A new species Leioproctus zephyr (Hymenoptera: Colletidae) is described from both sexes. Leioproctus zephyr sp. nov. is remarkable in featuring a large longitudinal ridge on the clypeus. This diagnostic morphological feature present in both sexes, along with various other distinctive characters including the male genitalia, female hind-tibial spur, and glossa morphology, clearly distinguish this species from all other Leioproctus. Along with these unique traits, L. zephyr cannot be classified into any of the existing subgenera of Leioproctus, sharing some, but not all, of the characters of the subgenera Ceratocolletes, Charicolletes, Protomorpha and Odontocolletes. DNA barcoding with the CO1 gene confirmed the sexes belonged to the same species and it did not match any previously barcoded species. This species is restricted to native vegetation remnants in the southwest Western Australian biodiversity hotspot, and is highly specialised, foraging only on a few species in the genus Jacksonia (Fabaceae). The unusual clypeus may be an adaptation for foraging on the keeled papilionaceous flowers. The limited number of sites this species has been collected from and its oligolectic diet suggest L. zephyr should be considered to be a species of conservation concern. Further taxonomic research is required to determine the phylogenetic position of this unusual Leioproctus.

Australia, biodiversity hotspot, colletid, DNA barcoding, new species, specialist

The genus Leioproctus Smith 1853 (Colletidae Lepeletier 1841), as currently described, is a highly diverse, speciose taxon (Almeida and Daforth 2009; Engel and Gonzalez 2022). In Australia, it is distributed throughout most of the continent, and the most recent guide to native bees of Australia divides this genus into 24 subgenera, with 193 named species (Houston, 2018), yet with hundreds awaiting description (Houston, personal comm.). New species are continually being described (e.g., Batley and Popic 2013; Leijs et al. 2018).

The systematics of Leioproctus requires clarification (Packer 2006; Almeida et al. 2019). A phylogeny combining morphological and molecular data of currently recognised subgenera is yet forthcoming. Even Leioproctus s. str. appears to be paraphyletic based on molecular phylogenies (Almeida and Danforth 2009; Almeida et al. 2019). In the magnum opus on the classification of native bees, Michener (2007) drew attention to how the sub-genera of Australian Leioproctus can be ill-defined, with partial intergradation among some taxa, as evident in the last revision of seven subgenera of Leioproctus (Maynard, 2013), A comprehensive and more thorough understanding of the Australian subgenera is hampered by the sheer diversity of many unusual species that are undescribed, and even of those that are described, many are described from only one sex. It is clear that an updated classification for the Australian Neopasiphaeinae is needed (Almeida et al. 2019; Engel and Gonzalez 2022). A new species, with a distinctive clypeus, that does not fit neatly into an existing subgenus is described, including its CO1 barcode, and data on its restrictive foraging and distribution range. This description will contribute to documenting and describing the diversity of Australian Neopasiphaeinae.

Specimens involved in the description were collected by the author with an entomological sweep-net (bag mesh size 0.9 × 0.3mm, Australian Entomological Supplies Pty Ltd) during surveys to sample native bee assemblages in residential gardens and bushland remnants within the urbanised region of the southwest Western Australian biodiversity hotspot (Prendergast et al. 2022; Suppl. material 1). The unusual appearance of this species and an inability to key the species out to subgeneric or species level from published keys led the author to contact Dr Terry Houston of the WA Museum to inquire whether he had seen this species before. This species was confirmed to be undescribed, lodged in the WA Museum and catalogued as Leioproctus (Protomorpha?) F188/M173. Further consultation with Dr Glynn Maynard who undertook the most recent revision of Australian Leioproctus (in part, Maynard 2013) confirmed that these specimens did not match described species.

Standard melittological terminology is used to describe the morphology (Michener, 2007). The following standard acronyms are used (following Michener (2007), Houston (1990) and Leijs et al. (2018)): HL head length; HW head width; AOD antennocular distance; IAD interantennal distance; OOD ocellocular distance; OAD ocelloantennal distance; metasomal sterna and terga are denoted S[segment number] and T[segment number], and flagellomeres are denoted F.

