The species of the cleptoparasitic (cuckoo) bee genus Epeolus Latreille (Hymenoptera: Apidae) occurring in Canada are revised. A total of 12 species are confirmed, with one additional species (E. ilicis Mitchell) listed as possibly occurring in Canada. Morphological comparisons of primary types and continuous variation within species, in addition to DNA barcode sequence analysis of recently collected specimens from across the range of each species support the following proposed synonymies: E. lanhami Mitchell, syn. n., and E. montanus (Cresson), syn. n., under E. americanus (Cresson); E. gabrielis (Cockerell), syn. n., E. geminatus Cockerell and Sandhouse, syn. n., and E. hitei Cockerell, syn. n., under E. compactus Cresson; E. arciferus Cockerell, syn. n., E. beulahensis Cockerell, syn. n., E. lutzi Cockerell, syn. n., E. lutzi dimissus Cockerell, syn. n., and E. pilatei Cockerell, syn. n., under E. minimus (Robertson); and E. humillimus Cockerell, syn. n., E. rubrostictus Cockerell and Sandhouse, syn. n., E. rufomaculatus Cockerell and Sandhouse, syn. n., and E. tristicolor Viereck, syn. n., under E. olympiellus Cockerell. The synonyms of E. americanus, E. compactus, and E. minimus proposed here were first proposed by Richard L. Brumley in an M.Sc. thesis published in 1965, but have until now not been validated. A dichotomous identification key to the Canadian species is presented, and their biology and life history is discussed and contrasted with that of Triepeolus Robertson and other cuckoo bees.

Epeolus olympiellus Cockerell. Photo by S. McCann

A high proportion (28%) of bees in the family Apidae (Hymenoptera: Apoidea) are cleptoparasites of nest-building bees (Cardinal et al. 2010). Cleptoparasitic (or cuckoo) bees appropriate the pollen food stores collected by females of their host species for their own offspring; the cleptoparasite invades the host nest and lays an egg in the brood cell. Subsequently, the host larva or egg (depending on the type of cleptoparasitic bee involved) is killed. Since female cleptoparasitic bees do not collect pollen to feed their brood, they lack the specialized pollen-carrying scopae characteristic of most nest-building bees. Most cleptoparasites are also wasp-like in appearance, exhibiting reduced hairiness, and typically have black and yellow and/or red colouration.
In the Nearctic region, the cleptoparasitic tribe Epeolini (Subfamily Nomadinae) is represented by Epeolus Latreille, Odyneropsis Schrottky (Griswold and Parker 1999), and Triepeolus Robertson (Robertson 1901). Of these, only Epeolus and Triepeolus occur in Canada, and they are the two most diverse genera in the entire tribe (Rightmyer 2004).

To date, no key to all the Canadian species has been published, although a key to the Epeolini of Ontario, the province with the greatest Epeolus diversity, exists (Romankova 2004). I have seen specimens from all provinces and territories in Canada except Newfoundland and Labrador and Nunavut (Map 14). I have verified locality records for 12 species in Canada (Table 1), but the key provided herein includes a thirteenth (E. ilicis Mitchell), which may occur in southern Ontario, whose Canadian voucher specimens (Romankova 2004) cannot be traced. The purpose of the present study is thus to provide a key to all species that might occur in Canada, and to redescribe them.

Table 1. Epeolus in Canada and associated Colletes host species. The nature of the evidence for all confirmed, hypothesized (based on personal assessment), or presumed (suspected and published) associations is indicated in the Discussion section of the taxonomic treatment of each species. Unless otherwise stated, confirmed associations are based on evidence of oviposition by female Epeolus within a Colletes nest, and hypothesized and presumed associations are based on spatial and temporal co-occurrence.

As the sexes in this genus are for the most part monomorphic (other than for typical sexually dimorphic characters), a single identification key for adult Epeolus species in Canada is presented. The identification key is based on external morphological differences that should be visible in dry, pinned specimens. In addition, species redescriptions of the sex opposite that of the primary type include only the key differences between females and males.

