Review of Perdita subgenus Procockerellia Timberlake (Hymenoptera, Andrenidae) and the first Perdita gynandromorph

Abstract A systematic study of Perdita subgenus Procockerellia Timberlake and the related subgenus Allomacrotera Timberlake results in the synonymy of the latter with the former, and two specific synonymies: Perdita (Hexaperdita) glamis Timberlake is a junior synonym of Perdita (Procockerellia) stephanomeriae Timberlake, while Perdita (Procockerellia) brachyglossa Timberlake is a junior synonym of Perdita (Cockerellia) imbellis Timberlake. Perdita (Procockerellia) moldenkei Timberlake is moved to subgenus Cockerellia Ashmead. A revised subgeneric diagnosis and key to the three included species are provided. Diagnoses of species are updated with novel characters; distributions and biological data are expanded. A gynandromorph of P. (Procockerellia) moabensis Timberlake, the first known in the genus Perdita, is reported.


Introduction
The panurgine genus Perdita Smith, 1853 (Hymenoptera: Andrenidae: Panurginae) is the most speciose bee genus in North America with 636 currently recognized species and 127 additional subspecies (Portman et al. 2016b). It is also diverse at the subgeneric level with 17 subgenera currently recognized (Michener 2007). Among these, the subgenera Procockerellia Timberlake, 1954 andAllomacrotera Timberlake, 1960 have had a complicated and intertwined taxonomic history.
Here, we assess the taxonomic status of these subgenera and revise the included species. Synonymies and changes presented here result in a single subgenus, Procockerellia, containing three species: P. (P.) albonotata, P. (A.) moabensis, and P. (A.) stephanomeriae. These changes reduce the total number of Perdita subgenera to 16 and the number of species to 634. A revised key to species of Procockerellia, and updated species accounts are presented. Lastly, during the course of this study, a gynandromorph of P. moabensis was discovered; its aberrant morphology is described. This specimen is the first described gynandromorph in the genus Perdita.  unknown, but they are presumably ground nesting bees like other species of Perdita. Both P. albonotata and P. moabensis are found throughout the flowering season from spring to fall, suggesting they are multivoltine. Thus, the flight period of Procockerellia matches the bloom period of the genus Stephanomeria, which contains species that collectively bloom from spring to fall (Gottlieb 1972). The paucity of collection events renders the phenology of P. stephanomeriae unclear.
Remarks. The relationship between Procockerellia, Allomacrotera and closely-related subgenera is ambiguous, though Prockerellia is clearly a member of the monophyletic group made up of the subgenera Callomacrotera, Cockerellia, Hexaperdita, Penta- perdita and Xeromacrotera Timberlake, 1954(Danforth 1996. The reduced number of maxillary palpi suggests an affinity to the subgenus Pentaperdita, which has the maxillary palpi 5-jointed (Timberlake 1954). Danforth (1996) suggested a close relationship to Cockerellia, though the species of Procockerellia also bear a general resemblance to the monotypic subgenus Xeromacrotera, which also has an uncertain phylogenetic relationship (Portman et al. 2016a).
The many similarities in scopal hairs ( Fig. 1), morphology, genitalia (Fig. 2), apical sterna (Figs 3,4), coloration and general gestalt (Figs 5, 6) all support a close relationship between Procockerellia and Allomacrotera that does not justify different subgenera. The structural similarities between S6, S7, and S8 in the males of P. albonotata and P. moabensis suggest that these are sister species, which would render Allomacrotera paraphyletic. In particular, P. albonotata and P. moabensis have S6 and S7 deeply divided and emarginate and S6 with pronounced lateral hair tufts; these characters are lacking in P. stephanomeriae (Figs 3,4). Timberlake (1980) and Michener (2007) also reported that Allomacrotera lacked lateral furrows in the flanks of the pronotum, but all three species contained in the two subgenera have the flanks of the pronotum moderately impressed.  The close relationship between P. albonotata and P. moabensis suggests two possible solutions to fix the classification of Procockerellia and Allomacrotera. Either (1) P. moabensis should be moved from Allomacrotera to Procockerellia, or (2) Allomacrotera and Procockerellia should by merged. If P. moabensis were to be moved to Procockerellia, it would eliminate the sole defining character of Allomactera (bifurcate hind tarsal claws in the male) because both P. moabensis and P. stephanomeriae share this character, while P. albonotata lacks it. In addition, the only remaining character unique to Allomacrotera would be the 3-jointed maxillary palpi. However, we agree with Timberlake (1954) that the reduction of maxillary palpi is more important for classification than the specific number of palpi. Indeed, a similar pattern can be seen in the Halictoides group of subgenus Perdita sensu stricto, which includes incredibly similar species in which the maxillary palpi collectively range in number from one to five (Timberlake 1958). Therefore, we have chosen to synonymize Allomacrotera with Procockerellia due to the shared characters of the corkscrew-shaped scopal hairs ( Fig. 1) and keel-shaped apical process of S8 (Fig. 3).
