen from side and evenly grading er surface, but not nearly as short as described into frontal area; parietals enlarged so that, for Neopasites (Rozen, 1966); adoral surface when head viewed from front (fig. 8), pari- at base not produced; mandibular apex bear- etals bulge on either side of midline; clypeus ing sharp teeth on both dorsal and ventral normal in length; frontoclypeal area in lateral edges; apex appearing rotated so that dorsal view (fig. 9) bulging beyond base of labrum dentate edge adoral (as also apparently true as is characteristic oflast instars ofNeopasites in Neopasites). Labiomaxillary region re- and Neolarra (Rozen, 1966). Labrum small, cessed and fused although apex of maxillae recessed; labral sclerite not evident; small but distinct from labium; maxillary sclerites not distinct, paired labral tubercles arising from discernible; maxillary palpus small, shorter 1991 ROZEN AND McGINLEY: CLEPTOPARASITIC BEE AND ITS HOST 11 than basal diameter, and less pronounced than REFERENCES antennal papilla; galea absent. Labium not divided into prementum and postmentum; Alexander, B., and J. G. Rozen premental 1987. Ovaries, ovarioles, and oocytes in par- sclerite not defined; labial palpus asitic bees (Hymenoptera: Apoidea). not produced, represented by single sensil- Pan-Pacific Entomol. 63: 155-164. lum. Salivary opening a small hole, without Burdick, D. J., and P. F. Torchio lips; hypopharynx not projecting as distinct 1959. Notes on the biology ofHesperapis reg- lobe, but hypopharngeal region extending ularis (Cresson) (Hymenoptera: Melitti- forward above salivary opening as seen in dae). J. Kansas Entomol. Soc. 32: 83- lateral view, about as in Neopasites cressoni 87. (Rozen, 1966: fig. 74); boundary between la- Hurd, P. D, Jr., and E. G. Linsley bium and hypopharynx poorly defined. 1975. The principal Larrea bees ofthe south- BODY: Integument without setae but many western United States (Hymenoptera: areas Apoidea). Smithsonian Contrib. Zool. finely, distinctly spiculate; abdominal 193: 74 pp. segment X not spiculate ventrally. Body Iwata, K. without spines or sclerotized tubercles. Body 1955. The comparative anatomy ofthe ovary form slender (as might be expected in larva in Hymenoptera. Part 1. Aculeata. Mu- that was not fully fed); intrasegmental lines shi 29: 17-34. not evident; body segments without dorso- Lind, H. lateral tubercles or other tubercles or swell- 1968. Nest provisioning cycle and daily rou- ings except venter of abdominal segment IX tine of behavior in Dasypoda plumipes distinctly produced, similar to same regions (Hym., Apidae). Entomol. Medd. 36: of Neopasites cressoni and presumably Neo- 343-372. larra Malyshev, S. I. pruinosa (Rozen, 1966: figs. 72, 79); 1927. The nesting habits of Dasypoda. Trav. abdominal segment X small, without dorsal Soc. Natural. Leningrad 57: 123-146. ridges, and seemingly positioned somewhat MacSwain, J. W., P. H. Raven, and R. W. Thorp dorsally on IX; anus apparently dorsal. Spira- 1973. Comparative behavior ofbees and On- cles small, subequal in size, appearing similar agraceae. IV. Clarkia bees of the west- to those ofNeopasites cressoni (Rozen, 1966: ern United States. Univ. California Publ. fig. 73) although with fewer subatrial cham- Entomol. 70: 80 pp. bers. Sexual characters unknown. Michener, C. D. MATERLAL STUDIED: One late-stage larva, 1936. Some California bees, Townsendiella 8 mi northwest Wickenburg, and Hesperapis (Hymen.: Apoidea). En- Maricopa Co., tomol. News 47: 181-185. Arizona, May 8, 1990 (J. G. Rozen) from nest Muller, H. of Hesperapis larreae. 1884. Ein Beitrag zur Lebensgeschichte der REMARKs: Comparison of this larva with Dasypoda hirtipes. Verh. naturhist. Ve- larvae of other Nomadinae is hampered be- rein d. Rheinlande, 5. Folge, 1: 1-52. cause it is not mature and not even certainly Rozen, J. G. a last instar. For that reason and because a 1966. The larvae of the Anthophoridae (Hy- phylogenetic reevaluation of the mature lar- menoptera, Apoidea). Part 2. The No- vae of the entire subfamily needs to be un- madinae. Am. Mus. Novitates 2244: 38 dertaken, we will not attempt a larva-based PP. detailed assessment ofrelationships ofTown- 1974. The biology oftwo African melittid bees (Hymenoptera, Apoidea). J. New York sendiella with other Nomadinae here. The Entomol. Soc. 82: 6-13. close similarity between Townsendiella and 1987. Nesting biology and immature stages of Neopasites (Biastini) is reflected in the above a new species in the bee genus Hesper- description and many of these characters are apis (Hymenoptera: Apoidea: Melitti- probably synapomorphies. Similarities with dae: Dasypodinae). Am. Mus. Novi- larval Neolarra are also evident although not tates 2887: 13 pp. so pronounced. Whether these similarities are Stage, G. I., and R. R. Snelling synapomorphies or the result ofconvergence MS. A revision of the Nearctic Melittidae: requires further study, particularly of Neo- the subfamily Dasypodinae (Hymenop- larra. tera: Apoidea). Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, New York 10024. L

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER.