The origin of the ancient Dembeni people of the Comoros Islands.

When the first Europeans reached the Comoros Islands in the late fifteenth century, they found a group of small independent city-states forming part of the 'Swahili Corridor', a system of small island and port-city states which stretched up the east coast of Africa, also including cities on Madagascar and the coasts of what today are Mozambique, Tanzania, and Kenya. These states were bound together by a common language, Swahili, and a common religion, and traded slaves and African goods with the Middle East and other areas of the Indian Ocean rim. Local legend had it that the people of the Comoros were descended from King Solomon and the Queen of Sheba, and that upon hearing of the Profit Mohammed, an emissary was sent from the islands, who returned bringing with him the Islamic faith, and built the first mosque in the archipelago on the island of Ngazidja. European ethnologists concluded that the Swahili culture came about as Arab traders moved down the coast of East Africa, taking local women as wives (or, in more lurid accounts, concubines). 

The position of the Comoros Islands in the Mozambique Channel. The island of Ngazidja is now more commonly known as 'Grand Comoros', and is home to Moroni, the capital of the Union of the Comoros, which gained independence from France in 1979 as the 'Federal Islamic Republic of Comoros', adopting its current name in December 2001. The island of Moare remained a French possession, now forming the Department of Mayotte. Wright (1975).

In the 1970s and 80s a series of archaeological digs were carried out in the Comoros. The earliest culture uncovered by these excavations was identified as the 'Dembeni Phase', in reference to the location where it was first discovered, close to the village of Dembeni on the island of Moare (now the French department of Mayotte). Overlying this in places were layers assigned to the more widespread 'Hanyundru Phase', which can be clearly associated with medieval Swahili culture, with buildings made from stone or coral and cement, and imported Middle Eastern Glassware.

Six sites were associated with the Dembeni Phase, Dembeni itself plus Old Sima on the island of Ndzuwani (now more commonly known as Anjouan), M'Bachile, Dzindani, and M'Beni, on the island of Ngazidja (now usually referred to Grand Comoros), and M'Ro Deoua on Mwali (now usually Mohéli). The Dembeni people built small-to-medium sized villages, comprising houses built from wooden posts with a mud covering, on beaches or ridge-tops, always close to lagoons, and with access to freshwater streams. They consumed significant amounts of seafood (Fish, Shellfish, and Turtles), and kept Carrle and Goats, and farmed crops such as Rice, Millet, and Coconuts, which are still common across the region today.

The Dembini people produced distinctive pottery, with two distinct styles found on the islands and thought to have been made locally. The first of these is plain, with a smooth exterior and a burnished interior, sometimes embellished with a band of incisions or shell imprints. The second is covered with a red slip, either on the interior or the exterior, or both, and is typically burnished and decorated with graphite. In both cases the clay used was course and contained particles of red, grey, and white grit, although the former was typically courser. This courser, plainer style was used to make jars and pots, many of which show signs of being used for cooking. The finer, more decorated, style was used to make shallow bowls, possibly for serving food. A single bead made from local pottery was found at Dembeni,

Examples of 'plain' Dembeni-ware pottery from (a) Dembeni, (b) Sima, (c) M'Ro Deoua, (d) Dembeni, (e) Sima, (f) Dembeni, (g) M'Beni, (h) Dembeni, (i) Dembeni, (j) Dembeni, (k) Dembeni, (l) M'Bachile, (m) M'Bachile, and (n) Sima. Carla Sinopoli in Wright (1984).

Dating of the Dembeni sites was difficult with the technology available in the 1970s and 80s. Attempts were made to obtain carbon¹⁴ dates from several items. A shell collected from a midden at Sima yielded a carbon¹⁴ date of 1550±70 years before the present. Since it was understood by this time that it was necessary to adjust dates from shells to take into account for the carbon¹⁴ content of the atmosphere and the ocean in which the shell had formed, this was interpreted as indicating a date of about 910 AD, with a 96% probability of a date between 785 and 1040 AD. A piece of charcoal from M'Bachile yielded a carbon¹⁴ date of 1140±60 years before the present, with a 96% chance of a date between 470 and 650 AD, while a second piece of charcoal from Sima yielded a carbon¹⁴ date of 1960±70 years before the present, with a 96% chance of a date between 155 BC and 215 AD. At first sight, these dates seem incompatible, sine the Dembeni culture is not thought to have persisted for many centuries, but dates obtained from charcoal represent the age of the original wood from which the charcoal was made, and if this wood came from the interior of a tree, may be much older than the burning event.