Following Packer (2006), the relative diameter and spacing for punctures (sculpture) are denoted by d and i, respectively. Other types of surface sculpturing follow Houston (Houston, 1975), as used in (Leijs et al. 2018). Measurements of key morphological features and relative head measurements were made on five specimens of each sex and averaged, and given in millimetres (Suppl. material 2). Specimens were observed with a Leica M205 C stereomicroscope, and measurements were made on high-resolution images taken with the same stereomicroscope and using the Leica auto-montage image stacking software. Images of key features were taken using a Nikon camera with Passport and Helicon image stacking software.

A sample (hind femur) of the female and male type and allotype were submitted to BOLD (Barcode of Life Database) for DNA barcoding using the cytochrome c oxidase subunit 1 (CO1) gene. The DNA barcode sequence, and other specimen information associated, can be accessed in BOLD via: as part of the Australasian and Pacific bee fauna Project (MSAPB): http://www.boldsystems.org/index.php/MAS_Management_DataConsole?codes=MSAPB.

The sequences were obtained from Canadian Centre for DNA Barcoding (CCDB) at the University of Guelph, Guelph, Ontario, Canada. Standard DNA sequencing protocols were carried out by CCDB (available online at: http://www.ccdb.ca/resources.php), using the PCR primers LepF1/LepR1. The barcoded vouchers are housed at the Museum of Western Australia. BOLD delineates molecular operational taxonomic units (MOTUs), which typically are in close concordance with species delineations based on traditional methods (Schmidt et al. 2015). The barcode index number (BIN) (Ratnasingham and Hebert 2013) is automatically assigned to a MOTU, which is incorporated into BOLD.

To ascertain the position of this species in relation to other Leioproctus and infer its placement within one of the described subgenera, a Taxon ID tree was created in BOLD using all specimens in the AUSBS project. The Taxon ID tree procedure uses varied distance metrics to generate a neighbour-joining (NJ) tree based on nucleotide similarity in the barcoded COI gene. Sequence alignment is automatically handled, with the Kimura 2 Parameter as the default distance model.

This new Leioproctus species is highly distinctive in its morphology. It does not conform to any of the subgenera in the latest revision (Maynard, 2014). This morphological distinctiveness of this species was supported from DNA barcoding studies. This species was in a cluster with four other species (none of which appear to have been formally described), with an average distance of about 15%. The closest species from an NJ tree based on sequenced species is an undescribed Leioproctus (Leioproctus sp. “CH13”​). It appears that the ridge is an autapomorphy, as none of the species in this clade have a ridge on the clypeus or a blunt thick apex of the hind tibial spur. Dissections of genitalia of the males revealed the S7 is comparatively simple for L. zephyr, being more complex in these other species. The only distinctive trait of L. zephyr shared with these other species is the short, robust S8 (R. Leijs, personal communication, 2020). Although sequencing with a single gene is insufficient to accurately represent evolutionary relationships, on the basis of these results it appears that clypeal protuberances can be homoplastic and represent convergent evolution in L. zephyr and Leioproctus (Ceratocolletes).

