The success of invading plants in island ecosystems has often been inferred to result from ‘invader complexes’, where cointroduced plants and their specialist pollinators can reciprocally enhance each other’s spread. However, it has also been suggested that in islands with low pollinator diversity, those pollinators should evolve into super-generalists that may be able to pollinate a wide range of exotic plants, enabling the spread of exotic weeds. Fiji has a very depauperate endemic bee fauna and previous studies have suggested that its only lowland bee species, Homalictus fijiensis (Apoidea: Halictidae), has a very wide range of host plants. However, those studies only included a small number of endemic flowering plants. Our study expands observations of bee–flower interactions to explore host plant ranges of H. fijiensis and introduced bee species to include a wider variety of native and introduced plant species. We show that H. fijiensis does have a wider host range than introduced bees, including Apis mellifera (Apoidea: Apidae), and an ability to exploit extrafloral nectaries and poricidal anthers that are not utilised by the introduced bee species. Our results support the hypothesis that super-generalism can evolve in islands where pollinator diversity is low, and that this may make those islands susceptible to weed invasions.

The Ektopodontidae are an enigmatic group of phalangeroid marsupials known from the late Oligocene to the Early Pleistocene of Australia. Although represented to date only by isolated teeth and several partial dentaries and maxillae, their highly distinctive dental morphology has allowed three genera and nine species to be distinguished. Here, we describe possibly the geologically oldest ektopodontid, Chunia pledgei sp. nov., from the Oligocene Pwerte Marnte Marnte fossil locality of central Australia. Phylogenetic analyses of Phalangeroidea, using 80 primarily dental characters framed by a molecular scaffold, support placement of the new taxon in the genus Chunia. The analyses failed to recover species of the genus Durudawiri in a monophyletic Miralinidae, indicating that they require systematic review. We also transfer the purported basal phalangerid Eocuscus sarastamppi to Miralinidae (Miralina sarastamppi comb. nov.). Additionally, the M1 specimens used to describe the Early to Middle Miocene miralinid genus Barguru, and three species therein, are reidentified as deciduous third premolars from early macropodoids. These findings imply that the Miralinidae are known only from the late Oligocene, whereas the oldest named phalangerids are from the Early Miocene. From a functional consideration of ektopodontid dental morphology, we infer support for prior suggestions of a granivorous and/or frugivorous diet for them. The relative stage-of-evolution expressed by the new taxon is comparable to those in the lower faunal zones of the Namba and Etadunna formations, which supports a late Oligocene age for the Pwerte Marnte Marnte assemblage. http://zoobank.org/urn:lsid:zoobank.org:pub:8881FF3A-4B9F-4085-A5C2-1BBF976CFB65 SUPPLEMENTAL DATA-Supplemental materials are available for this article for free at www.tandfonline.com/UJVP.

No abstract