*Carlos A. Fernandes, Biodiversity and Ecological Processes Group, Cardiff University, Park Place, PO Box 915, Cardiff CF10 3TL, UK. E-mail: email@example.com
In a large number of studies concerned with species movements between Africa and Eurasia, including the migrations of hominids out of Africa, a frequently-cited dispersal route is across a hypothetical land bridge in the southern Red Sea, which is suggested to have emerged during glacial sea-level lowstands. This paper, however, unequivocally demonstrates that palaeoceanographic and palaeoecological data are incompatible with the existence of Red Sea land bridges since the Miocene. The case is made by presenting the first quantitative history of water depth above the Red Sea sill for the last 470,000 years, a time period that includes the four most recent glacial–interglacial cycles, and by discussing the predictable consequences of any land bridge formation on the Red Sea sedimentary and microfossil records. The absence of post-Miocene Red Sea land bridges has extensive implications for biogeographic models in the Afro-Arabian region. Genetic, morphometric and palaeontological patterns reported in the literature cannot be related to dispersals over a land bridge, or in the case of marine organisms, separation of the Red Sea from the Indian Ocean by a land bridge. If such patterns in terrestrial species are only congruent with a southern Red Sea dispersal route, then they need to be considered in terms of sweepstake rafting, anthropogenic introduction, or in the particular case of the Out-of-Africa migration by modern humans, seafaring. The constraints imposed by our palaeoenvironmental record on biogeographic reconstructions within and around the Red Sea will hopefully encourage both the review of previous works and the preference for multidisciplinary approaches in future studies.
Evaporites, microfossils and water depth history in the Red Sea
The only place where a hypothetical land bridge might develop is in the southernmost Red Sea, at the Hanish Sill, with a present-day maximum water depth of 137 m (Werner & Lange, 1975). Emergence of that sill would isolate the Red Sea from the Indian Ocean. The high net evaporation over the Red Sea would then rapidly draw down the basin level, at a rate of approximately 2 m year−1 (Smeed, 2004). The resultant concentration of salt within the basin would cause dramatic deterioration of the Red Sea environment. Without doubt such deterioration would be registered in the Red Sea sedimentary and fossil record. Essentially, the basin would become sterilized within about 200 years, except for a small number of hypersalinity-tolerant species, and formation of evaporite deposits would start within about 600 years. Neither is observed in marine sediment cores covering the last 470,000 years – there are no basin-wide evaporite deposits, and marine faunas and floras that are unable to survive hypersaline conditions continue throughout (Rohling et al., 1998; Siddall et al., 2003).
Using the model of Siddall et al. (2003), we calculate that sea level must at all times during the last 470,000 years have remained more than 15 m above the level of the sill (Fig. 2). Note also that the strait would have remained at least 5 km wide during even the most severe sea-level lowstands. The most recent phase of basin-wide evaporite deposition occurred in the Red Sea during the middle to late Miocene around 12–6 Ma (Orszag-Sperber et al., 2001). Realistically this was the most recent period for which a land bridge can have separated the (proto-)Red Sea from the open ocean.
Implications for Afro-Arabian biogeography and Out-of-Africa models of Homo dispersal
Given this seemingly incontrovertible evidence against the emergence of land bridges in the southern Red Sea, we conclude that models of Plio-Pleistocene movements of terrestrial fauna between Africa and Arabia should only consider the hypotheses of dispersal via the Sinai route, ‘sweepstake’ rafting and/or anthropogenic introduction. Rafting is recognized to have played a role in several transmarine colonizations across distances that far exceed the width of the Red Sea, not only by invertebrates and small vertebrates but also by medium to large-sized mammals (de Queiroz, 2005).
Likewise, biogeographic studies of marine organisms in the Red Sea should not simply ascribe genetic differentiation across the strait to a land bridge separating the Red Sea from the Indian Ocean during past sea-level lowstands (Shefer et al., 2004). Alternative explanations may include strongly enhanced environmental (salinity) gradients and diminished water mass exchange through the strait at those times (Fenton et al., 2000).