In the WA Museum collection database, Houston tentatively placed this undescribed species in the subgenus L. (Protomorpha), however although this species exhibits some features characteristic of this subgenus (namely terga with pale apical hair bands; flagellum short, middle segments mostly broader than long or scarcely longer than broad; clypeus and supraclypeal area not flat, usually punctate, suture separating them distinct), other key features of Protomorpha, are lacking, including: females with striate pygidial plate (pygidial plate lacks any ornamentation or sculpturing); males with hind tibia and basitarsus elaborately expanded (no elaborations on these leg segments, tibia only slightly broader than is typical for male Leioproctus, no expansion of the basitarsus); males with robust body like that of females (although robust compared with some Leioproctus subgenera, female is distinctly more robust than the male); S7 of male with two large apical lobes (lobes, although present, are greatly reduced); mandibles simple, sharply pointed, without preapical tooth (mandibles broad, blunt, with preapical tooth); propodeum shorter than metanotum (propodeum is longer than metanotum). Similarly, this species exhibits features of Odontocolletes (which has features that are consistent with most of the major external features of Protomorpha), including the malar space absent, strong punctures on the dorsal surface of the mesosoma with smooth interspaces; it is also from the same geographic region as the majority of L. (Odontocolletes) species (Michener 2007; Maynard 2013). However, it lacks other key diagnostic traits, including red terga lacking apical hair bands and the diagnostic feature of Odontocolletes of a large, blunt, median tubercle on the metanotum (Michener 2007; Maynard 2014). And whilst like these subgenera, S7 of the male has two apical lobes, these wouldn’t be considered “large”, as is the case for Protomorpha and Odontocolletes. It also features some similarities with Charicolletes as described by Maynard (2013), including impressed facial fovea, strong punctures, and short antennal scapes. However, it is not metallic like Charicolletes, nor does it have a median metanotum tubercle. The morphologies of S7 and S8 of L. zephyr do not match those of the taxa illustrated in Maynard (2013). Whilst a number of taxa have two broad, flat apical lobes oriented laterally, few are as short in relation to the ventral processes, nor are they the same shape, as that of L. zephyr. Interestingly, S7 of L. zephyr is most similar to that of Goniocolletes parvus Maynard, 2013, however S8 and the genital capsule are morphologically dissimilar (compare with fig. 217–219 in Maynard (2013)). The genital morphology also bears some similarity to L. (Exleycolletes) argentifrons Smith, 1979 and L. (Leioproctus) macmillani Houston, 1991, but the ventral lobes are not as long in relation to the apical lobes, and distribution of hairs on the apical lobes are dissimilar, and the apical portion of S8 is narrower (see Maynard 2013, figs 48–49, 121–122). This species shares features with the subgenus Ceratocolletes, and appears to be most closely related to this genus, in being a stout-bodied, strongly punctate colletid; surface sculpture, on the metasoma in particular, having small, strong punctures with clearly defined, polished interspaces; the second to fourth metasomal terga in females and second to fifth metasomal terga in males with white, apical bands; malar space absent, and, notably, the clypeus with narrow, longitudinal, median, glabrous area (obscured by hair in males) (Maynard, 1993). However, unlike Ceratocolletes there is no distinct horizontal basal area on the propodeum and the basal area is not rounded onto the vertical area, and the propodeum in punctured, rather than smooth; in addition the scape does not attain the level of the median ocellus (Maynard 1993; Michener 2007). In the key to subgenera of Leioproctus with three submarginal cells by Maynard (2014), Ceratocolletes is separated along with Lamprocolletes and L. opaculus Cockerell 1929 from all other subgenera in having the jugal lobe of the hindwing not reaching cu-a. In L. zephyr, the tip of the jugal lobe extends just past the cu-a vein. The inner hind tibial spur of the female is also distinctly blunt and, unlike Ceratocolletes, does not have 11 long, fine teeth (Maynard 1993), instead having three blunt prongs only on the proximal section of the spur (see Fig. 7). The female’s pygidial plate is also not narrow and convex (Maynard 2014). The current species lacks the diagnostic paired lateral lobes on the male seventh sternum, and instead has only a single lateral lobe, and other features of the genitalia exhibit degrees of difference from Ceratocolletes. Only two species are currently recognised as belonging to Ceratocolletes. L. (Ceratocolletes) xanthosus Maynard 1993 has been collected from two areas in eastern Australia, and differs from the current species in the above differences related to the subgenus, as well as yellow colouration, golden hair in the males, and although the clypeus of the female is convex with a median ridge, it is not strongly protuberant (Maynard, 1993). The other species, Leioproctus (Ceratocolletes) antennatus also occurs only in southwest WA like L. zephyr and the apical hair bands on the terga in the female are incomplete medially (although these are also incomplete in the male of L. zephyr). However, in addition to the above differences at the subgeneric level, and like L. (Ceratocolletes) xanthosus, the antennae have yellow colouration, and males of L. (Ceratocolletes) antennatus have antennae with the apical segment expanded and flattened, in contrast to the unmodified antennae of L. zephyr. This unusual modification of the male’s antennae however is an autapomorphy and is absent in L. (Ceratocolletes) xanthosus. In Michener (2007)’s description of this subgenus, he notes that hind legs of the male are incrassate, trochanters toothed, tibiae bent, and tibial spurs reduced in size-features absent in L. zephyr, however it is noted that these features of the male’s hindlegs are only in one of the species, but which of the two species is not mentioned, and the description of Ceratocolletes by Maynard (1993) does not include these features. Although the first description of Ceratocolletes included a medially protuberant clypeus as a diagnostic feature of this subgenus (Michener, 1965), Maynard (1993), in placing L. xanthosus into Ceratocolletes, suggested that this feature was no longer subgenerically significant.

