Agapostemonfasciatus Crawford (Hymenoptera, Halictidae), a valid North American bee species ranging into southern Canada

Abstract Background Sweat bees of the genus Agapostemon Guérin-Méneville, 1844 (Hymenoptera: Halictidae) are common and widespread in the Americas. Despite distinct morphological characters that were recognised in earlier taxonomic treatments, Agapostemonfasciatus Crawford, 1901 has been considered a variety of A.melliventris Cresson, 1874 since the 1930s and later placed into synonymy under A.melliventris in the early 1970s. New information A more detailed study of morphology (including examination of type materials), distribution and genetic data (i.e. DNA barcodes) of these two taxa suggests they are not conspecific. As such, A.fasciatus is resurrected as a valid North American bee species. Agapostemonfasciatus ranges further north in North America than A.mellivenrtis, reaching the southern Prairies Ecozone of Canada (Alberta, Saskatchewan), while most records of A.melliventris are from the south-western United States and northern Mexico. More accurate distributions for both species can be modelled as specimens in collections are identified using the diagnostic features provided. However, additional work is required on the A.melliventris species complex in the southern United States as genetic data suggest that multiple taxa could be present.


Introduction
Bees of the genus Agapostemon Guérin-Méneville, 1844 (Hymenoptera: Halictidae) are commonly encountered throughout southern Canada and all of the United States and Mexico, the Antilles and into South America (Roberts 1972, Roberts and Brooks 1987, Janjic and Packer 2003. Roberts (1972) recognised 43 species from the Americas (though see Sheffield et al. (2021)), seven of these known from Canada (Sheffield et al. 2017). Sheffield et al. (2014) reported A. melliventris Cresson, 1874 from Saskatchewan, which was also a new record for Canada. Agapostemon melliventris, a species of apparent Sonoran origin (Porter 1983), is considered rather unique amongst the genus by usually having a metasoma that is at least partially amber coloured (Fig. 1) and being one of only four species in which females have the clypeus pale maculated apically (Fig. 2a, b, c), while all other species have a clypeus that is dark apically (Fig. 2d) (see the keys of Roberts (1972)). However, the colour of the metasoma of females of A. melliventris s. l. varies from mostly amber to black (Cockerell 1924, Roberts 1972) (and see Fig. 1) resulting in the synonymy of two taxa under A. melliventris: A. digueti Cockerell, 1924(synonymy by Sandhouse (1936) and A. fasciatus Crawford, 1901(synonymy by Roberts (1972). Variation in metasomal colour in female Agapostemon melliventris Cresson, 1874 s. l. based on the species concept of Roberts (1972).
Females of typical A. melliventris have at least the first three metasomal segments almost completely amber in colour (Cresson 1874) (Fig. 3a), while the females, originally described as A. digueti, usually have only the first two segments amber in colour (Fig. 3b), though variation was reported even within the type material (Cockerell 1924). In addition to partially sharing the amber metasomal colouration, the females of A. melliventris and A. digueti share pale maculations on the scape (Fig. 2a, b), one of the character states a b c d Figure 2.
assigned to this taxon (i.e. A. melliventris s. l.) in the morphology-based phylogeny of Janjic and Packer (2003). Females from the type series of A. fasciatus (Fig. 4), though sharing the clypeal colour pattern with these two taxa (Fig. 4c), differ in not having the scape maculated (Fig. 2c) and with little to no amber colouration on the metasoma (Fig. 4a, b).
The male of A. melliventris s. l., though known to Cockerell (1897), Titus (1900) and Hooker (1908), was not described until done so by Sandhouse (1936) and is also considered unique from most other North American Agapostemon in having the basal area of tergum 1 pale amber to yellow (Fig. 5a), not black ( Fig. 5b) (and see Roberts (1972)). Males in the type series of A. fasciatus also show other distinct morphological differences from males of A. melliventris and A. digueti, including the colour of the trochanters, being metallic green in A. fasciatus (Fig. 6a), but entirely yellow or with slight greenish tinges on the hind leg in the other two taxa (Fig. 6b) (Sandhouse 1936).
Despite noting these morphological differences in both sexes of these taxa and commenting on differences in distribution, Sandhouse (1936) failed to recognise A. fasciatus (as A. plurifasciatus (Vachal, 1903)) as a distinct species, instead recognising it as a variety of A. melliventris. Subsequently, only the variation of metasomal colour of females was used by Roberts (1972) to justify his synonymy of A. fasciatus under A. melliventris. Roberts (1972) indicated a sympatric distribution of A. fasciatus and A. melliventris, though he likely did not distinguish material thoroughly, based on the distinct a b Metasomal colour pattern in female Agapostemon melliventris Cresson, 1874.
The purpose here is to clarify the taxonomic status of these taxa for an upcoming review of Canadian bee species by re-examining morphology and analysing COI sequences and distribution of previously recognised units of A. melliventris s. l. in North America. In addition to the materials mentioned above, records were also mined from the literature (e.g. Porter (1983)) and GBIF.org (2023)), though identifications were likely based on Roberts (1972) concept of the species.

