Until the last decade there were twenty published shell midden sites relating to the Red Sea dated to the Holocene, mainly in the south: three on its south-west coast in Eritrea (Bar-Yosef Mayer and Beyin 2009), the remainder, bar one, on its eastern coastline in Southern Saudi Arabia and Yemen (see Edens and Wilkinson 1998; Durrani 2005). The exception is a site on the north-west coast of Egypt (Vermeersch et al. 2005). This concentration of sites towards the south may be due to research bias, but it also probably reflects the fact that the coastal morphology and ecology of the southern Red Sea is more conducive to coastal resource exploitation, with extensive archipelagos on both sides of the Red Sea, combined with shallow offshore topography/bathymetry, and many shallow bays that make for productive marine environments.
On the East side of the Red Sea six sites have been found in southwest Saudi Arabia, and seven sites in Yemen on the Tihama Plain. The Comprehensive Archaeological Survey Program of Saudi Arabia dated five shell midden sites to between 6.3–5.4 ka cal BP, four located on the mainland, and one on the Farasan Islands. The sites were lower than 1m in height, and often quite broad; ostrich egg shell was recovered from one site, along with bones of hunted wild animals (Zarins et al. 1981). An additional site was identified further to the north in Al Birk dating to 5.7 ka cal BP (Bailey et al. 2007), again being lower than 1m. In Yemen seven sites have been documented, dating to between 8–5 ka cal BP (Tosi 1986). These sites contain bones from both domesticated and hunted mammals and fish bones, and are less than 1m in height but some are quite extensive up to 150m across. A revival of interest in the region at the turn of the century resulted in further sites being identified and investigated, including the Al Birk site mentioned above.
In Eritrea, shell middens, including the earliest in the Red Sea at 8 ka cal BP (Beyin and Shea 2007; Bar-Yosef Mayer and Beyin 2009), demonstrate the existence of episodes of coastal exploitation during the mid-Holocene. However, the restricted number of sites (and their location inland on hill tops) is in contrast to the eastern side of the Red Sea where the majority of sites are located on Holocene palaeoshorelines. The Eritrean sites date between 8.5–5 ka cal BP, and are 50cm or less in depth, with evidence for ostrich eggshell working.
The remaining site on the west side of the Red Sea, in El Gouna, Egypt, is a 1.5m high mound dating to around 6.6 ka cal BP (Vermeersch et al. 2005). The mound is part of a group; layers of sand within the stratigraphy appear to indicate periods of abandonment.
Beginning in 2006, new work on the Farasan Islands, on the east side of the Red Sea, revealed a hitherto unsuspected abundance of shell midden sites (Bailey et al. 2007). Subsequent work has revealed 3000 sites that regularly reach heights of 3m, and occasionally heights of up to 6m; many of these are located directly on the contemporaneous palaeoshorelines (Meredith-Williams et al. 2014). To date twenty shell midden sites have been excavated, and two dated to 5.4–4.4 ka cal BP (Williams 2010). The excavations revealed a dominance of shell with very little matrix; fish bones were present, as well as bones of hunted mammals. Stone tool artefacts recovered from the excavations (from well stratified layers) and survey, demonstrate lithic sources not found on the Farasan Islands (basalt, obsidian, granite and schist). The closest sources for these materials is on the mainland; and their presence on the islands is therefore evidence of contacts with the mainland. Given the nature of the sites and this material it is therefore likely that fairly regularly seafaring of some description occurred during the formation of the sites.
Further research of sites on both sides of the Red Sea, using new techniques, has given tantalising insights into long-distance connections across the Red Sea (Khalidi 2010; Khalidi et al. 2010).