To clarify species limits and to give additional support for new synonymies reported here, the divergence levels in a 658 bp segment of the COI mitochondrial gene (DNA barcode) (Hebert et al. 2003a, b) were used in conjunction with morphology. Barcoding entailed the removal of a leg (the source of genetic material) from a bee for DNA extraction and gene amplification and sequencing at the Canadian Centre for DNA Barcoding (CCDB) in Guelph, Ontario, Canada. Barcode Index Numbers (BINs – automated code numbers given to unique barcode clusters) were assigned to sequences as short as >300 bp, although formal recognition of barcode compliant sequences requires a minimum length of 500 bp (Ratnasingham and Hebert 2007, 2013). To validate species designations of specimens and to check for contamination errors, sequences with unique BINs were compared to one another and to short, non-compliant sequences that clustered with compliant ones in a neighbour-joining (NJ) tree, based on Kimura's two-parameter distance model (Kimura 1980). Cases involving change in taxonomic status always prioritized morphological evidence over DNA barcoding, and barcoding merely confirmed what was already suspected to be continuous intraspecific variation in morphology. BINs are available for all species recorded in Canada except E. ilicis and are provided in the taxonomic treatment for each species. Sequences for “barcoded” specimens are published in BOLD (http://www.barcodinglife.org) in the “Epeolus of North America” project, and will be made available on GenBank (http://www.ncbi.nlm.nih.gov/genbank/) following a revision of all Nearctic Epeolus species north of Mexico.

Anatomical and taxonomic terms used generally follow Michener (2007), except I use the terms frontal and vertexal areas instead of frons and vertex, respectively, following Prentice (1998) and Dumesh and Packer (2013), as these are not clearly delimited structural features. Puncture density is quantified as the interspace (i) relative to the puncture diameter (d). MOD is an acronym for median ocellar diameter, used as a comparative measure for indicating the dimensions of smaller features, especially hair length. F with a number corresponds to one of 10 (for female) or 11 (for male) flagellomeres of the antenna. T with a number corresponds to one of six (for female) or seven (for male) exposed metasomal terga. S with a number corresponds to one of six (for female) or eight (for male) metasomal sterna. I use the term ferruginous to distinguish black or nearly black integument from that which is any of the following colours: light brown, mahogany, reddish brown, red, and rusty orange. All measurements comparing lengths and widths are based on the longest and widest margins of an anatomical feature of a specimen at the highest magnification that would allow measurement in eyepiece micrometer units. I use the term length to describe any measurement along the longitudinal axis of a bee, and width to describe any measurement along the lateral axis, except in reference to the longitudinal extent of the transverse metasomal fasciae, for which I use the term breadth. Measurements of the scape were made excluding the radicle. Rightmyer (2008) proposed several terms specific to epeoline/nomadine bees, which I have adopted (with exceptions) and redefine here for clarity. Paramedian bands are the two longitudinal anterior lines of pale tomentum (pubescence composed of short, matted hairs) on the mesoscutum (extending posteriorly from the anterior margin of the mesoscutum but not attaining its apex) found in most Epeolus species. In E. canadensis and E. compactus I do not consider as paramedian bands the anteromedial patch of pale tomentum barely separated by the admedian line. The transverse bands of Rightmyer (2008) I refer to as the basal and apical metasomal fasciae. The fasciae of T1 may be connected laterally by a longitudinal band of varying width. Discal patch refers to the dark medial region of T1 covered in brown to black tomentum that may be sparser than the off-white or yellow tomentum forming the basal and apical (when present) fasciae.