Like L. zephyr, both Ceratocolletes have only been collected on Fabaceae: Pultanaea spp. for L. (Ceratocolletes) xanthosus, and as with L. zephyr, L. (Ceratocolletes) antennatus have been recorded exclusively foraging on Jacksonia (Houston, 2000).

Looking at its phylogenetic relationships and cladistics groupings based on the Taxon ID Tree functionality in BOLD, the dendrogram generated from sequencing using the neighbour joining algorithm was not able to resolve its subgeneric grouping. Rather, it suggests that L. zephyr belongs to a distinct clade with a number of other undescribed Leioproctus (Mark Stevens, Remko Leijs, pers. comm. March 2022). The closest scientifically-described species were Leioproctus conospermi Houston 1989 (Supporting Information3), – an oligolectic species that features highly modified features as adaptations for foraging on the host, Conospermum (Houston, 1989), and Leioproctus excubitor Houston 1991, which has highly modified antennae in the male (Houston, 2018). Both of these species are currently placed in the subgenus Leioproctus. As such, the various features outlined above that L. zerphyr shares with various other subgenera (Protomorpha and Cladocerapis) are not taxonomically informative. It should be noted that phylogenetic analyses involving more than just the CO1 gene are required to further elucidate the taxonomic placement of L. zephyr. In particular, the dendrogram using just the CO1 gene is a phenetic result, used for illustrative purposes, and is merely suggestive but is not a reliable phylogenetic estimate; a rigorous phylogeny using more genes and sophisticated phylogenetic analyses is recommended (Ramírez et al. 2010; Trunz et al. 2016; Packer and Ruz 2017).

The remarkable feature about this new species is its highly distinct clypeus, featuring the medial ridge and protuberance, which is unusual for Leioproctus (Maynard, 2014). Only the monotypic subgenus Colletopsis Michener 1965, and the two species in L. (Ceratocolletes) feature a median ridge on the clypeus, which is especially pronounced in L. zephyr. This feature invites speculation about its evolution and function. Present in both sexes, it is unlikely to be due to sexual selection (although there is the possibility of mutual sexual selection acting on this feature). The male is relatively robust and broad metasoma compared with some Leioproctus males, but still relatively slimmer than that of the female. Patterns of sexual dimorphism vary across bees, and the relative size of the sexes can be considered to be informative about the relative sexual and natural selective forces acting up on the species. The relatively larger size of the female suggests that L. zephyr is non-territorial (Alcock and Houston 1996; Paxton 2005). The protuberance, although present in both species, is slightly larger relative to the bee’s head in the male (length of protuberance extending from clypeus in profile relative to head length 0.22 for the female compared with 0.29 for the male; see Supporting Information2), which may suggest a role for this protuberance in sexual selection. No instances of mating or nesting behaviour were observed but would be insightful for future studies.