Another way to determine the age of archaeological sites is by comparing goods from them to similar items from other sites with known chronologies. Although the majority of the recovered material from the Dembeni sites appears to have been local in origin, items from further afield were found at all locations, together representing about 4% of the recovered ceramics. 

Among the items found at M'Ro Deoua were several large basins and the neck of a large pithoi (large storage jar) made from an oxidised yellow-green clay. These items were made from a base-material which contained fragments of older pottery and fragments of igneous rock, and were decorated with chopped-grass impressions. Smaller fragments of the same material were also found at Dembeni and Sima. This pottery resembled similar material known from several sites in the Near East, at sites dated to between the eighth and tenth centuries AD, although this pottery had never been the subject of any detailed study.

A second type of non-local pottery, found at all sites except Dzindani, comprised fragments of large sandy-ware jars decorated with incisions and barbotine appliqué, with a blue-green glaze on the external surface and an speckled grey or grey-green glaze on the interior surface. Similar jars are known from Siraf on the Gulf Coast of Iran, where they first appear in the eighth century, and on Manda Island, Kenya, where they are found alongside ninth-to-tenth century Chinese ceramics. 

The most common form of imported pottery, found at all of the Dembeni sites, were small fine bowls with a white glaze, typically with a splash of coloured glaze on the interior; the colour of this varies, with the most common being cobalt, turquoise, or gold. Similar bowls are known at Susa in southwest Iran, where examples with cobalt on the interior appear around 720 AD, while turquoise does not appear until about 820 AD, as well as at Shiraf, where cobalt appears around 800 AD and turquoise around 850. This may offer some insight into the chronology of the Dembeni sites, as only white pottery is present at M'Ro Deoua, cobalt at Dembeni, and turquoise at M'Bachile and Sima.

As well as pottery, glass fragments were found at Dembeni, Sima, M'Ro Deoua, and M'Bachile. These were greenish, yellowish, clear, or cobalt blue, and round or cylindrical in shape. All appeared to have been blown, and resembled glassware forms widespread in the Near East. A number of iron items, and slag from iron forging were also found. While it was unclear where these had come from, iron is not a locally available resource in the Comoros. 

All of this appeared to show that the Dembeni people occupied the Comoros between about 800 and 1000 AD, and that from the very earliest, they had strong trading connections with the Near East. This apparently supported the idea that they were migrants from this area, rather than the east coast of Africa. However, two grave sites were also found associated with one of the Dembeni sites, Ngazidia. These both had their heads facing to the south, at odds with Islamic custom, making it highly unlikely that they were Muslims, which would almost certainly have been the case with Near Eastern emigrees of the period.

Founded in 2008, the Sealinks Project seeks to study the movements of ancient peoples and development of trading routes around the Indian Ocean. One of the problems this has sought to resolve is the peopling of Madagascar by settlers from Southeast Asia, and the subsequent adoption of Southeast Asian crops, such as Rice, Coconuts, and Mung Beans into the cultures of the Swahili Corridor. Madagascar was settled in the first millennium BC by hunter-gatherers from Africa, but agriculture appeared to have arrived on the island much later, with settlers from Southeast Asia, and today about 10% of Madagascar's flora is of Southeast Asian origin, either crop plants of weeds introduced along with them.

Between 2010 and 2013, archaeologists from the Sealinks Project carried out excavations at early sites on Madagascar and the Comoros, and settlements of the Swahili Corridor on the coast and offshore islands of East Africa, looking for early examples of both African and Southeast Asian crops, which were then dated using modern radiocarbon methods. Unsurprisingly, Southeast Asian crops were almost absent from the African sites, with only small quantities found at the largest and most active ports. On Madagascar, Southeast Asian crops appear abruptly around 1000 AD, with no evidence of any agricultural activity before this point. Surprisingly, the Comoron sites associated with the Dembeni Culture were also overwhelmingly dominated by Southeast Asian crops (only Sima produced small amounts of African crops such as Millet), and these appeared around 750 AD, significantly earlier than on Madagascar.

(A) Map of eastern Africa, including the Comoros and Madagascar, showing the locations of sites included in the crop study. The relative proportions of African and Asian crops are shown for each site (percentages based on numbers of identified specimens per site). (B) Chronological summary of African vs. Asian crop patterns by site from north to south. Crowther et al. (2016).