Redescriptions are based on primary type specimens, although other specimens (usually non-type sequenced) were referenced for comparison and to fill in information gaps. The description of the sex opposite that of the primary type was based on the allotype or lectoallotype specimen (if available), paratypes, or non-type specimens. Specimens for study were provided by entomological institutions, museums, and university collections across Canada and the United States of America (USA), and are indicated with the following acronyms, with full names provided in parentheses: AMNH (American Museum of Natural History, New York, NY), ANSP (Academy of Natural Sciences of Drexel University, Philadelphia, PA), BBSL (Utah State University USDA Bee Biology and Systematics Laboratory, Logan, UT), BIML (Patuxent Wildlife Research Center USGS Native Bee Inventory and Monitoring Lab, Laurel, MD), CAS (California Academy of Sciences, San Francisco, CA), CNC (Canadian National Collection of Insects, Arachnids and Nematodes, Ottawa, ON), CTMI (Central Texas Melittological Institute, Austin, TX), CUIC (Cornell University Insect Collection, Ithaca, NY), DEBU (University of Guelph Insect Collection, Guelph, ON), EMEC (University of California Essig Museum of Entomology, Berkeley, CA), FMNH (Field Museum of Natural History, Chicago, IL), FSCA (Florida State Collection of Arthropods, Gainesville, FL), INHS (Illinois Natural History Survey, Champaign, IL), KUNHM (University of Kansas Biodiversity Institute and Natural History Museum, Lawrence, KS), NCSU (North Carolina State University Insect Museum, Raleigh, NC), PCYU (Packer Collection at York University, Toronto, ON), ROM (Royal Ontario Museum, Toronto, ON), RSKM (Royal Saskatchewan Museum, Regina, SK), UCR (University of California Entomology Research Museum, Riverside, CA), and USNM (U.S. National Entomological Collection, National Museum of Natural History, Washington, D.C.).

In lists of specimens examined, the records from different localities are always separated with a semi-colon. A comma between records denotes that the collection locality is the same but at least one of the following is different: date, collector, and entomological institution. With regard to specimen occurrence records, there were instances in which locality data were rather vague, particularly true of older records, and localities straddled county lines. In such cases, I omitted the county name and indicated the contents of the specimen labels. The same was true if I was unable to pinpoint an indicated locality on a Google map.

The key and redescriptions are accompanied by images taken with a Canon EOS 40D digital SLR camera using the Visionary Digital BK Plus imaging system, focus stacked in Helicon Focus, and edited in PaintShop Pro. In preparation for study and imaging, terminalia were excised, cleared in KOH for up to six hours, and ultimately stored in glycerine, later transferred to genitalia vials pinned under the associated specimens.

Range maps were constructed in RStudio (version 0.97.248) using the following packages installed in R (version 2.15.0): maptools (Bivand and Lewin-Koh 2014), raster (Hijmans 2014), rgdal (Bivand et al. 2014), and rgeos (Bivand and Rundel 2014). Maps of Canada and the USA were plotted using projected shapefiles obtained from Statistics Canada (2015) and the U.S. Census Bureau (2015). Points of occurrence for a particular species are based on GPS coordinates accurate to at least two decimal degrees. Using customized functions in R, continuous ranges were estimated by forming a splined convex hull polygon, a method also used for preparing distribution maps for the International Union for Conservation of Nature (IUCN 2012), of georeferenced occurrence records (from the literature and observed voucher specimens).

Taxonomy

Specimens of Epeolus species are similar to those of Triepeolus in general appearance, and males can be particularly difficult to distinguish. In Epeolus, the male pygidial plate is generally wider basally, with the lateral margins convergent toward the apex (e.g. as in Epeolus ainsliei Crawford [Figure 1a] and E. olympiellus Cockerell [Figure 1b]). In Triepeolus, the pygidial plate is generally comparatively narrow (e.g. as in Triepeolus pectoralis (Robertson) [Figure 1c] and T. lunatus (Say) [Figure 1d]), and its lateral margins are typically somewhat concave or sinuate. Female Epeolus have a very distinct sixth sternum, which is often partly visible in pinned specimens even without dissection (Figure 2a) as two convergent spatulate lateral processes bearing setae modified into pointed denticles; the processes are joined by a large lobe-like disc, which is usually not visible unless excised (Figure 2b). By contrast, S6 in female Triepeolus has a pair of narrow, elongate, forceps-like processes with coarse spine-like setae, separated by a disc reduced to a narrow transverse bar (Figure 2c, 2d). The apices of these processes and their long spine-like setae are often visible without dissection in pinned specimens. These morphological differences between females of Epeolus and Triepeolus are presumably related to host specialization (Rightmyer 2004) and the mechanism whereby the female oviposits into the cell wall of its host’s nest or between the caps separating brood cells (Roig-Alsina 1991). The spinose setae of Triepeolus seem to be for digging holes in the soil walls of host cells (Torchio 1986) and/or may have a tactile function (Rightmyer 2004). In Epeolus, tooth-like setae on the lateral processes and the rigid attachment of these processes to the disc of the sternum indicate a saw-like function necessary for breaking through the tough polyester lining that separates brood cells and coats the cell walls of its host nest (Torchio and Burdick 1988). In at least one species of Epeolus, this process is aided by a glandular secretion that dissolves the polyester lining of the host nest on contact, and later resolidifies to close the gap (Torchio and Burdick 1988). Females of the two genera may be further distinguished by the pseudopygidial area – the medioapical region of T5 that generally changes slope (and may be elevated) from the rest of the tergum, and whose disc is flat or somewhat depressed and usually covered in shiny short hairs that are often uniform in length (Michener 2007). In Epeolus, the shape of this area is either campanulate (Figure 3a) or lunate (Figure 3b, 3c, 3d, 3e), whereas in Triepeolus it is more variable, and may be ovate or round (Figure 2c), quadrate, triangular, a shape intermediate between triangular and quadrate, or a shape more complex in outline. With one notable exception, the pseudopygidial area of Triepeolus is always relatively longer than in Epeolus (Rightmyer 2008); in the unusual Mesoamerican T. epeolurus Rightmyer, the transverse band of metallic setae on the pseudopygidial area (Figure 3f) is remarkably similar to that of some species of Epeolus, but is concave rather than arched in dorsal view. Another unusual feature of T. epeolurus is that the pseudopygidial setae reflect silver, whereas in most Triepeolus they reflect a golden colour (Rightmyer 2004).