Regardless of the evolutionary scenarios for its later stages, our record of minimum water depth above the Red Sea sill offers new constraints to the nature of the first step in the global colonization by our species (Fig. 1). If the Out-of-Africa migration entailed the crossing of the southern Red Sea, then the use of watercraft must be invoked in what would possibly be the earliest instance of seafaring in Homo sapiens (Straus, 2001; Bednarik, 2003).
In this context it is interesting that as early as 840,000 years ago (Morwood et al., 1998, 2004) Homo erectus, a species with lower cognitive capabilities, was able to colonize Flores, an Indonesian island on the eastern side of the Wallace's Line – a major biogeographic boundary associated with the permanent deep-water barrier that prevented dispersals from Sundaland into Wallacea by most other land mammals, even during glacial periods of lowered sea level (van den Bergh et al., 2001). Nevertheless, the absence of H. erectus remains in Australia and New Guinea, coupled with doubts about the presence in this species of the cultural and technological complexity needed for seafaring (Erlandson, 2001), raises the question of whether the colonization of Flores might have been the result of mostly passive accidental rafting(s) on vegetation mats (Diamond, 2004).
Because suggestions of planned maritime voyaging in H. sapiens before the colonization of Australasia also remain controversial (Erlandson, 2001), the hypothesis of a coastal dispersal around the Red Sea basin (Stringer, 2000) needs to be evaluated. Movement along the coasts would have provided a buffer against prevailing cool and dry conditions and could have been particularly fast if taking place during interstadials in southern Asia (Schulz et al., 1998), making this model also compatible with recent estimates of the coalescence time for all extant non-African human populations (Macaulay et al., 2005). Furthermore, this is a scenario that fits better with the apparently absent, or at least limited, role of the southern Red Sea passageway in later human migrations during the Upper Palaeolithic and Mesolithic (Luis et al., 2004).
Unfortunately, no modern human fossils confidently dated between 85,000 and 65,000 years ago have been found from Arabia, northeastern Africa and the Levant, yet such a find would help to identify the most likely route for the Out-of-Africa dispersal (Vermeersch et al., 1998; Shea, 2003). Similarly, data on the lithic industries from the Red Sea margins and environs that could be potentially relevant for the same issue are so far undated (Rose, 2004a) or remain unable to provide an unambiguous picture, due both to their paucity (Rose, 2004b) and regional incomparability; the latter a product of the diversity of how lithic assemblages have been described and analysed in separate localities and by different archaeologists (Vermeersch, 2001).
It is therefore clear that we need more data (genetic, palaeontological, archaeological and palaeoenvironmental) to resolve competing reconstructions of species dispersals through the Red Sea region. We hope that the far-reaching implications of the record here presented might stimulate such future investigations.
We thank Naama Goren-Inbar, Philip Van Peer, Jeffrey Rose, Simon Davis and Stephane Ostrowski for drawing our attention to relevant archaeological literature and/or for their comments on earlier versions of the manuscript. We are also grateful to the editor and two anonymous referees for their helpful suggestions. C.A.F. acknowledges support from the Fundação para a Ciência e a Tecnologia, grant BD/19755/99. E.J.R. acknowldeges support from The Natural Environment Research Council, project NE/C007152/1. M.S. holds a postdoctoral position funded by the European STOPFEN Network research project (HPRN-CT-2002-00221.)
Carlos A. Fernandes is a research associate in the Biodiversity and Ecological Processes Group at the University of Cardiff, UK. He is currently pursuing studies of molecular systematics, phylogeography and molecular adaptation in African and Eurasian carnivores, in which a major aim is to uncover clues regarding the biogeographic and ecological history of these continents.
Eelco J. Rohling is a Professor of Ocean and Climate Change at the National Oceanography Centre (NOC), Southampton, UK. His current research interests include high-resolution investigation of ocean/climate changes during the Neogene, the use of conservative properties and δ18O in oceanographic studies, and the integration of palaeoclimate research with archaeological records.
Mark Siddall is a physical oceanographer interested in palaeoceanography. During his PhD at the National Oceanography Centre (NOC), Southampton, UK, he used a model of Red Sea inflow/outflow to calculate a sea-level curve for the last 0.5 Myr. Now at the University of Bern, Switzerland, he continues to combine simple ocean models with palaeoceanographic data.