The species of Procockerellia are distinctive in Perdita due to their unique scopal hairs, which are especially long, dense, and tightly corkscrew-shaped, appearing crimped under all but the highest magnification ( Fig. 1). This type of scopal hair morphology is rare, and to our knowledge only occurs in one other group, the panurgine genus Panurgus Panzer, 1806 (Pasteels et al. 1983). The corkscrew hairs of Procockerellia encircle the hind tibia and basitarsus, and the tips are slightly clavate (Fig. 1C). Scopal hairs are also present on the hind femur and trochanter, though these are minutely branched rather than corkscrew-shaped. Similar to the related subgenera Callomacrotera, Cockerellia, Hexaperdita, Pentaperdita, Xeromacrotera (Danforth 1996), the species of Procockerellia initially pack dry pollen into the scopa and then cap it with pollen that has been moistened with nectar (Timberlake 1954, Norden et al. 1992, Portman and Tepedino 2017. Pollen loads on museum specimens indicate that P. albonotata and P. moabensis cap approximately the last 20% of the pollen load on the anterior face of the hind tibia with moistened pollen. The pollen on the trochanter, femur, basitarsus, and posterior face of the hind tibia are not moistened. The proportion of moist and dry pollen carried by P. stephanomeriae is unknown due to a lack of specimens with full pollen loads. The males of Procockerellia vary greatly in size. Similar to many other Perdita, the male head size increases and becomes more quadrate with larger body size ( fig. 7, Norden et al. 1992, Portman et al. 2016a). However, the distribution of head sizes is continuous, and there are not discrete classes as seen in the Perditini species Macrotera portalis (Danforth 1991). The function of the large, quadrate heads is unknown, though it could be used for inter-male aggression and/or grasping females during mating (Norden et al. 1992, Danforth andNeff 1992). 1 Frons and vertex strongly tessellate and dull (Fig. 6B); S1 medially with small, outflexed apical margin (Fig. 6H)  Apex of hind tibia with small nub above tibial spurs (Fig. 6G); T7 broadly rounded, without a point; hind tarsal claws simple; pronotal collar with prominent rounded nub laterally; metasoma generally with white or yellowish markings (Fig. 6D)  Diagnosis. Both sexes of P. albonotata have five maxillary palpi and the frons and vertex are weakly tessellate and slightly shining. The female generally has the most extensive facial markings in the subgenus, but the extent is highly variable. In the typical form, the face has the following markings: the clypeus is marked with white on the lateral margins and has a medial white band, the white paraocular marks are broadly triangular and reach the level of the antennal sockets, and the supraclypeal area can be dark or with a pair of white spots (Fig. 5A). The mandible is only slightly expanded on the inner margin and has a long apical tooth (Fig. 5D). The pronotal lobe can be white or dark. The white (sometimes yellowish) abdominal bands on T2-T5 are entire and curve towards the apical margins laterally (Fig. 5G).
The male is unique in Procockerellia due to the simple hind tarsal claws and the presence of an apical nub on the anterior face of the hind tibia, just above the tibial spurs (Fig. 6G). In addition, the male has the pygidial plate broadly and evenly rounded, and the discs of terga have small white markings, with the markings generally more pronounced on the apical terga (Fig. 6D).
Diagnosis. Both sexes of P. moabensis have the maxillary palpi 5-segmented and the frons and vertex are strongly tessellate and dull. Additionally, the head and mesosoma often have a more greenish (rather than bluish) cast. The female can be recognized by the relatively reduced facial markings which range from transverse lateral marks (Fig.  5B) to entirely absent. In addition, the mandibles are only slightly expanded medially (Fig. 5E) and the metasomal bands are curved towards the apical margins (Fig. 5H).
The male of P. moabensis is unique in having the apical margin of S1 slightly flexed out medially (Fig. 6H) as well as S8 bifurcate apically (Fig. 3E). Both characters are subtle but diagnostic in the male, as this is the only Perdita species known to have either of these characters. The male can be further distinguished by the strongly tessellate and dull frons and vertex, the evenly rounded pygidial plate, and the bidentate hind tarsal claws (though this may be hard to see).
Distribution. Colorado, Utah and Arizona: Colorado Plateau (Fig. 8B). Phenology. The gynandromorph is almost entirely female, except the left side of the head (Fig. 9), the left foreleg, and the left midleg are male. The face is split cleanly down the middle; the female half has an antenna with 12 antennal segments, whereas the male half has 13 antennal segments. The pronotal collar on the male side is slightly more protuberant than the female side, but not as much as in a typical male. The rest of the thorax, including the wings and hind legs, are entirely female. This unique specimen is the first known gynandromorph in Perdita. Further, this is the first recorded gynandromorph in Panurginae; all previously known gynandromorphs in the family Andrenidae have been restricted to the genus Andrena (Wcislo et al. 2004, Hinojosa-Díaz et al. 2012, although an intersex (blended male and female characters) Acamptopoeum submetallicum (Spinola) has been previously documented (Ramos and Ruz 2013).