Epeolus is represented by 102 valid species worldwide (Integrated Taxonomic Information System on-line database, http://www.itis.gov.) [Retrieved 11.ii.2016]. Based on my own knowledge in combination with records available on Discover Life (Ascher and Pickering 2016), 45 species were until the date of this publication recognized as occurring in North America excluding Mexico and the West Indies. The first species described as being a North American Epeolus, E. mercatus Fabricius, cannot be confidently assigned to Epeolus or Triepeolus, as the original description is vague and the type material apparently has been lost (Rightmyer 2008). Therefore, the numbers above do not include Epeolus mercatus Fabricius. Nonetheless, it would be surprising if this species did not represent another described species in one of these two genera. Brumley (1965) described an additional seven species (all from the American Southwest), but as he did not publish his work his names cannot be formally recognized. Apparently, one of these species, occurring in Arizona and Texas, USA, had already been described by Smith (1879) from Oaxaca, Mexico (Rightmyer 2008). None of the seven “new” species, however, are known to range into Canada. Despite the diversity of Epeolus in North America, with more known species than any other continent, the genus is poorly understood.

Several North American species of Epeolus were originally described as belonging to Phileremus Latreille and Triepeolus. Phileremus (Ammobates Latreille subgenus Ammobates Latreille s. str. in Michener 2007) included cleptoparasitic bee species in which the fore wing has two rather than three submarginal cells. This character is variable even within species (and sometimes specimens) of Epeolus, and Phileremus contained species from a large number of genera (mostly Nomadinae), including Ammobates, Ammobatoides Radoszkowski, Biastes Panzer, Epeolus, Dioxys Lepeletier and Serville, Holcopasites Ashmead, Melanempis Saussure, Neolarra Ashmead, Neopasites Ashmead, and Pasites Jurine (Ascher and Pickering 2016).

I synonymize 14 previously proposed names under those of four valid species. Epeolus americanus and E. minimus are similar to some species that are not treated here because they occur south of Canada only. They include a cryptic species revealed by DNA barcoding (BOLD:ACZ2142) within the “americanus group”, whose subtle morphological differences and collection date and locality record within Los Angeles County, California are shared with the holotype of E. asperatus Cockerell, which I have seen and examined. Also similar is the holotype of E. melectimimus Cockerell and Sandhouse. Epeolus barberiellus Cockerell is another species similar to E. americanus, with unique physical attributes and known to occur only in New Mexico and Texas. A species very similar to E. minimus is E. banksi (Cockerell), with unique physical attributes, and apparently restricted to parts of the mid-Atlantic and southeastern States. DNA barcode data are not yet available, but morphology suggests that specimens identified as E. banksi are clearly distinct from E. minimus. The names E. americanus and E. minimus antedate those of the abovementioned similar or cryptic species, and for the reasons stated herein I am confident that the new synonymies proposed are correct for the taxa in question, and do not apply to any other species.

