Comprehensive phylogeny, biogeography and new classification of the diverse bee tribe Megachilini: Can we use DNA barcodes in phylogenies of large genera?

Abstract

Classification and evolutionary studies of particularly speciose clades pose important challenges, as phylogenetic analyses typically sample a small proportion of the existing diversity. We examine here one of the largest bee genera, the genus Megachile - the dauber and leafcutting bees. Besides presenting a phylogeny based on five nuclear genes (5480 aligned nucleotide positions), we attempt to use the phylogenetic signal of mitochondrial DNA barcodes, which are rapidly accumulating and already include a substantial proportion of the known species diversity in the genus. We used barcodes in two ways: first, to identify particularly divergent lineages and thus to guide taxon sampling in our nuclear phylogeny; second, to augment taxon sampling by combining nuclear markers (as backbone for ancient divergences) with DNA barcodes. Our results indicate that DNA barcodes bear phylogenetic signal limited to very recent divergences (3-4 my before present). Sampling within clades of very closely related species may be augmented using this technique, but our results also suggest statistically supported, but incongruent placements of some taxa. However, the addition of one single nuclear gene (LW-rhodopsin) to the DNA barcode data was enough to recover meaningful placement with high clade support values for nodes up to 15 million years old. We discuss different proposals for the generic classification of the tribe Megachilini. Finding a classification that is both in agreement with our phylogenetic hypotheses and practical in terms of diagnosability is particularly challenging as our analyses recover several well-supported clades that include morphologically heterogeneous lineages. We favour a classification that recognizes seven morphologically well-delimited genera in Megachilini: Coelioxys, Gronoceras, Heriadopsis, Matangapis, Megachile, Noteriades and Radoszkowskiana. Our results also lead to the following classification changes: the groups known as Dinavis, Neglectella, Eurymella and Phaenosarus are reestablished as valid subgenera of the genus Megachile, while the subgenus Alocanthedon is placed in synonymy with M. (Callomegachile), the subgenera Parachalicodoma and Largella with M. (Pseudomegachile), Anodonteutricharaea with M. (Paracella), Platysta with M. (Eurymella), and Grosapis and Eumegachile with M. (Megachile) (new synonymies). In addition, we use maximum likelihood reconstructions of ancestral geographic ranges to infer the origin of the tribe and reconstruct the main dispersal routes explaining the current, cosmopolitan distribution of this genus.

Subgeneric classification and biology of the leafcutter and dauber bees (genus Megachile Latreille) of the western Palearctic (Hymenoptera, Apoidea, Megachilidae)

“…In its current delineation, this genus includes more than 1400 species (Ascher and Pickering 2016) in 55 subgenera (Michener 2007, Trunz et al 2016, and references therein). These subgenera are distributed in two groups based on nesting biology: the leafcutter species (or group 1 subgenera), which cut leaf discs to line their brood cells; and the dauber bees (or group 2 subgenera), which use resin or mud to build their cells (Michener 2007, Trunz et al 2016. The subgenus Creightonella Cockerell, 1908, the lone member of a third group according to Michener (2007), is considered to be a member of group 1 (Trunz et al 2016).…”

“…These subgenera are distributed in two groups based on nesting biology: the leafcutter species (or group 1 subgenera), which cut leaf discs to line their brood cells; and the dauber bees (or group 2 subgenera), which use resin or mud to build their cells (Michener 2007, Trunz et al 2016. The subgenus Creightonella Cockerell, 1908, the lone member of a third group according to Michener (2007), is considered to be a member of group 1 (Trunz et al 2016). The dauber bees have sometimes been placed in a separate genus, Chalicodoma Lepeletier, 1841(e.g.…”

“…Michener 1965, Pasteels 1965, Tkalců 1969; Trunz et al (2016) recently showed that the dauber lineages constitute a paraphyletic assemblage from which the leafcutters evolved, and the separation of the dauber bees into one (Chalicodoma) or a few well-delineated, monophyletic genera did not appear practical. Consequently, Trunz et al (2016) advocated the use of a classification close to Michener's generic classification, with a large genus Megachile that includes both leafcutter and dauber lineages with the exception of three comparatively small tropical groups segregated as separate genera: Gronoceras Cockerell, 1907, Heriadopsis Cockerell, 1931and Matangapis Baker & Engel, 2006(Trunz et al 2016). This classification is adopted here, placing all Palearctic species into one large genus Megachile that includes, among others, Chalicodoma, Creightonella and Pseudomegachile Friese, 1898 as subgenera.…”

“…This classification is adopted here, placing all Palearctic species into one large genus Megachile that includes, among others, Chalicodoma, Creightonella and Pseudomegachile Friese, 1898 as subgenera. With respect to the Palearctic fauna, the following further changes were also made to Michener's classification by Trunz et al (2016): the subgenus Parachalicodoma Pasteels, 1966 was merged with the subgenus Pseudomegachile; its species are considered here to form a species group of Pseudomegachile, referred to as the incana group; Eumegachile Friese, 1899 (with only one species, Megachile bombycina Radoszkowski, 1874) was merged into the subgenus Megachile; Eurymella Pasteels, 1965 was recognized as a valid subgenus of the genus Megachile and not as a synonym of Eutricharaea Thomson, 1872. Lastly, the morphologically unusual species M. giraudi Gerstäcker, 1869, placed alternately in the subgenera Eutricharaea (e.g.…”

“…Tkalců 1974) and Xanthosarus Robertson, 1903(e.g. Öz-bek and Zanden 1994, Ornosa et al 2007) is included here in Eutricharaea based on its phylogenetic position (Trunz et al 2016). All these classificatory changes were based on highly supported topologies (100% bootstrap support in all maximum likelihood analyses) and are adopted here.…”