The raised ridge and protuberance may be a point for muscle attachment of the mandibles (Grimaldi et al. 2005). X-ray micro-computed tomography (micro-CT) scans would shed light on whether this hypothesis has support. Another intriguing possibility is that this feature serves as a wedge to open up the keel of Jacksonia flowers. As a papilionaceous flower, the flowers of J. sericea have their fertile organs, and thus pollen and nectar “hidden” by a keel comprised of two ventral petals (Córdoba and Cocucci 2011). In order to access the floral rewards, pollinators must push down on the keel and the lateral petals of papillionate flowers (Córdoba and Cocucci 2011). This requires some force (Córdoba and Cocucci 2011), and it may be that the clypeus with its prominent protuberance of this specialised Leioproctus is used to wedge open the keel of its host flower. Although involving glossa or leg modifications rather than clypeus structure, other cases of unusual bee morphological structures have been linked to adaptations for accessing floral rewards in flowers that have limited access to flower visitors (e.g. Houston 1983; Pauw et al. 2017). Observations of other bee species have revealed behaviours that involve using the head and mouthparts to push or force themselves into flowers that have petal morphologies limiting access (Packer, 2004), including those with keeled flower parts (e.g. Westerkamp 1993; Raju and Rao 2006; Amaral-Neto et al. 2015). Although foraging observations were made in the field, the speed at which the bees foraged on flowers precluded being able to discern whether they performed this behaviour; specialised video-cameras recording this specie’s foraging behaviour and analysed in slow-motion play-back would be able to evaluate support for this hypothesised function.

Leioproctus zephyr has an extremely limited range of flowers it will forage on, namely a subset of species within the genus Jacksonia (Suppl. material 1). This contrasts with most Leioproctus species, which are often highly polylectic (Maynard, 2014). A notable exception in this region of SWWA are three species that are specialised on Conospermum (Proteaceae) (Houston, 1989). In this region where collections were made, most Leioproctus I collected or have observed have been recorded foraging on a range of Myrtaceae (Corymbia, Callistemon, Eucalyptus). The only other occurrence of a Leioproctus on a plant in the family Fabaceae was on Acacia (previously classified as a distinct family, Mimosaceae) (Maynard, 2014). Despite other Jacksonia (J. sternbergiana Bentham and J. furcellata Bonplande & de Candolle co-flowering, often in abundance, at sites where L. zephyr was collected, no specimens were ever observed foraging on these other related species.

The reason for this specialisation can only be speculated. It is unlikely to be due to avoiding competition, as J. sericea is frequently visited by Megachile Latreille 1802, a genus which is more typically associated with Fabaceae (Houston 2000; Prendergast and Ollerton 2021).

Leioproctus zephyr also appears to have a limited season of activity covering only two months in summer (December to January). The species was not observed after early January in the more recent collections by the author. Although the latest date the species has been collected was the end of January (January 19^th), this was a single collection forty years ago. As temperatures have risen by almost 1 °C over the last century, and rainfall has declines of 15% since the mid-70s, and it may be that climate change (Climate Council 2014) means that temperatures are now too hot for this species to remain active at this later collection date (Prendergast, 2022). Climate change may post a threat to L. zephyr by causing mismatches between the phenology of the emergence of the bee and its host plant (Hughes 2003; Pyke et al. 2016; Schleuning et al. 2016; Settele et al. 2016; Prendergast 2022).

This new species appears to be restricted to native vegetation reserves in the southwest Western Australian biodiversity hotspot (refer to Fig. 3). The only other population the species has been collected from is Pinjarrega Nature Reserve, almost 200 km² away from the other sites and thus well outside the flight range of the species (Zurbuchen et al. 2010). Whether other populations exist in the intervening region is unknown, however as J. sericea occurs in the intervening area, targeted surveys are recommended. It is also unknown whether the species still persists at this location, as it has been over two decades since the three specimens have been collected from this location. If it were to become extirpated, re-colonisation is therefore unlikely.

