Geodispersal as a Biogeographic Mechanism for Cenozoic Exchanges between Madagascar and Africa.

Islands are a natural consequence of Earth’s dynamic nature, and hundreds of thousands occur in our oceans. Most are classified as either continental (subaerial parts of continental shelves, connected to the mainland during sea level low-stands, and often inhabited by elements of the mainland biota) or oceanic (formed over oceanic plates and never connected to a continent, so that their biota arrives by dispersal). Madagascar comprises a large continental fragment (587,041 km2) that shares attributes with both continental and oceanic islands. Today it has an oceanic level of isolation, separated by the Mozambique Channel from Africa’s eastern shore by 420 km at its closest point. The channel is an extraordinarily ancient and formidable biogeographic barrier to the migration of terrestrial vertebrates, marking the first major rupture in the fragmentation of the Gondwana supercontinent, 157 to 120 Mya. For most of its extent, its depth reaches 2000 to 3000 m; eustatic sea level changes of one to several hundred meters could not moderate its effectiveness by very much. The Agulhas Current which flows down this channel is one of the fastest flowing currents in the world. Prior to this fragmentation in deep geological time, Madagascar was connected on all sides to the Gondwanan supercontinent, and undoubtedly hosted an ancient Gondwanan flora and fauna. With equal certainty, Madagascar’s biotic composition has altered repeatedly in response to the tectonic and climatological upheavals that accompanied geological and organismal evolution on Earth. A comprehensive study of phylogenetic relationships among 188 living taxa of Malagasy vertebrates led to reconstruct a complex scenario of biogeographic origins, including ancient (Cretaceous) elements, as well as a majority of Cenozoic (66 Mya-10 kya) colonists that have their closest relatives in either Africa or Asia. Phylogeographic investigations of the Malagasy flora show similar patterns. Biogeographic mechanisms are often viewed through a binary filter: either the organisms were in place before the landmasses sundered, or they arrived by dispersal. In the case of Madagascar, all organisms with African affinities that arrived after 120 Mya, and all those with Indian/Asian affinities that arrived after 83 Mya, are hence assumed to be products of long-distance dispersal; and dispersal is equated with flying, swimming or rafting across considerable marine barriers. Comparable methods of dispersal are proposed for Madagascar’s Cenozoic flora, although eolian transport has also been shown to be an effective long distance dispersal agent to faraway islands for a number of plants. There is a third alternative that has received insufficient consideration, in our view: i.e. geodispersal, or the expansion of floral and faunal ranges in response to the elimination of a prior biogeographic barrier. For Madagascar, this process implies the presence of episodic Cenozoic land connections between the island and eastern Africa, the elevation and submersion of which would yield a pattern of alternating periods of colonization and in situ diversification. The possibility of land bridges connecting Africa and Madagascar has been rejected repeatedly by scientists for a variety of reasons, primary among which is the limited and “unbalanced” higher-taxonomic composition of the island’s biota relative to that of the mainland. In 1940, George Gaylord Simpson published a renowned and much cited paper on mechanisms of insular colonization by mammals, proposing that they involve three potential migration routes: (1) corridors; (2) filter-bridges; and (3) sweepstakes. Corridors are pathways devoid of any physical or ecological barriers; filter-bridges are perennially open to some species and not to others; and sweepstakes are routes of sporadic, accidental, and highly selective dispersal from continent to island by means of either stepping-stones or natural rafts. According to Simpson, Madagascar’s mammal fauna comprises only four terrestrial lineages with links to Africa (lemurs, tenrecs, euplerid carnivorans, and nesomyine rodents) in terms of sweepstakes dispersal. Because Simpson worked within a framework of fixed continents, the only way he could envisage for animals to cross the Mozambique Channel was by floating on rafts of vegetation. Late Miocene taxa including large carnivores, paenungulates (chiefly elephants), apes, and ungulates (other than hippos) were considered non-starters as colonists, whereas a major contributing factor to successful dispersal for earlier and smaller mammals was chance: being in the right place at the right time. Small-bodied canids and felids, monkeys, shrews, and most rodents were simply “unlucky” and missed the raft. This argument continues to be cited: If there had been a land bridge, then “a greater variety of animals would have crossed" , as “all clades of that antiquity would have had equally probable chances of colonizing Madagascar”, and “large-scale invasions [would] almost certainly have ensued”. These predictions place a high value on taxonomic filtering, but fail to acknowledge the significance the habitat filtering; i.e. colonizers can only establish viable populations in habitats to which they are at least partially adapted. Habitat filtering may impose more restrictions on biogeography than dispersal, even on remote oceanic islands. Land bridges provide not only causeways, but habitats as well. This means that, while rafting must occur within an individual’s lifetime, geodispersal can occur over several generations. Another biological aspect that needs to be considered is the vulnerability of island biotas to extinction. The limited number of Malagasy clades alive today is unlikely to represent all of the lineages that ever colonized the island; but the absence of any Cenozoic fossils older than 26 kya renders them invisible to modern research.