1. Introduction

Aquatic resource exploitation is thought to have played an integral role in subsistence strategies of humans from early times (e.g. Archer et al. 2014), becoming increasingly important in the occupation of coastal areas (Bailey and King 2011; Erlandson 2001). While in some cases this could constitute a large part of the diet, it could also be vital as a supplementary food source during times of stress (e.g. Meehan 1977). Regions with high productivity and abundant marine resources can be attractive places for human occupation, during times of plenty as well as during famine. This is evidenced in part by the presence of shell middens on coastlines around the world, from a variety of periods from the Palaeolithic to present day (Bailey 2010). Shellfish are by no means the only major resource in coastal zones; their remains are simply the most visible (e.g. Bailey et al. 2007). Many other coastal and marine resources make coastlines very attractive places to live, including fish, marine and terrestrial mammals, birds, and a range of marine and terrestrial plants. These marine and terrestrial attractors often come together in environments such as estuaries, deltas and coastal lagoon systems.

Shell midden sites do not necessarily represent evidence for fully developed coastal economies relying solely on marine resources (Gutiérrez-Zugasti et al. 2011), even though they often dominate coastal archaeological assemblages. However, they can contain a wealth of other useful information on the past, from climatic proxy data, to artefacts, and evidence of other subsistence activities. They can also contain very well stratified deposits, increasing the resolution at which information can be derived from their investigation. They are usually better preserved than many other types of site, owing to the more durable nature of shells. Combined, these attributes make shell middens useful and unique for archaeological investigation of the coasts.

Detection of shell midden sites has previously been largely restricted to traditional field survey. Here we employ remote sensing methods refined and field-tested for locating shell midden sites on the Farasan Islands in the south-east Red Sea in Saudi Arabian waters, and use these methods to construct a predictive model of shell-midden location (Meredith-Williams et al. 2014). We then apply these methods and model to the inspection of satellite imagery over a wider area of the southern Red Sea, an exercise that serves both as a means of locating new sites and of further testing the model of site location developed in the Farasans. This exercise has resulted in the discovery of a large number of previously unknown shell middens particularly in the Dahklak Islands. We then compare the features of the shell middens in the Farasan and Dahlak Islands and examine what this may mean in terms of Red Sea crossings.