Historical biogeography of scarabaeine dung beetles Adrian L. V. Davis*, Clarke H. Scholtz and T. Keith Philips 1 Department of Zoology and Entomology, University of Pretoria, Pretoria, South Africa Abstract Aim (1) To review briefly global biogeographical patterns in dung beetles (Coleoptera: Scarabaeidae: Scarabaeinae), a group whose evolutionary history has been dominated by ecological specialization to vertebrate dung in warmer climates. (2) To develop hypo- theses accounting for the evolution of these patterns. Location Six principal biogeographical regions: Palaearctic, Oriental, Afrotropical, Australasia, Neotropical, Nearctic and five outlying islands or island groups harbouring endemic genera: Caribbean, Madagascar, Mauritius, New Caledonia, New Zealand. Methods Major patterns of tribal, generic and species distribution are investigated using cluster analysis, ordination, parsimony analysis of endemism and track analysis. Attempts are made to resolve biogeographical patterns with findings in the fields of plate tectonics, fossil and evolutionary history, plus phylogeny of both mammals and dung beetles. Results Because of conflict between published findings, it is uncertain at what point in time density of dinosaur dung, mammal dung or both became sufficiently great to select for specialized habits in dung beetles. However, biogeographical evidence would suggest a Mesozoic origin followed by further taxonomic radiation during the Cenozoic, poss- ibly in response to the increasing size and diversity of mammalian dung types in South America and Afro-Eurasia. Proportional generic distribution in fourteen tribes and subtribes showed four principal biogeographical patterns: (1) southerly biased Gondwanaland distribution, (2) Americas or (3) Madagascar endemism, and (4) nor- therly biased, Afro-Eurasian-centred distribution with limited numbers of genera also widespread in other regions. Proportional composition of faunas in eleven geographical regions indicated three principal distributional centres, East Gondwanaland fragments, Afro-Eurasia and the Americas. These patterns probably result from three principal long- term range expansion and vicariance events (Mesozoic: Gondwanaland interchange and fragmentation, Cenozoic: Afro-Eurasian ⁄ Nearctic interchange and the Great American interchange). It is suggested that old vicariance caused by the Mesozoic fragmentation of Gondwanaland leads to a high degree of regional endemism at generic or tribal level across one or more Gondwanaland tracks. In contrast, it is suggested that the more recent Cenozoic range expansions occurred primarily towards northern regions leading to endemism primarily at species level. These Cenozoic radiations were facilitated by the re-linking of continents, either because of tectonic plate movements (Africa to Eurasia in Miocene), climatically induced sea-level change (Afro-Eurasia to Nearctic in Miocene and Pleistocene), or similar coupled with orogenics (Nearctic to Neotropical in Pliocene). Speciation has followed vicariance either because of climatic change or physical barrier development. These recent range expansions probably occurred principally along an Afro-Eurasian land track to the Nearctic and Neotropical and an Americas land track northwards from the Neotropics to the Nearctic, with limited dispersal from Eurasia to *Correspondence: Department of Zoology & Entomology, University of Pretoria, 1 Lynnwood Road, Pretoria, 0002 South Africa. E-mail: adavis@zoology.up.ac.za 1 Present address: Department of Biology, Western Kentucky University, Bowling Green, KY 42101–3576, USA. Journal of Biogeography, 29, 1217–1256 Ó 2002 Blackwell Science Ltd