Specimens of
Additional DNA barcode sequences of material from North American were obtained following previously reported methods (Sheffield et al. 2017), supplementing material already in the Barcodes of Life Data System (BOLD: Ratnasingham and Hebert (2007)) as part of an ongoing campaign to build a comprehensive DNA barcode library for the bee fauna of Canada (Sheffield et al. 2017) and the World (Packer et al. 2010). All data were mapped using SimpleMappr (Shorthouse 2010) with taxa assigned according to morphology as per Crawford (1901) and Sandhouse (1936), as described above. Only specimen data for examined specimens of A. fasciatus are provided for this single taxon treatment; specimens from the AMNH were identifed by Corey Smith using photos and descriptions provided by the author.
Males of A. fasciatus can be separated from all other North American Agapostemon, except A. melliventris in having the basal area of tergum 1 coloured pale amber to yellow (Fig. 5a), not black (Fig. 5b). Males of A. fasciatus differ from A. melliventris by the colour of the trochanters, being metallic green in A. fasciatus (Fig. 6a), but entirely non-metallic yellow in A. melliventris (Fig. 6b). Differences in genitalia are not apparent.
In addition to these morphological differences in both sexes, the distribution of A. fasciatus is typically more northern across the Great Plains, extending into southern Canada (Alberta, Saskatchewan), with only a few records occurring in the southwestern United States (Fig. 7).