Comprehensive surveys that I conducted over 10 months failed to record this species in any residential gardens, which can be attributed to the lack of suitable foraging resources. Even at bushland remnants were Jacksonia sericea was flowering, this did not guarantee the presence of this species: for example, Piney Lakes Reserve has J. sericea patches, and is approximately only 4 km away from Wireless Hill where this species was recorded, yet no records were made at Piney Lakes. L. zephyrus was also not recorded on J. sericea at other bushland remnants surveyed in the City of Bayswater in 2020–22 (Prendergast 2021, 2022b). This region has become highly fragmented due to urbanisation and the associated loss of natural habitat for road and urban development. Other studies outside of Australia have also found that specialist species are underrepresented in small, isolated fragments in urbanised areas (Cane et al. 2006; Pauw 2007). Given the dependence of this new species on native vegetation remnants with J. sericea, efforts much be made to protect any native bushland that remains with these plants, and encourage bushland restoration initiative to plants patches of J. sericea to promote connectivity and increase the overall habitat area for this remarkable species.

There has been no formal conservation status of Leioproctus zephyr, but some recommendations can be made based on information regarding its distribution, phenology, habitat, and resource associations. 18 specimens have been collected by T. Houston along with 96 by the author (Suppl. material 1). The majority however (two thirds of all specimens) have been collected from a single site, Kings Park (Suppl. material 1). Moreover, it is unknown whether the species still exists at two of the sites surveyed by T. Houston, especially the site where the species was first collected (Neerabup National Park) in 1979, which is a great distance from where the other populations occur. L. zephyr is locally abundant at Kings Park, which is a large, intact area of remnant bushland that is under strong conservation legislation as an A-class Reserve, and is well-managed by the Government of Western Australia’s Kings Park and Botanic Gardens Authority, with two-thirds of this 400.6ha park being protected as managed bushland (Botanic Gardens and Parks Authority 2017). One of the sites where L. zephyr was recently collected is marked to be undergoing partial destruction to make way for urban development (Young, 2018); ongoing urbanisation of the matrix surrounding sites may also affect populations through edge effects, by increasing isolation, and preventing metapopulation dynamics.

Further surveys during December and January in areas where Jacksonia sericea is flowering are required to establish this species extent of occurrence. Ongoing monitoring is also required to detect any population trends. Preservation of J. sericea is of utmost importance for this species.

Native bees are suffering from a major taxonomic crisis, and without a scientific name, understanding their distribution, abundance, and conservation status is a challenge, which is contributing to the poor state of conservation of invertebrates, including in a megadiverse country like Australia (Braby, 2018). Describing and naming this Leioproctus will enable it to receive conservation attention, as well as serve as a springboard for further taxonomic work on the diverse Leioproctus in Australia. This species is moreover morphologically distinct, featuring a modified clypeus, is oligolectic, and appears restricted to a few locations in the southwest Western Australian biodiversity hotspot. DNA barcoding has reinforced its distinct position and offers inspiration for further research into the taxonomy and systematics of Australian native bees and Hymenoptera at large.

I would like to thank Dr Glynn Maynard (Department of Agriculture, Fisheries, Forestry Office of the Chief Plant Protection Officer) for checking her collection to ensure this was in fact an undescribed species, Dr Terry Houston (Museum of Western Australia) for identifying the corresponding specimens in the WAM collection, Dr Nik Tatarnic (Museum of Western Australia) for providing me information on the previous specimen collections, and Prof Laurence Packer (York University), Dr Remko Leijs (South Australia Museum) and Prof Jason Gibbs (University of Manitoba) for their time and expertise in providing useful feedback on improving the manuscript. Thank you also to Phil Patterson and Dr Nik Tatarnic for databasing these specimens so they are safe and secure in WAM. An extra big thank you for Dr Nik Tatarnic for going through thousands of specimens to find my types and take some extra photos for this manuscript whilst I was on the other side of the country. I am also grateful to the editor Dr Jack Neff and the anonymous reviewers for their constructive comments. Finally, I’d like to thank my Maremma dog Zephyr for always being there for me to give me a big doggy grin and unconditional love.

Funding for the DNA barcoding was made possible through awards I received from Flow Hive and Graduate Women of WA. The surveys in which I collected specimens were part of my PhD project, funded by a Forrest Research Foundation scholarship. Permission to survey parks and reserves was granted by Kings Park and Botanic Gardens Authority, local councils, and DBCA Fauna Collecting Licence.

No ethics approval was required.