This raised some surprising questions. The reliance of the Dembeni people on Southeast Asian, rather than African or Middle Eastern crops, suggests that they did not come from either of these areas, but were instead of Southeast Asian origin, potentially part of the great Austronesian migration of peoples from the Malaysian Peninsula through the Islands of Indonesia and on as far as Hawai'i and New Zealand, but also westward to Madagascar. With the Comoros having apparently been settled before Madagascar, it is possible that the larger island was secondarily settled by the Dembeni people, although it is also possible that older sites on Madagascar have not been located.

The modern people of the Comoros are not obviously Southeast Asian. They appear similar to the peoples of East Africa, and speak a language (Comorian) which belongs to the Sabaki language group, which is otherwise spoken along the Tana River of Kenya, and surrounding parts of the Kenyan coast. Their closest cultural links are with the Swahili culture of coastal Kenya, Tanzania, and Mozambique, itself thought to have derived from the Middle East through trade along the east coast of Africa. Possibly this could indicate that the Dembeni people, although the first settlers of the Comoros, did not survive contact with later settlers from East Africa or the Middle East.

With this in mind, a project led by scientists from the Laboratoire d’Anthropologie Moléculaire et Imagerie de Synthèse at the Université de Toulouse carried out a study in which the genomes of volunteers from the Comoros and coastal Swahili communities in Kenya were compared to an extant database, containing genomes from Africa, Madagascar, the Middle East, Southeast Asia, South Asia, East Asia, and Europe.

Surprisingly, despite long-held assumptions about their having Middle Eastern ancestry, the Kenyan Swahili populations were found to be of entirely sub-Saharan African ancestry. The Kenyan Swahili populations were found to have a homogeneous structure, all more closely related to one another than to any other group, and to be more closely related to Bantu-speaking populations than any other African group. Some of the Kenyan Swahili individuals had recent Somali ancestry, but otherwise the Swahili populations were strongly isolated from Somali and Ethiopian populations. The Comorian and Malagasy populations were more closely related to the Kenyan Swahili populations than to any other African population, although both had admixtures of Southeast Asian ancestry, which made up between 33 and 39% of the ancestry of the Malagasy populations, and 8-9% of the ancestry of the Comorians. Curiously, despite being further south than the Kenyan Swahili populations, the Comorians also had some Middle Eastern ancestry, which made up 6-7% of their genomes.

A statistical model of the ancestry of modern Comorians suggested that their common ancestry formed as a distinct population between 1197 and 792 years before the present, with the admixture of a Swahili and a Southeast Asian group. Different islands within the group have also had subsequent influxes of genetic material from Swahili, Middle Eastern, and Southeast Asian populations. The Madagascan populations showed a more recent origin, with an admixture between a Swahili Group and an Island Southeast Asian group between 876 and 742 years before the present (this population appears to have arrived in the area separately, and not to have migrated from the Comoros). 

Admixture Scenario for Populations along the Swahili Corridor as Estimated by GLOBETROTTER and ADMIXTURE Dark red arrows represent Swahili gene flow; light green arrows represent Island Southeast Asian Banjar or Malay; a yellow arrow represents Middle Eastern gene flow; and the purple arrow represents gene flow from the Horn of Africa. The pink arrow represents gene flow from central and southern Bantu speakers. Dates refer to the last detectable admixture event; dates below pie charts refer to the admixture event between the Swahili and Banjar or Malay; dates in italics represent secondary gene flow. Sex-biased gene flows are represented by male and female symbols in the tip of the arrows; note that they are not presentin Malagasy Antemoro and Comorian Moheli. Brucato et al. (2018).

Thus, the most likely scenario is that the Dembeni people reached the Comoros Islands from Peninsula Malaysia or Island Southeast Asia around 750 AD. They brought with them staple crops from Southeast Asia, as well as Cattle and Goats, and fished in the islands' lagoons and surrounding seas. They exchanged trade goods with the Near East, probably through early East African Swahili intermediaries, with whom they may have intermarried. Despite this cultural exchange, the Dembeni appear to have remained a distinct culture till around 1000 AD, when they were abruptly replaced by the Hanyundru culture, which appears to have been more distinctly Swahili in nature. Modern Comorians derive about 8-9% of their genetic ancestry from this group, and still farm many Southeast Asian crops, but speak a language of apparent Kenyan origin, and are culturally close to the Swahili populations of East Africa.

See also...