When did bumblebees reach South America? Unexpectedly old montane species may be explained by Mexican stopover (Hymenoptera: Apidae)

Abstract

A problem for understanding bumblebee biogeography is that if bumblebees dispersed from Asia through North America to South America, if they are poor at long-distance dispersal with establishment over sea, and if the land bridge between North and South America was not established until c. 3 Ma BP, then there is an apparent conflict with the divergence among currently endemic South American lineages having been dated as early as 15–17 Ma. Using the first complete phylogenetic trees for all known and accepted extant species of the groups involved, we show how this conflict could be resolved. We suggest that characterizing bumblebees as being associated generally with temperate flower-rich meadows conflates divergent habitat specializations between two early lineages, associated with northern lowland grasslands and with southern montane grasslands respectively, which may have driven divergences in behaviour and in biogeographic processes. First, for most of the lowland grassland group of bumblebees, estimated dates of divergence are consistent with dispersal to South America via the land-bridge corridor that opened at c. 3 Ma, followed by extant endemic lineages diverging in situ within South America. In contrast, for the second group that occupies montane grassland habitats (and for a few montane lineages of the ‘lowland’ group), we suggest that dispersal to South America at c. 3 Ma could be consistent with older divergence for currently endemic species if: (1) many of the extant South American lineages had already diverged outside the region before 3 Ma in neighbouring Mesoamerica; and (2) they had been constrained within the high mountains there, dispersing southwards into South America only once the isthmus corridor had become established; and (3) some of those ancestral montane lineages had become extirpated from Mesoamerica during subsequent warm climatic fluctuations. This interpretation re-emphasizes that biogeographic studies need to consider habitat-specific dispersal models that change through time.