Month
Remarks. Timberlake (1980) reported that the male of this species had only four maxillary palpi. Examination of many specimens (including the female holotype) reveals that the species always has five maxillary palpi. Figures 2G-I, 3C, F, 4E-F, 5C, F, I, 6C, F, I, 8B

Perdita (Procockerellia) stephanomeriae Timberlake
Perdita (Procockerellia) stephanomeriae Timberlake, 1954: 404, ♀;Timberlake 1960: 132, ♂. Diagnosis. Both sexes have the maxillary palpi 3-jointed (whereas the other two species of Prockerellia have 5-jointed maxillary palpi) and the frons and vertex are barely tessellate and strongly shining (e.g. Fig. 6C). The transverse dorso-lateral carina on the pronotal collar found in both sexes is distinctive; other Procockerellia have a rounded nub laterally. The female has the face marked with white laterally on the clypeus and a triangular mark on the lateral area (Fig. 5C), similar to lighter females of P. albonotata. The female can be further recognized by the broad median expansion of the mandibles (Fig. 5F) and narrowly interrupted metasomal bands that don't curve to the apical margin laterally (Fig. 5I). The male is unique in having a small point apically on the pygidial plate (Fig. 6I). It can be further distinguished by the bidentate tarsal claws and the lack of light bands on the metasoma (Fig. 6F).
Distribution. Nevada and California: Mojave and Sonoran Deserts (Fig. 8B) Griswold;Pinto Ridge (36.2422 -114.5493): 2 ♂ 5 ♀, 9 Oct 1998, T.L. Griswold. Remarks. As a result of this study, P. stephanomeriae is hereby returned to its original subgeneric assignment, Procockerellia. This species appears rare, especially compared to P. albonotata and P. moabensis, which can be common and locally abundant. Extensive all season sampling conducted in 1998, 2004, 2005 in Clark County, Nevada in the eastern Mojave Desert yielded large numbers of Procockerellia. It is therefore interesting that while P. albonotata was widely distributed and abundant, P. stephanomeriae was rarely detected.
The holotype of P. glamis was examined and found to clearly match P. stephanomeriae. The mouthparts of the holotype of P. glamis are not exposed, which likely led Timberlake (1980) to incorrectly place and describe the species in subgenus Hexaperdita since he could not see the reduced number of palpi. Remarks. Perdita (Procockerellia) brachyglossa was described from a single female specimen. Timberlake (1971) reported that the female had four maxillary palpi. Examination of the holotype revealed four maxillary palpi on one side and one palp on the other. In every morphological character except the palpi, the holotype of P. brachyglossa clearly matches P. imbellis. It seems likely that the mouthparts of the holotype of P. brachyglossa are either damaged or aberrant. Due to a lapsus calami in Timberlake (1980), Perdita albomaculata Timberlake, 1980, Perdita imbellis Timberlake, 1968, and Perdita luculenta Timberlake, 1968 were all referred to as being in subgenus Cockerellula Strand, 1932. They all belong in subgenus Cockerellia.

Perdita (Cockerellia) moldenkei Timberlake, 1980
Perdita (Procockerellia) moldenkei Timberlake, 1980: 14 Remarks. Timberlake's description of P. moldenkei reports the lone type specimen as having five maxillary palpi. However, examination of the type reveals that the specimen clearly has six maxillary palpi on both sides. Timberlake must have miscounted the number of maxillary palpi and described P. moldenkei in the incorrect subgenus as a result. Perdita moldenkei is clearly a member of subgenus Cockerellia, and is hereby moved to that subgenus. Based on the examination of the type of P. moldenkei, we believe that it is a likely synonym of P. verbesinae Cockerell, 1896. However, we have not examined the various syntypes and varieties of P. verbesinae in order to confirm this; examination of the syntypes of P. verbesinae should be addressed in a broader revision of Cockerellia.