Analysis
In addition to the distinctive morphological characters of both sexes of A. fasciatus described above, which are consistent across all material examined, it also differs from A. melliventris genetically, based on results from the Barcode Gap Analysis tool in BOLD. Agapostemon fasciatus is presently assigned to Barcode Index Number (BIN, after Ratnasingham and Hebert (2013)) BOLD:ACU3900 and differs by 7.99% sequence divergence from other "A. melliventris" in BOLD (currently represented by at least three BINs BOLD:AAJ1185, BOLD:ABY2743 and BOLD:AAN8220, which will be the subject of a subsequent paper on the remaining members of species complex not found in Canada). Amongst these four BINs, the mean distance in sequence divergence averages 8.82%, suggesting multiple additional taxa. Distribution of confirmed specimens of Agapostemon fasciatus Crawford, 1901 (red dots -data provided above) and A. melliventris Cresson, 1874 (yellow dots, including type material of A. digueti Cockerell, 1924 shown in blue dots) and material identified as A. melliventris s. l., but not examined from GBIF.org (2023) and as recorded in literature (underlying grey boxes). Crawford (1901) mentioned that 18 female and 37 male species formed the type series (i.e. syntypes), though none of these was specifically designated as a lectotype within any subsequent publication by that author. However, Sandhouse (1936) indicated that Crawford selected a lectotype female and lectoallotype male from his syntype series (these also examined by Cockerell (1937) and Roberts (1972)) from material at the USNM (see above). Roberts (1972) indicated that he also examined the type material of A. fasciatus and A. digueti and believed they were conspecific with A. melliventris, based on variable metasomal colour of the females alone and pointed out that, until he synonymised A. fasciatus with A. melliventris, it had been considered by some authors to be of subspecific rank (e.g. Sandhouse (1936)), though others continued to consider it a valid species (e.g. Cockerell (1937), Fischer (1950). His (Roberts 1972) argument for the synonymy was also based on his belief that A. fasciatus was sympatric with A. melliventris in the south-western part of the range (namely Arizona) where intermediate forms were "too common" (cf. discussion of variation). However, there is no reason why distinct species cannot have overlapping ranges and it is likely that he made this comment based on female metasomal colour alone (which varies even with A. melliventris s. str.); he may not have examined or noted the other characters indicated above that seem to be consistent within each taxa (i.e. scape maculations in females and leg colouring in males). When these are considered, a differing distribution of the two taxa is observed, with only a small area of sympatry (Fig. 7). Thus, some of the material identifed as A. melliventris from the Great Plains reported recently by Portman et al. (2022) may actually be A. fasciatus. Roberts (1972) also commented on the nomenclature of A. fasciatus. Vachal (1903) considered Agapostemon a subgenus of Halictus Latreille, 1804, making Crawford's species a junior secondary homonym of Halictus fasciatus Nylander, 1848 (= Halictus tumulorum (Linnaeus, 1758)), so Vachal (1903) proposed the new name, Halictus (Agapostemon) plurifasciatus Vachal 1903. The same also occurred for A. coloradensis Crawford, 1901, which Vachal (1903 renamed H. (A.) coloradinus Vachal, 1903. These specific epithets were used by Sandhouse (1936), though Cockerell (1937) (1950)). However, Roberts (1972) indicated, based on the International Code of Zoological Nomenclature (citing ICZN (1964)), "secondary homonyms rejected before 1961 cannot be revived," specifically citing the example of A. coloradinus which has been used as a valid species name since Vachal (1903) (e.g. Sandhouse (1936, Roberts (1972), Hurd (1979), Moure and Hurd (1987)). Roberts (1972) also suggested that, if Crawford's A. fasciatus were to be recognised as a valid taxon (as is done so here), then A. plurifasciatus should be used, but this is not the case as (from ICZN (1999): 59.3. Secondary homonyms replaced before 1961 but no longer considered congeneric): "A junior secondary homonym replaced before 1961 is permanently invalid unless the substitute name is not in use [applicable to A. plurifasciatus since it was synonymised with A. melliventris by Sandhouse (1936) and Roberts (1972)] and the relevant taxa are no longer considered congeneric, in which case the junior homonym is not to be rejected on grounds of that replacement". As such, A. fasciatus is resurrected from synonymy with A. melliventris and Crawford's original species epithet is considered valid. By contrast, as A. coloradinus has remained in use in a number of taxonomic (e.g. Sandhouse (1936), Roberts (1972), Hurd (1979), Moure and Hurd (1987), Janjic and Packer (2003), Scott et al. (2011)) and ecological works (e.g. Lavigne (1976), Berger et al. (1985), Osborn et al. (1988)) since Vachal (1903), it is the appropriate name for that taxon (see ICZN (1999)).

Discussion
Specimens from at least three additinal BINs are tentatively identified in BOLD as A. melliventris, mostly all matching the general discription of A. melliventris, but suggesting that there are possible additional species in this species group. Members of BOLD: AAJ1185 (seven male specimens) are from one location on the Texas/Mexico border and members assigned to BOLD:ABY2743 are from Arizona (two female, three male specimens). Most interesting are the four specimens of BOLD:AAN8220, also from Arizona, though with the single female having both the clypeus apically and scape maculated, though with a metasoma that is similar to A. fasciatus (based on images available on BOLD). These materials will be covered in a subsequent work on the A. melliventris species complex, which is currently in progress.