Analysis of family-level relationships in bees (Hymenoptera: Apiformes) using 28S and two previously unexplored nuclear genes: CAD and RNA polymerase II.

We analyzed a combined data set of two protein-coding nuclear genes (CAD and RNA polymerase II) and a nuclear ribosomal gene (28S D2-D4 region) for 68 bee species and 11 wasp outgroups. Our taxon sampling included all seven extant bee families, 17 of 20 subfamilies, and diverse tribes. Wasp outgroups included the two families most closely related to bees: Crabronidae and Sphecidae. We analyzed the combined and single gene data sets using parsimony and Bayesian methods, which yielded largely congruent results. Our results provide reasonably strong support for family and subfamily-level relationships among bees. Our data set strongly supports the sister-group relationship of the Colletidae and Stenotritidae, and places Halictidae as sister to this clade combined. Our analyses place the Melittidae and the long-tongued (LT) bee clade (Apidae+Megachilidae) near the base of the tree with Colletidae (and Stenotritidae) in a fairly highly derived position. This topology ("Melittidae-LT basal") was obtained in previous morphological studies under certain methods of character coding. A more widely accepted tree topology that places Colletidae (and/or Stenotritidae) as sister to all other bees ("Colletidae basal") is not supported by our data. The "Melittidae-LT basal" hypothesis may better explain patterns in the bee fossil record as well as historical biogeography of certain bee groups. Our results provide new insights into higher-level bee phylogeny and indicate that CAD, RNA polymerase II, and 28S are useful data sets for resolving Cretaceous-age divergences in bees and other Hymenoptera.