Biology

All Epeolus species for which host use has been assessed are cleptoparasites of Colletes Latreille, the type genus of the family Colletidae (Michener 2007). The reproductive biology and immature stages of Epeolus were first described for E. pusillus Cresson in association with Colletes ciliatoides Stephen (Torchio 1965) and C. compactus compactus Cresson (Rozen and Favreau 1968). Both host species construct a single cell at the end of a lateral tunnel that branches from the meandering, mostly vertical main tunnel. Rozen and Favreau (1968) noted female E. pusillus flying swiftly 15–20 cm above the ground, slowing down over what presumably to them seemed to be nest entrances – one female flew quickly toward a burrow from which a host Colletes had previously been collected, descended, and re-emerged within a minute. When Rozen and Favreau (1968) excavated the brood cell, they found that it had an Epeolus egg attached, positioned between the inner and outer envelopes of the cell lining.

There is some indication that female Epeolus repeatedly visit and inspect the nest or nests of their host species of Colletes, likely to confirm the suitability of the nest site and ensure that they are present at the right time for oviposition. For instance, Graenicher (1906) reported that upon discovering a C. eulophi Robertson nest (about midday), a female E. minimus (Robertson) began crawling over the ground with quivering wings. The female approached the nest from various angles without entering. The female Epeolus then perched motionless on a small plant, or twig at times, about 20 cm above the nest entrance while the female Colletes returned with provisions. The female Epeolus preened herself at that time, and again after the host female left before the Epeolus herself entered the nest for about one minute. The female then emerged and examined the surrounding area. The process of examining the nest entrance, perching, entering the nest, and examining the surrounding area was repeated within a particular day and on different days (confirmed by marking of the female Epeolus specimen). To be successful, the female Epeolus must avoid detection by the host. In Central Europe, Bogusch (2003) twice observed a female C. similis Schenck successfully defending a nest from a female E. variegatus (L.).

Like other Nomadinae, Epeolus females enter unsealed cells while the host is foraging during the nest provisioning stage. Whereas Colletes eggs were found to be attached to the inner polyester lining of the cell, the egg of E. pusillus was laid between the inner and outer polyester linings of the double-layered nest lining of its host (Rozen and Favreau 1968). Where the egg is laid depends on the host and type of nest constructed. Torchio and Burdick (1988) documented two strategies used by E. compactus Cresson. Its host species, C. kincaidii Cockerell, may reuse abandoned nests. In this case, E. compactus inserts its eggs between the inner lining of the burrow and residual lining (assuming it is intact) from previous nest use, because there is sufficient space and presumably also to protect the egg from getting wet. Interestingly, rates of cleptoparasitism were higher for reused nests. Torchio and Burdick (1988) found overall rates of nest parasitism of C. kincaidii by E. compactus to be as high as nearly 18%. If the nest was newly founded by the female Colletes host (and only a single polyester layer separates the cell from bare ground), E. compactus instead attaches its eggs to the caps of completed cells separating the brood cells (Torchio and Burdick 1988). Although the egg is exposed within the already completed cell, the larva hatches into the cell that was incomplete when the parent Epeolus oviposited. In some instances, multiple eggs may be deposited through a cell cap, but it is not known if these belong to the same or multiple female Epeolus. Oviposition through the cellophane-like cell lining of another colletid genus (Scrapter Lepeletier and Serville) has similarly been documented in the nomadine cleptoparasitic genus Sphecodopsis Bischoff (Rozen and Michener 1968 – as Pseudodichroa). Rozen (1968) suggested that Sphecodopsis females puncture the lining and poke a hole in the sand outside the cell (where the egg is to be embedded) using the heavily sclerotized, median process of S6.

Rozen and Favreau (1968) observed that when the larva of E. pusillus hatched, it immediately found and killed the host egg. Similarly, Torchio and Burdick (1988) found that the larva of E. compactus killed the host egg or larva using its long, sickle-shaped mandibles, and combated the other Epeolus larvae in superparasitized host cells until a single survivor remained. In the case of E. pusillus, the rate of larval development was found to be much faster than that of the host (C. compactus compactus in this case), and by the time the cleptoparasite larva went into diapause, neighbouring representatives of its host species had consumed less than half of their provisions (Rozen and Favreau 1968).